JP6543365B2 - Error correction method and apparatus for input method based on user input speed - Google Patents

Description

  The present invention relates to the field of input methods, and more particularly to error correction techniques for input methods based on user input speed.

  When the user inputs text in the user device, 1) error correction is performed by the text error correction model, and 2) based on the drop point information at the time of input, based on the positional relationship from the drop point to each key Error correction is performed by the two conventional means of determining how to perform error correction on the input code. Among them, the above-mentioned text error correction model is inflexible, the quality of the word output can only be ensured by a strict sorting policy, and environment and scene information at the time of user input are often overlooked. It is easy to misjudge depending on factors such as screen size, resolution, age of the user to use, hand size, usage habit of one hand / both hands etc. The quality is lower.

  Therefore, one of the problems of the technique to be solved by the person skilled in the art is to provide an error correction means of a more efficient, accurate and flexible input method.

  An object of the present invention is to provide an error correction method and apparatus for an input method based on user input speed.

According to one aspect of the present invention, there is provided an error correction method of an input method based on user input speed. The error correction method is
Obtaining (a) the current single key input speed triggered;
Adjusting a weight when an input code corresponding to the current single key is corrected to an input code corresponding to another key according to the input speed;
Performing error correction on the input code corresponding to the current single key based on the adjusted weight to obtain at least one input code after error correction;
And (d) determining at least one candidate word corresponding to the whole input according to at least one input code after error correction corresponding to each key in one whole input.

Preferably, the error correction method is
The average input speed of the key corresponding to each input code in the overall input code is acquired according to the input time of the overall input code corresponding to the one overall input, and the at least one candidate corresponding to the overall input is acquired The method further includes the step (e) of adjusting the sorting of the words.

Preferably, the step (e) comprises
Calculating the average input speed of a key corresponding to each input code in the overall input code according to the input time of the overall input code (e1);
Determining a corresponding error correction model based on the average input speed (e2);
Adjusting the sorting of the at least one candidate word corresponding to the global input according to the error correction model (e3).

Preferably, according to the input speed, adjusting weights when an input code corresponding to the current single key is corrected to an input code corresponding to another key is:
Increasing the weight when an input code corresponding to a key prone to an erroneous touch is corrected to an input code corresponding to another key;
Reducing its own weight when an input code corresponding to a key susceptible to an erroneous touch is corrected to an input code corresponding to another key;
Increasing an own weight when an input code corresponding to a key that is unlikely to cause an erroneous touch is corrected to an input code corresponding to another key;
Reducing the weight when the input code corresponding to the key that is less likely to cause an erroneous touch is corrected to the input code corresponding to another key.

Preferably, the determination of the degree of difficulty of occurrence of an erroneous touch of the key is
Judgment according to the level hierarchy of single key input speed,
Judging according to the score weight of single key input speed,
Judging according to the level of the fitting function map of the single key input speed and the current input environment,
Determining according to the score weight of the fitting function map between the single key input speed and the current input environment, and / or at least one of the following.

Preferably, determining a corresponding error correction model based on the average input speed
An error correction level model with a single said average input speed,
An error correction score model with a single said average input speed,
Error correction level model by fitting the average input speed with other environmental parameters,
An error correction score model by fitting the average input speed with other environmental parameters.

Preferably, the step (e) further comprises:
Noise elimination processing is performed on a sequence consisting of key-in timing of each input code in the entire input code to obtain a noise elimination time series;
The input time of the entire input code is determined according to the noise elimination time series,
Obtaining the average input speed of the overall input code according to an input time of the overall input code, and adjusting sorting of the at least one candidate word corresponding to the overall input.

Preferably, the noise elimination process is
Deleting the timing when the key-in timing of the input code is greater than the upper threshold of the predetermined timing or smaller than the lower threshold of the predetermined timing;
Deleting the timing when the ratio of the key-in timing of the input code to the value of the timing of the current input average is larger than the upper threshold of the predetermined ratio or smaller than the lower threshold of the predetermined ratio;
Deleting the timing when the input input code has a map value with the fitting function of the current input environment greater than the upper threshold of the predetermined map or smaller than the lower threshold of the predetermined map; Includes at least one of them.

Preferably, the error correction method further comprises
Determine the user group where the user is located according to the characteristics of the user of the trigger key,
And optimizing the error correction method according to the user group information.

Preferably, the error correction method further comprises
Get custom information of user's input history,
And optimizing the error correction method according to the conventional information of the input history.

  According to another aspect of the present invention, there is provided an error correction device of an input method based on a user's input speed.

The error correction device
An acquisition device for acquiring the current single key input speed triggered;
An adjusting device for adjusting a weight when an input code corresponding to the current single key is corrected to an input code corresponding to another key according to the input speed;
An acquisition device that performs error correction on an input code corresponding to the current single key based on the adjusted weight, and acquires at least one input code after error correction;
And a determining device for determining at least one candidate word corresponding to the whole input according to at least one input code after error correction corresponding to each key in one whole input.

Preferably, the error correction device further comprises:
The average input speed of keys corresponding to each input code in the overall input code is obtained according to the input time of the overall input code corresponding to the one overall input, and the at least one candidate word corresponding to the overall input A sorting device for adjusting the sorting of

Preferably, the sorting device is
A calculation unit for calculating the average input speed of the key corresponding to each input code in the entire input code according to the input time of the entire input code;
A first determining unit for determining a corresponding error correction model based on the average input speed;
A first adjusting unit for adjusting the sort of the at least one candidate word corresponding to the overall input according to the error correction model.

Preferably, adjusting a weight when an input code corresponding to the current single key is corrected to an input code corresponding to another key according to the input speed is:
Increasing the weight when an input code corresponding to a key prone to an erroneous touch is corrected to an input code corresponding to another key;
Reducing its own weight when an input code corresponding to a key susceptible to an erroneous touch is corrected to an input code corresponding to another key;
Increasing an own weight when an input code corresponding to a key that is unlikely to cause an erroneous touch is corrected to an input code corresponding to another key;
Reducing the weight when the input code corresponding to the key that is less likely to cause an erroneous touch is corrected to the input code corresponding to another key.

Preferably, the determination of the occurrence difficulty level of the key wrong touch is
Judgment according to the level hierarchy of single key input speed,
Judging according to the score weight of single key input speed,
Judging according to the level of the fitting function map of the single key input speed and the current input environment,
Determining according to the score weight of the fitting function map between the single key input speed and the current input environment, and / or at least one of the following.

Preferably, determining a corresponding error correction model based on the average input speed
An error correction level model with a single said average input speed,
An error correction score model with a single said average input speed,
Error correction level model by fitting the average input speed with other environmental parameters,
An error correction score model by fitting the average input speed with other environmental parameters.

Preferably, the sorting device further comprises:
A noise cancellation unit for performing a noise cancellation process on a sequence consisting of key-in timings of input codes in the entire input code to obtain a noise cancellation time series;
A second determining unit for determining an input time of the entire input code according to the noise elimination time series;
And a second adjusting unit for acquiring the overall input speed of the overall input code according to the input time of the overall input code, and adjusting sorting of the at least one candidate word corresponding to the overall input.

Preferably, the noise elimination process is
Deleting the timing when the key-in timing of the input code is greater than the upper threshold of the predetermined timing or smaller than the lower threshold of the predetermined timing;
Erasing the timing when the ratio of the key-in timing of the input code to the value of the current input average timing is larger than the upper limit threshold of the predetermined ratio or smaller than the lower limit threshold of the predetermined ratio;
Deleting the timing when the map value of the fitting function of the input code with the current input environment of the input code is larger than the upper threshold of the predetermined map or smaller than the lower threshold of the predetermined map; At least one of

Preferably, the error correction device further comprises:
Determine the user group where the user is located according to the characteristics of the user of the trigger key,
The error correction apparatus is optimized according to the information of the user group.

Preferably, the error correction device further comprises:
Get custom information of user's input history,
The error correction device is optimized in accordance with the customary information of the input history.

  Preferably, the present invention further provides a computer readable storage medium. The computer readable storage medium stores computer code, and when the computer code is executed, the method according to any one of the above is performed.

  Preferably, the invention further provides a computer program product. Where the computer program product is executed by a computer device, the method according to any one of the above is performed.

Preferably, the invention further provides a computing device. The computer device is
One or more processors,
A memory for storing one or more computer programs;
The one or more processors implement the method according to any one of the above when the one or more computer programs are executed by the one or more processors.

  Compared to the prior art, the present invention has the following advantages.

  The present invention analyzes the big data to mine the relationship between the user input speed and the error correction from a temporal point of view to support accurate error correction by the error correction model. The present invention continuously acquires the input speed when each single key is input by the user, and the input code corresponding to the current single key corresponds to the other according to the false touch rate corresponding to the input speed. In the input method, the input speed is made to be one of the elements for correcting the input code by the error correction model by converting the input code corresponding to the key into the input code corresponding to the adjustment of the weight in the case of correction. The accuracy rate of the error correction by the error correction model is improved by further correcting the input code which is evaluated temporally and determined by the spatial thing such as the user's triggered keyboard drop point To improve the quality of user input.

Other features, objects and advantages of the present invention will become more apparent from the detailed description of the non-limiting example based on the following drawings.
It is a schematic diagram which shows the structure of the error correction apparatus of the input method based on the user input speed which concerns on one aspect of this invention. It is a figure which shows the relationship between the said incorrect touch rate and the input time of a single key. It is an example of input display. 5 is a flowchart illustrating an error correction method of an input method based on user input speed according to an aspect of the present invention.

  The same or similar reference numerals in the drawings indicate the same or similar parts.

  Before describing the exemplary embodiments in more detail, some exemplary embodiments are described as processes or methods depicted as flowcharts. Although each operation is described as sequential processing in the flowchart, some of the operations can be performed in parallel, in parallel or simultaneously. In addition, the order of each operation can be newly arranged. The process may be discontinued when the operation is completed, but may include additional steps not included in the drawing. The processing can correspond to a method, a function, a procedure, a subprogram, a subprogram, and the like.

  The "computer device" referred to in context is also referred to as "computer" and refers to an intelligent electronic device capable of operating a predetermined program or command to perform a predetermined processing process such as numerical operation and / or logic operation. It has a processor and a memory, and the processor may execute successive commands stored in the memory to execute predetermined processing steps, or may execute predetermined processing steps by hardware such as ASIC, FPGA, or DSP. It may be realized by combining the above two. Computer devices include, but are not limited to, servers, personal computers, laptop computers, tablet computers, smart phones, and the like.

  The computer device includes a user device and a network device. In addition, although the said user device contains a personal computer, a smart phone, a PDA etc., it is not limited to them. The network devices include, but are not limited to, a single network server, a server group consisting of a plurality of network servers, or a cloud consisting of a large number of computers or network servers by Cloud Computing. Cloud computing is a kind of distributed computing, and is a super virtual computer in which a plurality of loosely coupled computers are concentrated. The computer device may realize the present invention by a single operation, or may be connected to a network to realize the present invention by an interaction operation with another computer device in the network. Also, the network where the computer device is located includes, but is not limited to, the Internet, WAN, MAN, LAN, VPN network, and the like.

  It should be described that the user device, the network device, the network and the like are merely examples, and any other existing or future computer device or network that can appear in the future is included in the protection scope of the present invention, as long as it can be applied to the present invention. Included here in the form of

  The methods described below, some of which are illustrated in the flowcharts, may be implemented by hardware, software, firmware, middleware, microcode, hardware description language, or any combination thereof. When implemented by software, firmware, middleware or microcode, program code or code sets for performing the required tasks may be stored on a machine or computer readable medium (eg, storage medium). The processor (s) can perform the necessary tasks.

  The specific configurations and functions disclosed herein are exemplary only, and serve to illustrate the purpose of exemplary embodiments of the present invention. However, the present invention may be embodied specifically with a plurality of substitution means, and is not interpreted as being limited to the embodiments described herein.

  It should be understood that although the technical words "first", "second" etc. may be used here to describe each unit, these units are not limited to these technical words. The use of these technical words is to separate one unit from another. For example, the first unit may be referred to as a second unit, and similarly the second unit may be referred to as a first unit, without departing from the scope of the illustrative embodiment. good. As used herein, the term "and / or" includes any and all combinations of one or more of the listed items.

  It should be understood that when one unit is described as being "connected" or "coupled" to another unit, it may be directly connected or coupled to the other unit or via a middle unit It is good. On the other hand, when one unit is described as being "directly connected" or "directly coupled" to another unit, it means that it does not go through the middle unit. Other words should be interpreted that describe the relationship between units in a similar format (eg "located between ..." and "located directly between ..." and "... "Adjacent" and "directly adjacent to ..." etc.).

  The technical terms used herein are for the purpose of describing specific embodiments only, and are not intended to limit the exemplary embodiments. As used herein, the singular forms "one", "one" and "the" include plural unless the context clearly dictates otherwise. It should be further understood that as used herein, the term "comprising" and / or "including" define the presence of the described features, integers, steps, operations, units and / or components, and / or more. The presence or addition of other features, integers, steps, operations, units, components and / or combinations thereof of

  Also, in some permutation implementation schemes, the functions / actions mentioned may occur in a sort different from that shown in the figures. For example, depending on the function / operation, the two drawings shown in succession may in fact be performed essentially simultaneously or in reverse.

  FIG. 1 is a schematic view showing a configuration of an error correction device of an input method based on a user input speed according to one aspect of the present invention. The error correction device 1 includes an acquisition device 101, an adjustment device 102, an acquisition device 103, and a determination device 104.

  Here, the error correction device 1 is located, for example, in a user device, and the user device includes, but is not limited to, a personal computer, a portable personal computer, a tablet personal computer, a smartphone, a PDA, and the like.

  Those skilled in the art should understand that the user device is only an example, and user devices that can appear in the future or in the future are included in the protection scope of the present invention and applicable in the present embodiment, if applicable to the present invention. Be It should be further understood by those skilled in the art that the user device here refers to a smart electronic device that executes a predetermined process such as numerical operation and / or logic operation by operating a predetermined program or command. A processor and a memory are provided, and the processor executes successive commands stored in the memory to execute predetermined processing steps, or hardware such as ASIC, FPGA, DSP or the like executes predetermined processing steps, or It may be realized by a combination of the two.

  The acquisition device 101 acquires the current input speed of the single key to be triggered. Specifically, when the user performs an input operation and presses or touches a key on the keyboard of the user device, the acquisition device 101 interacts with the user device to provide, for example, an application interface (API provided to the input application) ) One or more times to obtain the current single key input speed triggered by the user collected by the user device sensor. It is noted that the keyboard includes, but is not limited to, an actual keyboard or a virtual keyboard on a user device screen.

  Those skilled in the art should understand that the false touch rate corresponding to the input speed described above is merely exemplary, and the map relationship corresponding to the false touch rate of the actual input speed is analyzed for big data, It is an erroneous touch value which is statistical.

  The adjusting device 102 adjusts the weight when correcting the input code corresponding to the current single key to the input code corresponding to the other key based on the input speed. Specifically, the false touch rate corresponding to the input time of a single key is statistics based on researches and tests, with the time for single input for each key as a group box. Here, the single key input time is an interval between two adjacent single keys being triggered. FIG. 2a shows the relationship between the false touch rate and the single key input time, and the single key input speed can be calculated based on the single key input time, ie, input speed = 1 / input time It is. As shown in FIG. 2a, the false touch rate gradually decreases as the single key input speed decreases, ie, as the single key input time increases. That is, as the user's input speed is faster, an input error is more likely to occur. Therefore, the adjusting device 102 adjusts the weight when correcting the input code corresponding to the current single key to the input code corresponding to the other key based on the input speed, and the existing weight The accuracy rate of user input is improved by applying to the error correction model and supporting the error correction by the existing error correction model. For example, if the input speed of one single key is relatively fast, the adjusting device 102 corrects the input code corresponding to the current single key to the input code corresponding to the other key based on the input speed. If the input speed of one single key is relatively slow, the adjustment device 102 makes the input code corresponding to the current single key correspond to the other key based on the input speed. Reduce the weight w corrected to the input code. The specific numerical value of the weight w may be set according to a predetermined rule. For example, the total weight value tw is set to 1, w is an arbitrary numerical value between 0-1 and a specific numerical value of w is determined according to a predetermined rule.

  Note that adjusting the weight when the input code corresponding to the current single key is corrected to the input code corresponding to another key based on the input speed is 1) prone to false touch Increasing the weight when the input code corresponding to the key is corrected to the input code corresponding to the other key, and 2) the input code corresponding to the key where the erroneous touch is likely to occur correspond to the other key Decrease own weight when corrected to code, 3) increase own weight when input code corresponding to a key that is less likely to be erroneously corrected to input code corresponding to another key And 4) reducing the weight when an input code corresponding to a key that is unlikely to cause an accidental touch is corrected to an input code corresponding to another key, but is not limited thereto. Specifically, when the input speed is high, for example, when the input speed of the current single key exceeds a predetermined speed threshold vf, the current single key is a key that is likely to cause an erroneous touch. Increase the weight to correct keys prone to false touch to other keys, and decrease the weight of the current single key itself. Note that the weight is a first weight, and the first weight indicates the weight of the current single key and the weight at which the current single key is corrected to another key. For example, taking the QWERTY full keyboard as an example, it is assumed that the current single key of the obtained user is "r". At this time, if the current single key “r” is a key prone to an erroneous touch, ie, if the current single key input speed is relatively fast, the user's desired trigger is Decrease the first weight where the key is "r" itself and increase the first weight where the user's desired trigger key is the "r" peripheral key "e", "f", "t" Do. Note that the key of the user's desired trigger is the user's purpose key, the current single key is the trigger key acquired by the user device, and the current single key is not necessarily the user's purpose key. is not. Thus, the current single key itself also has a corresponding first weight so that the error correction model can combine other parameters based on the first weight of each key to determine the user's purpose key. ing. On the other hand, if the input speed of the current single key is lower than a predetermined speed threshold vs, ie if the current input speed of the single key is relatively low, then the current single key Is a key that is less likely to cause an erroneous touch, and increases its own first weight of the key that is less likely to cause an erroneous touch, and an input code corresponding to the key that is less likely to cause an erroneous touch corresponds to another key Reduce the first weight if it is corrected to For example, if the current single key of the obtained user is "i" and the current single key "i" at this time is a key that is unlikely to cause an erroneous touch, Increase the first weight where the key is "i" itself and decrease the first weight where the user's desired trigger key is the "r" peripheral key "u", "k", "o" Do.

  Preferably, the determination process of the occurrence of the key mistouch includes, but is not limited to:

  1) Judgment based on the level hierarchy of single key input speed. For example, according to the speed range of input speed, the level of input speed of single key is divided into high level and low level. When the input speed of the single key is high, a second weight corresponding to the high level is obtained for the single key, and the single key is erroneously touched based on the second weight. It is possible to determine whether it is a key that is hard to occur or a key that is prone to false touch. When the input speed of the single key is at a low level, a second weight corresponding to the low level is obtained for the single key, and the single key is erroneously touched based on the second weight. It is possible to determine whether it is a key that is hard to occur or a key that is prone to false touch. The level at which the single key input speed is hierarchized according to the range of the input speed may be other plural means.

  2) Judgment based on score weight of single key input speed. Specifically, when a single key is triggered by the user, the acquisition device 101 acquires the current input speed of the single key triggered and scores for the single key based on the input speed of the single key. I take the. For example, by setting a function having an input speed as a variable, a score corresponding to each different input speed is calculated, and a second weight corresponding to the input speed of the single key is obtained based on the score. Based on the second weight, it is determined whether the single key is a key that is likely to cause an erroneous touch or a key that is unlikely to cause an erroneous touch.

  3) Determine based on the input speed of a single key and the level of the fitting function map of the current input environment. Specifically, a fitting function is generated based on two parameters of the input speed of the single key and the current input environment, and it is integrally determined which level the single key is at. It is determined whether the current single key is a key that is prone to an erroneous touch or a key that is unlikely to cause an accidental touch according to the level. The current input environment is, for example, whether or not the light beam is sufficient, the current input device is a personal computer, a tablet or a mobile phone, or the user device is in a stationary state or in a moving state. Or, what is the movement speed.

  4) Determine based on the input speed of a single key and the score weight of the fitting function map of the current input environment. Specifically, the single key is scored according to the input speed of the single key and the current input environment. For example, by setting a fitting function having an input speed and a current input environment as variables, a score corresponding to each single key is calculated in different input speeds and different input environments, and the score is calculated according to the score. Acquire a second weight corresponding to the input speed of a single key, and based on the second weight, whether the single key is a key prone to false touch or a key unlikely to false touch To judge.

  Those skilled in the art should understand that the means for adjusting the weight when the input code corresponding to the current single key is corrected to the input code corresponding to the other key based on the input speed is merely illustrative. Means for adjusting the weight when an input code corresponding to the current single key is corrected to an input code corresponding to another key based on the input speed that can appear in the future or in the future If it is applied to the present invention, it is included in the protection scope of the present invention, and is included in the form incorporated herein.

  The acquisition device 103 performs error correction on the input code corresponding to the current single key based on the adjusted weight, and acquires at least one input code after error correction. Specifically, the acquisition device 103 corrects the error correction model corresponding to the input method corresponding to the error correction device 1 based on the weight which is the first weight. Here, by using the current single key input speed triggered as one parameter of the error correction model, the error correction model can be evaluated and calculated from the viewpoint of time when performing error correction. become. In addition, for the input code corresponding to the current single key, combined with at least one non-on-screen triggered historical key belonging to the same one general input as the current single key. The error correction is performed to obtain at least one input code after the error correction. Here, the one full input consists of the current single key and the non-on-screen triggered triggered history key. That is, for all input codes that are not input on screen, the candidate word corresponding to the triggered history key that is not input on screen also changes each time a single additional key is input. For example, it becomes like the input process shown by FIG. 2 b. FIG. 2 b is an example of the input display, and one whole input is “shoifashi”. If the user continuously inputs "zainali" without performing on-screen input, then the entire input at this time becomes "shoifashizainali".

  For example, if the weight when the input code corresponding to the current single key adjusted by the adjustment device 102 is corrected to the input code corresponding to the other key, the error correction device 1 causes the current correction. The rate at which the input code corresponding to a single key is corrected to another input code is large, and the error correction model corresponding to the input method has a high probability that the current single key is corrected to another key. Become. An appropriate existing error correction model is selected according to the adjusted weight, the adjusted weight is used for the existing error correction model, and error correction is performed with other parameters in the existing error correction model. That is, error correction is performed from a temporal point of view, the current single key input speed is introduced into the existing error correction model as a factor of the existing error correction model, correct error correction is performed on the input code, and acquisition is performed. The apparatus 103 obtains at least one input code after error correction by the error correction model. The at least one input code after the error correction may still be the input code corresponding to the current single key, or the input code corresponding to the current single key may be another key. It may be corrected to the corresponding input code. For example, when the input interface is a touch-type virtual QWERTY full keyboard interface, the error correction device 1 detects all the keys in the keyboard according to the keyboard drop point position L triggered by the user's input. Calculate the distances s1, s2 ... sn from the center point to the drop point position L, calculate the probability that the current single key triggered by the user becomes each key according to the distance, and thereafter calculate the probability Based on the probabilities or in combination with other parameters, it is obtained that the input code corresponding to the current single key triggered by the user is "s". And, the input speed of “s” is 9 alphabets / second, and the acquisition apparatus 101 acquires the input speed of the input code “s” as 9 alphabets / second. For example, if the threshold of the input speed is 5 alphabets / second, the current single key may be selected when the input speed of the input code exceeds the input speed threshold, ie, when the input speed belongs to the high level. The weight is increased if the corresponding input code is corrected to the input code corresponding to the other key, and the input code corresponding to the current single key is corrected to the self “s”, ie When the user's purpose key decreases the weight that is still "s" itself, the adjusting device 102 determines that the input speed exceeds the input speed threshold according to the input speed, "s" is "w". Increase the weight corrected to "w" to w1, increase the weight corrected "s" to "a" to w2, increase the weight corrected to "d" to w3 and correct to "x" Increase the weight to w4 and the weight of the user's purpose key is still "s" itself to w5, and the acquirer 103 reduces the weights w1, w2, w3, w4, w5, and the current single key before Integrally determine the probability of correcting the current single key to another key based on the history key of, for example, “s” is calculated to “w”, “a”, “d”, “x” respectively by calculation The probability of being corrected is P1, P2, P3 and P4, respectively, and the probability that the user's purpose key is "s" is P5, and the user's purpose key is "P1", P2, P3, P4 and P5. Confirm as "a". Integrally determine the probability that the current single key will be corrected to another key based on the current single key previous on-screen non-history key, and combine it with the error correction method of the existing error correction model Perform error correction on the input code corresponding to the current single key. For example, if the error-corrected input alphabet before "s" is the initial "p", both "p" and "s" are adjacent to "s" in the alphabet "s" keyboard because both are vocals. The weights corrected to the prosodic "a" increase.

  Those skilled in the art should understand how to adjust the weight when the input code corresponding to the current single key is corrected to the input code corresponding to the other key based on the input speed threshold. The means for determining if is only exemplary and the weight when the input code corresponding to the existing single key which can appear in the future or in the future is corrected to the input code corresponding to the other key A means for determining how to adjust, if applicable to the present invention, is included in the protection scope of the present invention and included here in the form of incorporation. Here, when acquiring device 103 acquires at least one input code after error correction, it is registered only in the background of the system without displaying it on the input interface, that is, it is still displayed to the user that it is displayed on the input interface. Input code corresponding to the triggered key.

  Here, by analyzing the big data, the relationship between the user's input speed and the error correction is mined from a temporal point of view to support accurate error correction by the error correction model. Also, the error correction device 1 continuously obtains the input speed for each single key input by the user, and the input code corresponding to the current single key corresponds to the other key according to the input speed. By setting the input speed as one of the factors for correcting the input code by the error correction model by adjusting the weight in the case of correction to the corresponding input code, the input method is evaluated from the time point of view, By further correcting the input code determined based on the spatial viewpoint of the triggered keyboard drop point, the accuracy rate of error correction by the error correction model is improved, and user input Improved the quality.

The determination device 104 determines at least one candidate word corresponding to the whole input based on at least one input code after error correction corresponding to each key in one whole input. Specifically, the acquiring unit 103 sequentially acquires at least one input code after error correction, and after the acquiring unit 103 acquires an input code after error correction corresponding to each key of one full input. The determining device determines at least one candidate word corresponding to the whole input. For example, following the above example, after correcting the alphabet "s" to the prosody "a" adjacent to the "s" in the keyboard, "pa" is composed of the alphabet "a" and the previous alphabet "p". , And the determining device 104 is configured to select at least one candidate word corresponding to “pa”, for example,
Confirm Furthermore, the determination device 104 determines the corresponding at least one candidate word based on the at least one input code before the error correction, and the at least one candidate corresponding to the at least one input code before the error correction. The priority grade of the word is lower than the at least one candidate word corresponding to the at least one input code after error correction. For example, at least one candidate word corresponding to the input code "ps" before error correction
Priority grade such as
Lower than.

  The above input process is continuous, and the user's input continues. In the process of inputting the input code, as long as the on-screen input is not executed, the input code included in the current one full input increases rapidly and the newly increased input code is error corrected. By combining with the input history code that is not on-screen input, at least one candidate word corresponding to the entire input dynamically changes.

  Preferably, the error correction device 1 further comprises a sorting device (not shown).

  The sorting apparatus obtains an average input speed of keys corresponding to each input code in the entire input code based on an input time of the entire input code corresponding to the one entire input, and corresponds to the entire input. Adjust the sort of at least one candidate word. Specifically, after one complete input by the user is completed, the sorting apparatus is based on the input time of the entire input code corresponding to the acquired current entire input and the number of input codes in the current entire input process. The average input speed of the key corresponding to each input code in the current overall input code is calculated, and then the sort of the at least one candidate word corresponding to the overall input is adjusted according to the average input speed.

  Preferably, the sorting device comprises a calculation unit (not shown), a first determination unit (not shown), and a first adjustment unit (not shown).

  Further, the calculation unit calculates the average input speed of the key corresponding to each input code in the entire input code based on the input time of the entire input code. Specifically, the calculation unit may calculate the average input speed of the key corresponding to each input code in the overall input code based on the input time of the overall input code and the number of input codes included in the overall input code. calculate.

The first determination unit determines a corresponding error correction model based on the average input speed. The error correction model is 1) an error correction level model based on the average input speed alone, 2) an error correction score model based on the average input speed alone, 3) fitting the average input speed with other environmental parameters Error correction level models, 4) including, but not limited to, error correction score models by fitting the average input speed with other environmental parameters. The first adjusting unit adjusts the sort of the at least one candidate word corresponding to the general input based on the error correction model. Specifically, when an error correction level model with a single average input speed is adopted as the error correction model, for example, the level of the average input speed is divided into high level, middle level and low level. When the average input speed is low, each input code in the overall input increases the weight to be the input code initially acquired by the acquisition device 101, and each input code in the overall input is erroneous By decreasing the weight to be the input code after correction, the candidate word corresponding to the input code in the entire input becoming the input code initially acquired by the acquisition device 1 is positioned at the head. Similarly, when the average input speed is at a high level, weights for each input code in the overall input to become an input code after error correction are increased, and each input code in the overall input is acquired by the acquisition device 1 By decreasing the weight to be the input code acquired first, the candidate word corresponding to each input code in the entire input becoming the input code after error correction is positioned at the head. For example, when "shoifashi" is input by the user, when the average input speed is at a high level, this means that the probability of erroneous input in the entire input is high at this time, and the input code after error correction "shoufashi" Corresponding candidate word, ie
Candidate words with high degree of correlation with “shoufashi”, etc. at the beginning of the candidate word list, and candidate words corresponding to the initial overall input code “shoifashi” not subjected to error correction processing, ie
Position etc after the candidate word list. Similarly, if the average input speed is low, replace the sort. When the error correction score model with the single average input speed is adopted for the error correction model, the score of the error correction score is a continuous linear function or non-linear function related to the average input speed, and the average input Different as the speed changes. The first adjustment unit acquires a corresponding error correction score based on the magnitude of the input speed, and thereafter an input code for which each input code in the entire input is first acquired by the acquisition device 1 based on the score And each input code in the entire input is an input code after error correction, and the sort is determined in the candidate word list of candidate words corresponding to both. When an error correction level model by fitting the average input speed and other environmental parameters is adopted for the error correction model, the error correction level corresponds based on the fitting of the average input speed and other environmental parameters. The error correction level is determined, and based on the error correction level, each input code in the overall input is an input code initially acquired by the acquisition device 1, a weight, and each input code in the overall input is erroneous The weight which is the input code after correction is adjusted, and the sort in the candidate word list of the candidate word corresponding to both is decided. When an error correction score model by fitting the average input speed and other environmental parameters is adopted as the error correction model, the score of the error correction score relates to the fitting of the average input speed with other environmental parameters. It is a continuous linear or non-linear function, which varies according to the change of the average input speed or other environmental parameters. Thereafter, based on the score, each input code in the entire input is a weight that is an input code initially acquired by the acquisition device 1, and each input code in the entire input is an error-corrected input code The weights are adjusted, and the sort in the candidate word list of candidate words corresponding to both is determined.

  Preferably, the sorting apparatus further comprises a noise cancellation unit (not shown), a second determination unit (not shown), and a second adjustment unit (not shown).

  The noise cancellation unit performs noise cancellation processing on a sequence consisting of key-in (input) timing for each input code in the entire input code to obtain a noise cancellation time series. Specifically, one sequence is formed by the timing (time order) of the input for each input code in one full input by the user, and the noise elimination process is performed on the sequence. For example, when an input code is input by the user and then the input process is stopped under other circumstances, the timing corresponding to the input code far exceeds the upper threshold of the predetermined timing at the time of normal input, and from the sequence The timing corresponding to the input code is erased. Preferably, the noise cancellation process 1) erases the timing when the timing of a certain input code is larger than the upper threshold of the predetermined timing or smaller than the lower threshold of the predetermined timing; 2) Erasing the timing when the ratio of the timing of a certain input code to the value of the current input average timing is larger than the upper threshold of the predetermined ratio or smaller than the lower threshold of the predetermined ratio; 3) deleting the timing when the map value of the fitting function of the input code with the current input environment is larger than a predetermined map upper threshold or smaller than a predetermined map lower threshold; At least one of

  The second determination unit determines the input time of the entire input code based on the noise cancellation time series. Specifically, the input time of the overall input code should be equal to the sum of all the timings in the noise elimination time series, and the number of timings erased by the noise elimination process should be removed from the number of overall input codes.

  The second adjusting unit obtains the overall input speed of the overall input code and the sort of the at least one candidate word corresponding to the overall input based on the input time of the overall input code. Specifically, the second adjustment unit obtains the average input speed of the overall input code based on the number of input codes corresponding to the input time of the overall input code and the number of noise cancellation input codes removed. Adjusting the sort of the at least one candidate word corresponding to the global input.

  Preferably, the error correction device 1 further comprises a first optimization device (not shown). The first optimization device determines the user group in which the user is located based on the characteristics of the user of the trigger key, and optimizes the error correction device 1 based on the user group information. Specifically, the error correction device 1 draws the user according to each feature of the user by learning for a predetermined time, and likes the user group corresponding to the drawn image by the big data, for example, rapid input. Mining young men, middle-aged women who like to seriously input. Each feature of the user includes the user's overall input speed, the input speed for each user's specific key, the user's own manufactured word, the user's frequent input, etc. It is not limited. The first optimization device optimizes the error correction device 1 by performing threshold adjustment of each term with respect to the error correction device 1 by a feedback system.

  The error correction device 1 further includes a second optimization device (not shown). The second optimization apparatus acquires customary information of the user's input history, and optimizes the error correction apparatus based on the customary information of the input history. Specifically, the error correction device 1 learns a predetermined time and based on the user's drop point situation, the user's input practice and method, for example, whether to get used to one-handed input or two-handed input, horizontal screen input Whether you can get used to input on the vertical screen or the vertical screen. The drop point situation of the user includes, but is not limited to, the degree of discreteness of the drop point of the input by the user, the input speed of the specific key pair, the false touch rate of the input of the specific key, and the like. The second optimization device acquires customary information of the user's input history and optimizes the error correction device 1 by adjusting the threshold of each term in the error correction device 1 according to the input convention and system.

  FIG. 3 is a flowchart illustrating an error correction method of an input method based on a user's input speed according to an aspect of the present invention.

  In step S301, the error correction device 1 acquires the current input speed of the single key triggered. Specifically, when an input operation is performed by the user and a key on the keyboard of the user device is pressed or touched, in step S301, the error correction device 1 interacts with the user device, for example, to the input application. By calling the provided application interface (API) one or more times, the user's current single key input speed triggered by the user collected by the user device sensor is obtained. Also, the keyboard includes, but is not limited to, an actual keyboard and a virtual keyboard on a user device screen.

  It should be understood by those skilled in the art that the false touch rate corresponding to the input speed is merely exemplary, and the corresponding map relationship between the actual input speed and the false touch rate is analyzed and statistically analyzed for big data. It is the numerical value of the wrong touch.

  In step S302, the error correction device 1 adjusts the weight when the input code corresponding to the current single key is corrected to the input code corresponding to another key based on the input speed. Specifically, the false touch rate corresponding to the input time of a single key is statistics based on research and experiments, with the time of single input for each key as a group box. Also, the single key input time is an interval of time at which two adjacent single keys are triggered. FIG. 2a shows the relationship between the false touch rate and the single key input time, and the single key input speed can be calculated based on the single key input time, ie, input speed = 1 / input time It is. As shown in FIG. 2a, the false touch rate gradually decreases as the single key input speed decreases, ie, as the single key input time increases. That is, as the user's input speed is faster, an input error is more likely to occur. Therefore, in step S302, the error correction device 1 adjusts the weight when the input code corresponding to the current single key is corrected to the input code corresponding to the other key based on the input speed, The weight is applied to the existing error correction model to support error correction by the existing error correction model, thereby improving the accuracy of user input. For example, if the input speed of a single key is relatively fast, in step S302, the error correction device 1 inputs the input code corresponding to the current single key to another key based on the input speed. If the weight w corrected to the code is increased, or if the input speed of a single key is slow, the error correction device 1 selects an input corresponding to the current single key based on the input speed in step S302. Reduce the weight w that the code is corrected to the input code that corresponds to the other key. The specific numerical value of the weight w may be set according to a predetermined rule. For example, the total weight value tw is set to 1, w is an arbitrary value between 0-1 and a specific value of w is determined according to a predetermined rule.

  Note that adjusting the weight when the input code corresponding to the current single key is corrected to the input code corresponding to another key based on the input speed is 1) prone to false touch Increasing the weight when the input code corresponding to the key is corrected to the input code corresponding to the other key, and 2) the input code corresponding to the key where the erroneous touch is likely to occur correspond to the other key Decrease own weight when corrected to code, 3) increase own weight when input code corresponding to a key that is less likely to be erroneously corrected to input code corresponding to another key And 4) reducing the weight when an input code corresponding to a key that is unlikely to cause an accidental touch is corrected to an input code corresponding to another key, but is not limited thereto. Specifically, if the input speed is relatively high, for example, the input speed of the current single key exceeds a predetermined speed threshold vf, the current single key is a key which is likely to cause an erroneous touch. Thus, the weight at which a key prone to false touch is corrected to another key is increased, and the weight of the current single key itself is decreased. Note that the weight is a first weight, and the first weight indicates the weight of the current single key and the weight with which the current single key is corrected to another key. For example, taking the QWERTY full keyboard as an example, the current single key of the obtained user is "r". At this time, if the current single key "r" is a key prone to an erroneous touch, that is, if the current single key input speed is relatively fast, the user's desired trigger key is The first weight that becomes "r" itself is decreased, and the first weight that the user's desired trigger key becomes "r" 's peripheral keys "e", "f", "t" is increased respectively. Note that the key of the user's desired trigger is the user's purpose key, the current single key is the trigger key acquired by the user device, and the current single key is not necessarily the user's purpose key. is not. Thus, the current single key itself also has a corresponding first weight so that the error correction model can combine other parameters based on the first weight of each key to determine the user's purpose key. There is. On the other hand, if the input speed of the current single key is lower than the predetermined speed threshold vs, that is, if the current input speed of the single key is relatively low, then the current single key is It is a key that is unlikely to cause an accidental touch, and the first weight of the key that is unlikely to cause an accidental touch is increased, and the input code corresponding to the key that is unlikely to cause an accidental touch is Reduce the first weight if it is corrected. For example, if the current single key of the obtained user is "i" and the current single key "i" at this time is a key that is unlikely to cause an erroneous touch, Increase the first weight where the key is "i" itself and decrease the first weight where the user's desired trigger key is the "r" peripheral key "u", "k", "o" Do.

  Preferably, the determination process of the occurrence of the key mistouch includes, but is not limited to:

  1) Judgment based on the level hierarchy of single key input speed. For example, according to the speed range of input speed, the level of input speed of single key is divided into high level and low level. When the input speed of the single key is high, a second weight corresponding to the high level is obtained for the single key, and the single key is erroneously touched based on the second weight. It is possible to determine whether it is a key that is hard to occur or a key that is prone to false touch. When the input speed of the single key is at a low level, a second weight corresponding to the low level is obtained for the single key, and the single key is erroneously touched based on the second weight. It is possible to determine whether it is a key that is hard to occur or a key that is prone to false touch. The level at which the single key input speed is hierarchized according to the range of the input speed may be other plural means.

  2) Judgment based on score weight of single key input speed. Specifically, when a single key is triggered by the user, the error correction device 1 acquires the current input speed of the single key triggered in step S301, and the single key is input based on the input speed of the single key. Take a score for one key. For example, by setting a function having an input speed as a variable, a score corresponding to each different input speed is calculated, and a second weight corresponding to the input speed of the single key is obtained based on the score. Based on the second weight, it is determined whether the single key is a key that is likely to cause an erroneous touch or a key that is unlikely to cause an erroneous touch.

  3) Determine based on the input speed of a single key and the level of the fitting function map of the current input environment. Specifically, a fitting function is generated based on two parameters of the input speed of the single key and the current input environment, and it is integrally determined which level the single key is at. It is determined whether the current single key is a key that is prone to an erroneous touch or a key that is unlikely to cause an accidental touch according to the level. The current input environment is, for example, whether or not the light beam is sufficient, the current input device is a personal computer, a tablet or a mobile phone, or the user device is in a stationary state or in a moving state. Or, what is the movement speed.

  4) Determine based on the input speed of a single key and the score weight of the fitting function map of the current input environment. Specifically, the single key is scored according to the input speed of the single key and the current input environment. For example, by setting a fitting function having an input speed and a current input environment as variables, a score corresponding to each single key is calculated in different input speeds and different input environments, and the score is calculated according to the score. Acquire a second weight corresponding to the input speed of a single key, and based on the second weight, whether the single key is a key prone to false touch or a key unlikely to false touch To judge.

  Those skilled in the art should understand that the means for adjusting the weight when the input code corresponding to the current single key is corrected to the input code corresponding to the other key based on the input speed is merely illustrative. Means for adjusting the weight when an input code corresponding to the current single key is corrected to an input code corresponding to another key based on the input speed that can appear in the future or in the future If it is applied to the present invention, it is included in the protection scope of the present invention, and is included in the form incorporated herein.

  In step S303, the error correction device 1 performs error correction on the input code corresponding to the current single key based on the adjusted weight, and acquires at least one input code after error correction. . Specifically, in step S303, the error correction device 1 corrects the error correction model corresponding to the input method corresponding to the error correction device 1 based on the weight which is the first weight. Here, by using the current single key input speed triggered as one parameter of the error correction model, the error correction model can be evaluated and calculated from the viewpoint of time when performing error correction. become. In addition, for the input code corresponding to the current single key, combined with at least one non-on-screen triggered historical key belonging to the same one general input as the current single key. The error correction is performed to obtain at least one input code after the error correction. Here, the one full input consists of the current single key and the non-on-screen triggered triggered history key. That is, for all input codes that are not input on screen, the candidate word corresponding to the triggered history key that is not input on screen also changes each time a single additional key is input. For example, it becomes like the input process shown by FIG. 2 b. FIG. 2 b is an example of the input display, and one whole input is “shoifashi”. If the user continuously inputs "zainali" without performing on-screen input, then the entire input at this time becomes "shoifashizainali".

  For example, in step S302, if the weight when the input code corresponding to the current single key adjusted by the error correction device 1 is corrected to the input code corresponding to another key is relatively large, error correction The rate at which the input code corresponding to the current single key is corrected to another input code by the device 1 is increased, and the current single key is corrected to the other key by the error correction model corresponding to the input method The probability of being An appropriate existing error correction model is selected according to the adjusted weight, the adjusted weight is used for the existing error correction model, and error correction is performed with other parameters in the existing error correction model. That is, the error correction is performed from the temporal point of view, the current single key input speed is introduced into the existing error correction model as a factor of the existing error correction model, and the error correction is performed correctly on the input code. In S303, the error correction device 1 obtains at least one input code after the error correction by the error correction model. The at least one input code after the error correction may still be the input code corresponding to the current single key, or the input code corresponding to the current single key may be another key. It may be corrected to the corresponding input code. For example, when the input interface is a touch-type virtual QWERTY full keyboard interface, the error correction device 1 detects all the keys in the keyboard according to the keyboard drop point position L triggered by the user's input. Calculate the distances s1, s2 ... sn from the center point to the drop point position L, calculate the probability that the current single key triggered by the user becomes each key according to the distance, and thereafter calculate the probability Based on the probabilities, or in combination with other parameters, it is obtained that the input code corresponding to the current single key triggered by the user is "s". And, the input speed of "s" is 9 alphabets / second, and in step S301, the error correction device 1 acquires the input speed of the input code "s" as 9 alphabets / second. For example, if the threshold of the input speed is 5 alphabets / second, the current single key may be selected when the input speed of the input code exceeds the input speed threshold, ie, when the input speed belongs to the high level. The weight is increased if the corresponding input code is corrected to the input code corresponding to the other key, and the input code corresponding to the current single key is corrected to the self “s”, ie If the user's objective key still reduces the weight of "s" itself, in step S302, the error correction device 1 determines that the input speed exceeds the input speed threshold according to the input speed, "s" Increases the weight corrected to “w” to w 1, the weight “s” corrects to “a” increases to w 2, the weight corrected to “d” to w 3 Increase the weight corrected to "x" to w4, decrease the weight whose user's purpose key is still "s" itself to w5, and in step S303, the error correction device 1 adds the weights w1, w2,. Integrally determine the probability of correcting the current single key to another key based on w3, w4, w5, and the previous history key of the current single key, eg "s" is "w" by operation The probability of being corrected to “a”, “d”, and “x” is P1, P2, P3, and P4, respectively, and the probability that the user's objective key is “s” is P5, and P1, P2, and so on. The user's objective key is determined as "a" based on P3, P4 and P5. Integrally determine the probability that the current single key will be corrected to another key based on the current single key previous on-screen non-history key, and combine it with the error correction method of the existing error correction model Perform error correction on the input code corresponding to the current single key. For example, if the error-corrected input alphabet before "s" is the initial "p", both "p" and "s" are adjacent to "s" in the alphabet "s" keyboard because both are vocals. The weights corrected to the prosodic "a" increase.

  Those skilled in the art should understand how to adjust the weight when the input code corresponding to the current single key is corrected to the input code corresponding to the other key based on the input speed threshold. The means for determining if is only exemplary and the weight when the input code corresponding to the existing single key which can appear in the future or in the future is corrected to the input code corresponding to the other key A means for determining how to adjust, if applicable to the present invention, is included in the protection scope of the present invention and included here in the form of incorporation. Here, when the error correction device 1 acquires at least one input code after error correction in step S303, it is registered only in the background of the system without displaying it on the input interface, that is, displayed on the input interface. Is the input code that still corresponds to the key triggered by the user.

  Here, by analyzing the big data, the relationship between the user's input speed and the error correction is mined from a temporal point of view to support accurate error correction by the error correction model. Also, the error correction device 1 continuously obtains the input speed for each single key input by the user, and the input code corresponding to the current single key corresponds to the other key according to the input speed. The input method is evaluated from the time point of view by setting the input speed as one of the factors by which the input code is corrected by the error correction model by adjusting the weight in the case of correction to the corresponding input code. Further corrects the input code determined based on the spatial thing of the keyboard's falling point triggered by the SIC, thereby improving the accuracy rate of the error correction by the error correction model and the quality of the user input. Improved.

In step S304, the error correction device 1 determines at least one candidate word corresponding to the entire input, based on at least one input code after error correction corresponding to each key in one entire input. Specifically, in step S303, the error correction device 1 sequentially acquires at least one input code after error correction, and in step S303, the error correction device 1 performs error correction corresponding to each key of one entire input. In step S304, the error correction device 1 determines at least one candidate word corresponding to the entire input. For example, following the above example, after correcting the alphabet "s" to the prosody "a" adjacent to the "s" in the keyboard, "pa" is composed of the alphabet "a" and the previous alphabet "p". In step S304, the error correction device 1 transmits at least one candidate word corresponding to "pa", for example,
Confirm Furthermore, in step S304, the error correction device 1 determines the corresponding at least one candidate word based on the at least one input code before the error correction, and at least the at least one input code before the error correction. The priority grade of one candidate word is lower than at least one candidate word corresponding to at least one input code after error correction. For example, at least one candidate word corresponding to the input code "ps" before error correction
Priority grade such as
Lower than.

  The above input process is continuous, and the user's input continues. In the process of inputting the input code, as long as the on-screen input is not executed, the input code included in the current one full input increases rapidly and the newly increased input code is error corrected. By combining with the input history code that is not on-screen input, at least one candidate word corresponding to the entire input dynamically changes.

  Preferably, the error correction method further includes step S305 (not shown).

  In step S305, the error correction device 1 obtains the average input speed of the key corresponding to each input code in the overall input code based on the input time of the overall input code corresponding to the one overall input, and the overall Adjust the sort of the at least one candidate word corresponding to the input. Specifically, after one complete input by the user is completed, the error correction device 1 receives the input time of the entire input code corresponding to the acquired current entire input in step S305 and the input code in the current entire input process. Calculating the average input speed of the key corresponding to each input code in the entire input code of this time based on the number and the sort of the at least one candidate word corresponding to the entire input according to the average input speed adjust.

  Preferably, step S305 includes sub-step S3051 (not shown), sub-step S3052 (not shown), and sub-step S3053 (not shown).

  In addition, in sub-step S3051, the error correction device 1 calculates the average input speed of the key corresponding to each input code in the entire input code based on the input time of the entire input code. Specifically, in the sub-step S3051, the error correction device 1 generates a key corresponding to each input code in the overall input code based on the input time of the overall input code and the number of input codes included in the overall input code. Calculate the average input speed of

  In substep S3052, the error correction device 1 determines a corresponding error correction model based on the average input speed. The error correction model is 1) an error correction level model based on the average input speed alone, 2) an error correction score model based on the average input speed alone, 3) fitting the average input speed with other environmental parameters Error correction level models, 4) including, but not limited to, error correction score models by fitting the average input speed with other environmental parameters.

In sub-step S3053, the error correction device 1 adjusts the sort of the at least one candidate word corresponding to the entire input based on the error correction model. Specifically, when an error correction level model with a single average input speed is adopted as the error correction model, for example, the level of the average input speed is divided into high level, middle level and low level. When the average input speed is at a low level, each input code in the overall input increases the weight which is the input code initially acquired in step S301 by the error correction device 1, and each input code in the overall input The candidate word corresponding to the fact that each input code in the entire input is the input code initially acquired by the error correction device 1 is placed at the beginning by decreasing the weight where each of the input codes is error corrected. Let Similarly, when the average input speed is at a high level, each input code in the overall input increases a weight that is an input code after error correction, and each input code in the overall input is an error correction device 1 A candidate word corresponding to the fact that each input code in the entire input is an error-corrected input code is located at the top by decreasing the weight which is the input code initially obtained by the. For example, when "shoifashi" is input by the user, when the average input speed is at a high level, this means that the probability of erroneous input in the entire input is high at this time, and the input code after error correction "shoufashi" Corresponding candidate word, ie
Candidate words with high degree of correlation with “shoufashi”, etc. at the beginning of the candidate word list, and candidate words corresponding to the initial overall input code “shoifashi” not subjected to error correction processing, ie
Position etc after the candidate word list. Similarly, replace the sort if the average input speed is at a high level. When the error correction score model with the single average input speed is adopted for the error correction model, the score of the error correction score is a continuous linear function or non-linear function related to the average input speed, and the average input Different as the speed changes. In sub-step S3053, the error correction device 1 obtains a corresponding error correction score based on the magnitude of the input speed, and thereafter, based on the score, each input code in the entire input is first by the error correction device 1 The weight which is the acquired input code and the weight which is the input code after each of the respective input codes in the whole input are error-corrected are adjusted, and the sort in the candidate word list of candidate words corresponding to both is decided. When an error correction level model by fitting the average input speed and other environmental parameters is adopted for the error correction model, the error correction level corresponds based on the fitting of the average input speed and other environmental parameters. The error correction level is determined, and based on the error correction level, each input code in the overall input is a weight that is an input code initially obtained by the error correction device 1, and each input code in the overall input is The weight which is the input code after error correction is adjusted, and the sort in the candidate word list of candidate words corresponding to both is decided. When an error correction score model by fitting the average input speed and other environmental parameters is adopted as the error correction model, the score of the error correction score relates to the fitting of the average input speed with other environmental parameters. It is a continuous linear or non-linear function, which varies according to the change of the average input speed or other environmental parameters. Thereafter, based on the score, each input code in the overall input is a weight that is an input code initially obtained by the error correction device 1, and each input code in the overall input is an input code after error correction A certain weight is adjusted, and the sort in the candidate word list of candidate words corresponding to both is determined.

  Preferably, step S305 further includes sub-step S3054 (not shown), sub-step S3055 (not shown), and sub-step S3056 (not shown).

  In sub-step S3054, the error correction device 1 performs noise cancellation processing on the sequence consisting of the timing of input for each input code in the overall input code to obtain a noise cancellation time series. Specifically, one sequence is formed by the timing of the input for each input code in one complete input by the user, and noise elimination processing is performed on the sequence. For example, when an input code is input by the user and then the input process is stopped under other circumstances, the timing corresponding to the input code far exceeds the upper threshold of the predetermined timing at the time of normal input, and from the sequence The timing corresponding to the input code is erased. Preferably, the noise cancellation process 1) erases the timing when the timing of a certain input code is larger than the upper threshold of the predetermined timing or smaller than the lower threshold of the predetermined timing; 2) Erasing the timing when the ratio of the timing of a certain input code to the value of the current input average timing is larger than the upper threshold of the predetermined ratio or smaller than the lower threshold of the predetermined ratio; 3) deleting the timing when the map value of the fitting function of the input code with the current input environment is larger than a predetermined map upper threshold or smaller than a predetermined map lower threshold; At least one of

  In sub-step S3055, the error correction device 1 determines the input time of the entire input code based on the noise cancellation time series. Specifically, the input time of the overall input code should be equal to the sum of all the timings in the noise elimination time series, and the number of timings erased by the noise elimination process should be removed from the number of overall input codes.

  In sub-step S3056, the error correction device 1 obtains the overall input speed of the overall input code and the sort of the at least one candidate word corresponding to the overall input based on the input time of the overall input code. Specifically, in sub-step S3056, the error correction device 1 determines the average input of the total input code based on the number of input codes corresponding to the input time of the total input code and the number of noise cancellation input codes removed. Get the speed and adjust the sort of the at least one candidate word corresponding to the global input.

  Preferably, the method further comprises step S306 (not shown). In step S306, the error correction device 1 determines the user group in which the user is located based on the characteristics of the user of the trigger key, and optimizes the error correction device 1 based on the user group information. Specifically, the error correction device 1 draws the user according to each feature of the user by learning for a while, and a user group corresponding to the image according to the big data, for example, a young man who likes rapid input, seriousness Mining middle-aged women who like to enter. Each feature of the user includes, but is not limited to, the user's overall input speed, the input speed for each user's specific key, words manufactured by the user, frequent input of the user, etc. I will not. In step S306, the error correction device 1 optimizes the error correction device 1 by performing threshold adjustment of each term with respect to the error correction device 1 using a feedback system.

  The method further includes step S307 (not shown). In step S307, the error correction device 1 acquires customary information of the user's input history, and optimizes the error correction device based on the customary information of the input history. Specifically, the error correction device 1 learns for a while, and based on the user's drop point situation, the user's input customs and methods, for example, whether to get used to one-handed input or two-handed input, horizontal screen input and It is determined which of the inputs on the vertical screen can be used. The drop point situation of the user includes, but is not limited to, the degree of discreteness of the drop point of the input by the user, the input speed of the specific key pair, the false touch rate of the input of the specific key, and the like. In step S307, the error correction device 1 acquires customary information of the user's input history, and adjusts the threshold of each item in the error correction device 1 according to the input convention and method to optimize the error correction device 1 Do.

  The invention further provides a computer readable storage medium. Computer code is stored on the computer readable storage medium. When the computer code is executed, any one of the above methods will be executed.

  The invention further provides a computer program product. Where the computer program product is executed by a computer device, any one of the above methods will be performed.

The invention further provides a computing device. The computer device is
One or more processors,
And a memory for storing one or more computer programs.

  When the one or more computer programs are executed by the one or more processors, the one or more processors are caused to execute any one of the methods.

  It should be noted that the present invention may be implemented by software and / or a combination of software and hardware. For example, each device of the present invention may be implemented employing a dedicated integrated circuit (ASIC) or any other similar hardware device. In one embodiment, the software program of the present invention can implement the steps or functions by being executed by a processor. Similarly, the software program (including associated data structures) of the present invention may be stored on a computer readable storage medium such as a RAM memory, a magnetic or optical driver or floppy disk and similar devices. Also, some steps or functions of the present invention may be realized by hardware, for example, a circuit that cooperates with a processor to perform each step or function.

  As will be appreciated by those skilled in the art, the present invention is not limited to the details of the above-described exemplary embodiments, but can be embodied in other embodiments without departing from the spirit or basic characteristics of the present invention. It may be done. Thus, in any case, the examples are illustrative and non-restrictive. Since the scope of the present invention is not limited by the above description but by the appended claims, it is intended that all changes within the meaning and scope of equivalent elements falling within the scope of the claims be included in the present invention. Become. Not all codes in the claims are limited to the related claims. Also, obviously, the word "comprising" does not exclude other units or steps, and the singular does not exclude a plurality. A plurality of units or devices described in the system claims may be realized by one unit or device via software or hardware. The first, second, etc. words represent names and do not indicate any particular sort.

Claims (23)

An error correction method of an input method based on a user's input speed,
Obtaining (a) the current single key input speed triggered;
Adjusting a weight when an input code corresponding to the current single key is corrected to an input code corresponding to another key according to the input speed;
Performing error correction on the input code corresponding to the current single key based on the adjusted weight to obtain at least one input code after error correction;
According to at least one of the input code after error correction corresponding to each key in a single entire input, and step (d) to determine at least one candidate word corresponding to the entire input, only including,
The step (c) is
Comprehensively determining a probability of correcting the current single key to another key based on the adjusted weight and the previous history key of the current single key, c1.
Performing error correction on the input code corresponding to the current single key on the basis of the probability to obtain at least one input code after error correction, at least after the error correction One input code may still be the input code corresponding to the current single key, or the input code corresponding to said current single key is corrected to the input code corresponding to the other key An error correction method , including step c2, which may be The error correction method is
The average input speed of the key corresponding to each input code in the overall input code is acquired according to the input time of the overall input code corresponding to the one overall input, and the at least one candidate corresponding to the overall input is acquired The error correction method according to claim 1, further comprising the step (e) of adjusting the sorting of words. In the step (e),
Calculating the average input speed of a key corresponding to each input code in the overall input code, according to the input time of the overall input code (e1);
Determining a corresponding error correction model based on the average input speed (e2);
Adjusting the sorting of the at least one candidate word corresponding to the global input according to the error correction model (e3);
The error correction method according to claim 2, comprising Adjusting a weight when an input code corresponding to the current single key is corrected to an input code corresponding to another key according to the input speed;
Increasing the weight when an input code corresponding to a key prone to an erroneous touch is corrected to an input code corresponding to another key;
Reducing its own weight when an input code corresponding to a key susceptible to an erroneous touch is corrected to an input code corresponding to another key;
Increasing an own weight when an input code corresponding to a key that is unlikely to cause an erroneous touch is corrected to an input code corresponding to another key;
Reducing the weight when an input code corresponding to a key that is unlikely to cause an accidental touch is corrected to an input code corresponding to another key;
The error correction method according to any one of claims 1 to 3, comprising at least one of the following. Judgment of the degree of difficulty of false touch of the key
Judgment according to the level hierarchy of single key input speed,
Judging according to the score weight of single key input speed,
Judging according to the level of the fitting function map of the single key input speed and the current input environment,
Judging according to the score weight of the fitting function map of the single key input speed and the current input environment,
The error correction method according to claim 4, comprising at least one of the following. Determining a corresponding error correction model based on the average input speed (e2),
An error correction level model with a single said average input speed,
An error correction score model with a single said average input speed,
Error correction level model by fitting the average input speed with other environmental parameters,
An error correction score model by fitting the average input speed with other environmental parameters;
The error correction method according to claim 3 , comprising at least one of the following. In the step (e),
Noise elimination processing is performed on a sequence consisting of key-in timing of each input code in the entire input code to obtain a noise elimination time series;
The input time of the entire input code is determined according to the noise elimination time series,
The average input speed of the overall input code is obtained according to the input time of the overall input code, and sorting of the at least one candidate word corresponding to the overall input is adjusted.
The error correction method according to claim 2 or 3 , further comprising: The noise elimination process is
Erasing the timing when the key-in timing of the input code is larger than the upper threshold of the predetermined timing or smaller than the lower threshold of the predetermined timing;
Deleting the timing when the ratio of the key-in timing of the input code to the value of the timing of the current input average is larger than the upper threshold of the predetermined ratio or smaller than the lower threshold of the predetermined ratio; ,
Deleting the timing when the map value of the input code with the fitting function of the current input environment is larger than the upper threshold of the predetermined map or smaller than the lower threshold of the predetermined map;
The error correction method according to claim 7, comprising at least one of the following. The error correction method is
Determine the user group where the user is located according to the characteristics of the user of the trigger key,
Further comprising optimizing the error correction method according to information of the user group
The error correction method according to any one of claims 1 to 8. The error correction method is
Get custom information of user's input history,
The method further includes optimizing the error correction method according to the input information of the input history.
The error correction method according to any one of claims 1 to 9. An error correction apparatus for an input method based on a user's input speed, comprising:
An acquisition device for acquiring the current single key input speed triggered;
An adjusting device for adjusting a weight when an input code corresponding to the current single key is corrected to an input code corresponding to another key according to the input speed;
An acquisition device that performs error correction on an input code corresponding to the current single key based on the adjusted weight, and acquires at least one input code after error correction;
According to at least one of the input code after error correction corresponding to each key in a single entire input, e Preparations and determine device, the determining at least one candidate word corresponding to the overall input,
The acquisition device is
A device that integrally determines a probability of correcting a current single key to another key based on the adjusted weight and a history key before the current single key;
An apparatus for performing error correction on an input code corresponding to the current single key based on the probability and acquiring at least one input code after error correction, the at least one after the error correction One input code may still be the input code corresponding to the current single key, or the input code corresponding to said current single key corrected to the input code corresponding to the other key May be, including devices,
Error correction device of input method based on user's input speed. The error correction device
The average input speed of the key corresponding to each input code in the overall input code is obtained according to the input time of the overall input code corresponding to the one overall input, and the at least one corresponding to the overall input is acquired The apparatus further comprises a sorting device that adjusts sorting of candidate words.
The error correction device according to claim 11. The sorting device
A calculation unit for calculating the average input speed of the key corresponding to each input code in the entire input code according to the input time of the entire input code;
A first determining unit for determining a corresponding error correction model based on the average input speed;
A first adjusting unit for adjusting the sort of the at least one candidate word corresponding to the entire input according to the error correction model;
The error correction device according to claim 12, comprising: Adjusting the weight when the input code corresponding to the current single key is corrected to the input code corresponding to the other key according to the input speed is:
Increasing the weight when an input code corresponding to a key prone to an erroneous touch is corrected to an input code corresponding to another key;
Reducing its own weight when an input code corresponding to a key susceptible to an erroneous touch is corrected to an input code corresponding to another key;
Increasing an own weight when an input code corresponding to a key that is unlikely to cause an erroneous touch is corrected to an input code corresponding to another key;
Reducing the weight when an input code corresponding to a key that is unlikely to cause an accidental touch is corrected to an input code corresponding to another key;
The error correction device according to any one of claims 11 to 13, comprising at least one of the following. Judgment of the occurrence difficulty level of the key false touch is
Judgment according to the level hierarchy of single key input speed,
Judging according to the score weight of single key input speed,
Judging according to the level of the fitting function map of the single key input speed and the current input environment,
Judging according to the score weight of the fitting function map of the single key input speed and the current input environment,
The error correction device according to claim 14, comprising at least one of the following. Determining a corresponding error correction model based on the average input speed is:
An error correction level model with a single said average input speed,
An error correction score model with a single said average input speed,
Error correction level model by fitting the average input speed with other environmental parameters,
An error correction score model by fitting the average input speed with other environmental parameters;
The error correction device according to claim 13 , comprising at least one of the following. The sorting device further comprises
A noise cancellation unit for performing a noise cancellation process on a sequence consisting of key-in timings of input codes in the entire input code to obtain a noise cancellation time series;
A second determining unit for determining an input time of the entire input code according to the noise elimination time series;
And a second adjusting unit for acquiring the overall input speed of the overall input code and adjusting the sort of the at least one candidate word corresponding to the overall input in accordance with the input time of the overall input code. The error correction device according to claim 12 or 13 . The noise elimination process is
Deleting the timing when the key-in timing of the input code is greater than the upper threshold of the predetermined timing or smaller than the lower threshold of the predetermined timing;
Deleting the timing when the ratio of the key-in timing of the input code to the value of the average timing of the current input is larger than the upper threshold of the predetermined ratio or smaller than the lower threshold of the predetermined ratio;
Deleting the timing when the map value of the fitting function of the input code with the current input environment is larger than the upper threshold of the predetermined map or smaller than the lower threshold of the predetermined map;
The error correction device according to claim 17, comprising at least one of the following. The error correction device further comprises
A first optimization device is provided that determines a user group in which the user is located according to the characteristics of the user of the trigger key, and optimizes the error correction device according to the information of the user group. The error correction device described in one. The error correction device
The error according to any one of claims 11 to 19, further comprising a second optimization device for acquiring customary information of the user's input history and optimizing the error correction device according to the customary information of the input history. Correction device. Computer code is stored,
A computer readable storage medium on which the method of error correction according to any of claims 1 to 10 is carried out when said computer code is executed. By being executed by a computer device, to realize the error correction method according to any one of claims 1 to 10, the computer program. One or more processors,
A memory in which one or more computer programs are stored,
The error correction method according to any one of claims 1 to 10 is implemented by the one or more processors when the one or more computer programs are executed by the one or more processors. Computer device.