CN105912697B - A kind of optimization method and device of conversational system knowledge base - Google Patents
A kind of optimization method and device of conversational system knowledge base Download PDFInfo
- Publication number
- CN105912697B CN105912697B CN201610262475.7A CN201610262475A CN105912697B CN 105912697 B CN105912697 B CN 105912697B CN 201610262475 A CN201610262475 A CN 201610262475A CN 105912697 B CN105912697 B CN 105912697B
- Authority
- CN
- China
- Prior art keywords
- information
- answer
- user
- answer information
- question
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A kind of optimization method and device of conversational system knowledge base, this includes: target question and answer information acquiring step, judges whether the assessment parameter of question and answer information to be analyzed meets optimisation criteria, such as meets, then is analysed to question and answer information as target question and answer information；Update answer set generation step, the problem of based on target question and answer information the corresponding answer information set of acquisition of information, the degree of correlation parameter of each answer information in answer information set is calculated separately, and parameter generates the update answer set for problem information in target question and answer information according to the degree of correlation.This method can be realized the update of the active to conversational system database, so that conversational system, which can export, is more in line with user's use habit and desired answer, and then improve the user experience and user's viscosity of conversational system.
Description
Technical field
The present invention relates to human-computer interaction technique fields, specifically, being related to a kind of optimization method of conversational system knowledge base
And device.
Background technique
For traditional human-computer interaction, human-computer interaction mainly passes through mouse, keyboard and touch screen etc. by user
Device is interacted with equipment such as computer, mobile phones.And between man-machine information interaction amount volatile growth is just being presented,
Traditional man-machine interaction mode greatly affected the efficiency and effect of human-computer interaction.
Most natural, the most convenient and fast interactive mode that people are accustomed to is natural language interaction, thus by conversational system come
Realize that efficiently human-computer interaction becomes more and more common selection.However, some inquiries of the existing conversational system for user
Satisfactory answer can not be provided by asking questions, this also just affects the user experience of entire conversational system, so that the use of product
Family viscosity is poor.
Summary of the invention
To solve the above problems, the present invention provides a kind of optimization methods of conversational system knowledge base, comprising:
Target question and answer information acquiring step, judges whether the assessment parameter of question and answer information to be analyzed meets optimisation criteria, such as
Meet, then using the question and answer information to be analyzed as target question and answer information；
The problem of updating answer set generation step, the being based on the target question and answer information corresponding answer information of acquisition of information
Set, calculates separately the degree of correlation parameter of each answer information in the answer information set, and according to the degree of correlation parameter
Generate the update answer set for problem information in the target question and answer information.
According to one embodiment of present invention, in the target question and answer information acquiring step, judge described to be analyzed ask
Whether the assessment parameter for answering information is less than default assessment threshold value, if it is less, using the question and answer information to be analyzed as target
Question and answer information.
According to one embodiment of present invention, it in the update answer set generation step, is asked according in question and answer information
Whether topic information and answer information have semantic identical centre word to determine the degree of correlation parameter of answer information, wherein semantic
The quantity of identical centre word is more, and the degree of correlation parameter of answer information and problem information is bigger.
According to one embodiment of present invention, when obtaining the answer information set, target is chosen according to preset rules
User, and the problems in target question and answer information information is pushed to the target user, the target user is obtained for institute
The answer information that problem information is fed back is stated, to obtain the question and answer information aggregate.
According to one embodiment of present invention, when choosing the target user,
The user property for obtaining conversational system different user, judges whether user property meets default problem push request,
If met, corresponding user is determined as the target user, wherein the user property includes in item set forth below
Any one or several:
Subscriber identity information, customer position information, age of user information, user gender information；
Or, obtain conversational system different user interaction scenarios and/or interaction topic, judge the interaction scenarios and/or
Whether interaction topic meets default problem push request, if met, corresponding user is determined as the target user.
The present invention also provides a kind of optimization devices of conversational system knowledge base, comprising:
Target question and answer data obtaining module, is used to judge whether the assessment parameter of question and answer information to be analyzed to meet optimization mark
Standard such as meets, then using the question and answer information to be analyzed as target question and answer information；
Acquisition of information is corresponding answers for the problem of updating answer set generation module, being used for based on the target question and answer information
Case information aggregate calculates separately the degree of correlation parameter of each answer information in the answer information set, and according to the correlation
It spends parameter and generates the update answer set for being directed to problem information in the target question and answer information.
According to one embodiment of present invention, the target question and answer data obtaining module is configured to judge described to be analyzed ask
Whether the assessment parameter for answering information is less than default assessment threshold value, if it is less, using the question and answer information to be analyzed as target
Question and answer information.
According to one embodiment of present invention, the update answer set generation module is configured to ask according in question and answer information
Whether topic information and answer information have semantic identical centre word to determine the degree of correlation parameter of answer information, wherein semantic
The quantity of identical centre word is more, and the degree of correlation parameter of answer information and problem information is bigger.
According to one embodiment of present invention, the update answer set generation module is configured to obtaining the answer letter
When breath set, target user is chosen according to preset rules, and push asking in the target question and answer information to the target user
Information is inscribed, the target user is obtained and is directed to the answer information that described problem information is fed back, to obtain the question and answer information
Set.
According to one embodiment of present invention, the update answer set generation module is configured to choosing the target use
When family,
The user property for obtaining conversational system different user, judges whether user property meets default problem push request,
If met, corresponding user is determined as the target user, wherein the user property includes in item set forth below
Any one or several:
Subscriber identity information, customer position information, age of user information, user gender information；
Or, obtain conversational system different user interaction scenarios and/or interaction topic, judge the interaction scenarios and/or
Whether interaction topic meets default problem push request, if met, corresponding user is determined as the target user.
Conversational system knowledge base optimization method provided by the present invention is carried out by user's Request Log to mass users
Analysis, finds question and answer information bad or unreasonable in these logs as far as possible, then initiatively obtains through various channels
The answer information of the problems in this kind of question and answer information information, and the judgement of legitimacy is carried out to answer information, it will finally conform to
The answer information asked is inserted or updated into database.By the update to conversational system database, conversational system can be exported
It is more in line with user's use habit and desired answer, to improve the user experience and user's viscosity of conversational system.
The method achieve the self study of conversational system knowledge base and closed loop feedbacks.Wherein, this method can make pair
Telephone system can carry out rationalization improvement to question and answer information independence bad and unreasonable in knowledge base, so that this is
The quality of system can have the promotion of duration.Meanwhile this method can also make conversational system realize autonomous learning, so also
The manual mode of learning that must rely on maintenance personnel can be detached from.
This method is asked by the way that the problems in target question and answer information information is pushed to user again to obtain user for this
The answer for inscribing information, just realizes the closed loop feedback of " user-conversational system-user ", therefore conversational system is also in this way
The stability of itself is kept when by external interference.It is based on this closed loop configuration, what conversational system can continue connects
Receive the feedback from user's perception level, and continued to optimize according to the feedback of user knowledge library so as to adjust itself
Output, so that this output can meet the expectation of user.
In addition, conversational system knowledge base optimization method provided by the present invention can also by according to the attribute of user come to
Specific user pushes the problems in target question and answer information information.Since the attribute of this kind of user inputs target with to conversational system
The attribute of the user of problem information is same or like in question and answer information, therefore is also obtained with by this kind of user and is directed to mesh
The higher answer information of more accurate or compatible degree of problem information in question and answer information is marked, and is determined according to this kind of answer information
The update answer set of above problem information out also will be more accurate.
Other features and advantages of the present invention will be illustrated in the following description, also, partly becomes from specification
It obtains it is clear that understand through the implementation of the invention.The objectives and other advantages of the invention can be by specification, right
Specifically noted structure is achieved and obtained in claim and attached drawing.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is required attached drawing in technical description to do simple introduction:
Fig. 1 is the implementation flow chart of conversational system knowledge base optimization method according to an embodiment of the invention；
Fig. 2 is the implementation flow chart of conversational system knowledge base optimization method in accordance with another embodiment of the present invention；
Fig. 3 is the implementation flow chart of the conversational system knowledge base optimization method of another embodiment according to the present invention；
Fig. 4 is the structural schematic diagram of conversational system knowledge base optimization device according to an embodiment of the invention.
Specific embodiment
Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings and examples, how to apply to the present invention whereby
Technological means solves technical problem, and the realization process for reaching technical effect can fully understand and implement.It needs to illustrate
As long as not constituting conflict, each feature in each embodiment and each embodiment in the present invention can be combined with each other,
It is within the scope of the present invention to be formed by technical solution.
Meanwhile in the following description, for illustrative purposes and numerous specific details are set forth, to provide to of the invention real
Apply the thorough understanding of example.It will be apparent, however, to one skilled in the art, that the present invention can not have to tool here
Body details or described ad hoc fashion are implemented.
In addition, step shown in the flowchart of the accompanying drawings can be in the department of computer science of such as a group of computer-executable instructions
It is executed in system, although also, logical order is shown in flow charts, and it in some cases, can be to be different from herein
Sequence execute shown or described step.
For the conversational system in intelligent robot, from the perspective of user, better user experience, then just
Mean higher user's viscosity.Currently, experience the most intuitive is the problem of user is inquired for dialogue robot
Answer whether meet the expectation of user.For example, when user initiates inquiry " always alone " to dialogue machine people, it is right
For existing dialogue robot, the answer of feedback is likely to " I don't have the slightest idea about the meaning of your remarks ".It is possible thereby to see
Out, existing dialogue robot can not meet the daily interaction demand of user well.
Talking with robot is to realize that the dialogue between robot and user is interacted using conversational system, wherein dialogue
System determines corresponding answer using the knowledge base of itself the problem of input according to user.The present invention is exactly using above-mentioned
Characteristic provides a kind of optimization method of new conversational system knowledge base, so that conversational system can utilize knowing after optimization
Know the answer that library is more accurate, reasonably determines out the problem of user is inputted.The optimization method by persistently learning incessantly
It practises, the quality of conversational system knowledge base is constantly promoted with this.
It should be pointed out that the promotion of the quality of the conversational system knowledge base of present invention meaning, can both refer in knowledge base
The improvement of answer information corresponding to problem information can also refer to the expansion of problem information and its answer information in knowledge base.
In order to clearly illustrate the realization principle of conversational system knowledge base optimization method provided by the present invention, realize
Process and advantage are further described this method below in conjunction with different embodiments.
Embodiment one:
Fig. 1 shows the implementation flow chart of the optimization method of conversational system knowledge base provided by the present embodiment.
As shown in Figure 1, method provided by the present embodiment obtains the survey of question and answer information to be analyzed in step s101 first
Comment parameter.Wherein, accessed assessment parameter is preferably question and answer evaluation system to be analyzed to this method in step s101
The scoring of question and answer information.For the problems in question and answer information information, the scoring of answer information is higher, that is to say, bright user
It is higher to the satisfaction of the answer information；, whereas if the scoring of answer information is lower, that is to say, bright user believes the answer
The satisfaction of breath is lower, and the information that this kind of answer information also exactly needs to optimize.
Therefore, in the present embodiment, this method in step s 102 assessment parameter obtained in judgment step S101 whether
Meet default optimisation criteria.If parameter of testing and assessing meets default optimisation criteria, party's rule is in step s 103 by step S101
In question and answer information to be analyzed as target question and answer information.Wherein, target question and answer information is question and answer information to be optimized.
Specifically, in the present embodiment, since acquired assessment parameter is question and answer evaluation system to this method in step s101
System is directed to the scoring of answer information, therefore this method passes through whether the parameter that judges to test and assess is less than default assessment threshold in step s 102
Value judges whether question and answer information to be analyzed is target question and answer information.
Such as user's input the problem of for information " always alone ", answer information that conversational system is fed back
" I don't have the slightest idea about the meaning of your remarks " obviously pleases oneself without decree user, therefore its corresponding scoring also just will necessarily be very low.
If the scoring corresponding to the answer information is less than default scoring threshold value, mean that needs carry out the answer information
Optimization, so that user's conversational system when inputting problem information can feed back the answer for enabling user more satisfied.
As shown in Figure 1, this method is in step S104 based on target question and answer information after determining target question and answer information
Problem information obtains corresponding answer information aggregate.In the present embodiment, when determining the target question and answer in conversational system database
After information, this method can be pushed to the problems in target question and answer information information under specific scene the user of conversational system,
And user is recorded to the feedback of the problem information, to obtain answer information aggregate corresponding to problem information.
For example, this method in step s 103 determined by target question and answer information include: problem information " it is total oneself one
People " and answer information " I don't have the slightest idea about the meaning of your remarks ", in step S104, this method can by problem information " it is total oneself one
It is personal " being pushed to the user of conversational system, (user herein may be that the user of the input problem information may also be that other are more
A different user).The information pushed aiming at the problem that, user can give the answer feedback of oneself to dialogue system in dialog procedure
System, such this method can obtain answering corresponding to problem information in target question and answer information by collecting the feedback answer of user
Case information aggregate.
It should be pointed out that in other embodiments of the invention, this method can also be obtained by other rational methods
The answer information set for corresponding to problem information in target question and answer information is taken, the invention is not limited thereto.
It is mostly by multiple and different as answering answer information included in information aggregate after obtaining answer information set
User fed back, therefore wherein both may be comprising the answer information that can enable the user more satisfied, it is also possible to comprising can not
Customer satisfaction system answer information is enabled, also just needs to screen answer information aggregate in this way, is asked using being determined to as target
Answer the information of the alternative answer of problem information in information.
Specifically, as shown in Figure 1, this method calculates separately answer in step s105 after obtaining answer information set
The degree of correlation parameter of each answer information in information aggregate, and parameter generation is asked for target according to the degree of correlation in step s 106
Answer the update answer information set of problem information in information.
In the present embodiment, degree of correlation parameter is used to characterize the degree of contact between answer information and problem information, and judges
An important research topic in always artificial intelligence field is contacted between problem information and answer information.In order to enable right
Telephone system can work as efficiently as possible, and this method calculates answer information collection using preset rule in the present embodiment
The degree of correlation parameter of each answer information in conjunction.
Specifically, in the present embodiment, this method is each to calculate separately using multiple rules judged suitable for the degree of correlation
The degree of correlation of the lower question and answer information of rule, followed by the weight of each rule to the degree of correlation being calculated using each rule
It is weighted, to finally obtain the degree of correlation parameter of a certain answer information totality.
Wherein, the preset rules that this method is utilized preferably include centre word rule, and this method is according in question and answer information
Whether answer information and problem information have semantic identical centre word to determine the degree of correlation parameter of answer information.If problem
Information and identical centre word semantic in answer information are more, then the value of answer information and the degree of correlation parameter of problem information
Also bigger；, whereas if problem information and identical centre word semantic in answer information are fewer, then answer information and problem
The value of the degree of correlation parameter of information is also just smaller.
It should be pointed out that in other embodiments of the invention, this method is in the degree of correlation parameter for calculating answer information
When used prediction can also include other rule of reason, the invention is not limited thereto.
In the present embodiment, after obtaining the degree of correlation parameter of each answer information in answer information set, this method can be by
Each answer information in answer information aggregate is ranked up according to the value size of degree of correlation parameter, and it is forward to choose sequence
N (value of N can be set according to actual needs) a answer information is as update answer set, the update answer set
The alternative answer of problem information, thus just realizes the optimization to conversational system knowledge base as in target question and answer information.
It should be pointed out that can be carried out according to actual needs using update cycle of this method to conversational system knowledge base
Setting (such as can be by focus on realizing daily to conversational system knowledge to the question and answer information being collected into daily
Library updates primary), the invention is not limited thereto.Meanwhile this method is when being executed, acquired mass data not only includes number of users
According to further including the data of question and answer evaluation system and user evaluation.Specifically, these data include but is not limited to: user's
The answer information of problem information and conversational system, the source of conversational system answer information, the related data of user, answer information
The time etc. that score data and data generate.
It can be seen from the above description that conversational system knowledge base optimization method provided by the present embodiment realizes dialogue system
The self study and closed loop feedback for knowledge base of uniting.Wherein, this method can enable conversational system to bad in knowledge base
Rationalization improvement is carried out with unreasonable question and answer information independence, so that the quality of the system there can be mentioning for duration
It rises.Meanwhile this method can also make conversational system realize autonomous learning, can also be detached from must rely on maintenance personnel in this way
Manual mode of learning.
This method is asked by the way that the problems in target question and answer information information is pushed to user again to obtain user for this
The answer for inscribing information, just realizes the closed loop feedback of " user-conversational system-user ", therefore conversational system is also in this way
The stability of itself is kept when by external interference.It is based on this closed loop configuration, what conversational system can continue connects
Receive the feedback from user's perception level, and continued to optimize according to the feedback of user knowledge library so as to adjust itself
Output, so that this output can meet the expectation of user.
Such as the problems in target question and answer information as described above information " always alone ", existing dialogue system
Fed back answer information of uniting is that no decree user is satisfied " I don't have the slightest idea about the meaning of your remarks ".And utilize the present embodiment
After provided method is updated conversational system knowledge base, conversational system can then export such as that " I accompanies always the body at you
The answer information on side ", the answer information are obviously more in line with user's expectation.
Embodiment two:
Fig. 2 shows the implementation flow charts of the optimization method of conversational system knowledge base provided by the present embodiment.
As shown in Fig. 2, method provided by the present embodiment obtains the survey of question and answer information to be analyzed in step s 201 first
Parameter is commented, and judge whether assessment parameter meets default optimisation criteria in step S202, if met, in step S203
Using question and answer information to be analyzed acquired in step S201 as target question and answer information, target question and answer information as needs to carry out excellent
The question and answer information of change.
It should be pointed out that in the present embodiment, the realization principle and realization process and reality of step S201 to step S203
It applies the realization principle of step S101 to step S103 in example one and realizes that process is identical, therefore details are not described herein.
Existing conversational system is not distinguish user in carry out problem push, this may cause conversational system that can incite somebody to action
Problem information is pushed to unsuitable user, and the answer information that this kind of user is fed back is likely to cause answer information unreliable
And inaccuracy.
For example, when conversational system has pushed problem information " how is the cosmetics of X board ", male user to a male user
The answer of the problem information usually can not be accurately provided, thus also may to conversational system feed back such as " I does not know " or
The answer information of " I does not use cosmetics ".
In view of the above-mentioned problems, method provided by the present embodiment preferably determines that problem is believed by the user property of user
The target user for ceasing push, to obtain more accurate, reliable answer information.
In the present embodiment, after determining the target question and answer information in conversational system database, this method can ask target
The specific user that the problems in information information is pushed to conversational system is answered, and records user to the feedback of the problem information, thus
Obtain answer information aggregate corresponding to problem information.
Specifically, as shown in Fig. 2, after determining target question and answer information, this method obtains dialogue system in step S204
The attribute of different user in system, and user property accessed in judgment step S204 in step S205 whether meet it is default
Problem push request.If the user property of user meets default problem push request, party's rule is in step S206
Corresponding user is determined as target user, and pushes the problems in target question and answer information information to target user.
In the present embodiment, for determining whether user is that the user property of target user preferably includes: user identity letter
Breath, customer position information, age of user information and user gender information etc..Such as age of user information, if
The age of age of user involved in problem information and certain user are in same age bracket in target question and answer information, then the user
Attribute also just meet default problem push request.
Certainly, in different embodiments of the invention, when this method is judged in step S205, both can only make
With in item listed above a certain item or a few items carry out the differentiation of target user, also can use not listed above other
Reasonable item or other reasonable items and above-mentioned a certain item or a few combinations carry out the differentiation of target user, the present invention is not limited to
This.
After pushing the problems in target question and answer information information to target user, this method passes through dialogue in step S207
System obtains target user and is directed to the answer information that problem information is fed back, to obtain answer information set.
After obtaining answer information set, this method calculates separately each answer in answer information set in step S208
The degree of correlation parameter of information, and it is true from answer information combination according to the degree of correlation parameter of each answer information in step S209
Make update answer set.
It should be pointed out that in the present embodiment, the realization principle and realization process of step S207 to step S209 and implementation
The realization principle of step S104 to step S106 and realization process are similar in example one, therefore details are not described herein.
In the present embodiment, when being updated to the answer information in target question and answer information, the updated problem information
Answer information will include preferably multiple answer informations, when such user and conversational system interact, conversational system is directed to
Same problem information can be multiple and different to user feedback answer information, so as to avoid user when inquiring same problem pair
Telephone system makes user be fed up with due to always exporting identical answer, and which further improves the user's body of conversational system
It tests and user's viscosity.
It can be seen from the above description that conversational system knowledge base optimization method provided by the present embodiment is in one institute of embodiment
On the basis of the method for offer, by pushing the problems in target question and answer information letter to specific user according to the attribute of user
Breath.Due to this kind of user attribute with to conversational system input target question and answer information in the attribute of user of problem information it is identical or
It is similar, therefore be also obtained with by this kind of user and to be directed to the more accurate of problem information in target question and answer information or agree with
Higher answer information is spent, and the update answer set for the above problem information determined according to this kind of answer information also will more
It is accurate.
It should be pointed out that in other embodiments of the invention, this method can also be according to other reasonable parameter (examples
Such as interaction scenarios and/or interaction topic) from multiple users of conversational system target user is chosen, the invention is not limited thereto.
Such as in one embodiment of the invention, the execution step S301 to S309 of the optimization method of conversational system knowledge base and implementation
Step S201 in example two is roughly the same to step S209, unlike, in this embodiment, this method institute in step s 304
Acquisition is the current interaction topic of the different user of conversational system, and is to choose mesh using interaction topic in step S305
Mark user's.
In the specific application process, if a certain user and be " football " to the interaction topic between change system, and target
Problem information is for " AlphaGo " in question and answer information, then it is clearly very that problem information, which is pushed to the user, at this time
Lofty, in this case, will be unable to guarantee the answer information that user information pushed aiming at the problem that is fed back can
By property.Therefore the push strategy meeting of this problem information is so that the user being pushed feels that interactive process is interfered, thus
Reduce the user experience of conversational system.
And method provided by the present embodiment is exactly based on interactive topic to choose mesh from a large number of users of conversational system
User is marked, when the current interactive topic of user is related to topic involved in problem information in target question and answer information, at this time will
Problem information, which is pushed to the user not only and not will use family, feels lofty, additionally it is possible to user be allowed to feel that conversational system is that have " thought "
's.In this case, the obvious quality of answer information that user is fed back for problem information is higher, more can accurately send out
User is mirrored really to answer the problem information.
It should be pointed out that in other embodiments of the invention, the optimization method of the conversational system knowledge base can incite somebody to action
Multiple and different parameters (such as attribute, interaction topic and interaction scenarios of user), which combines, carrys out more accurate selection target use
Family, the present invention are similarly not so limited to.
It is also desirable to, it is noted that in the foregoing description, before the assessment parameter for obtaining question and answer information to be analyzed, go back
The question and answer information to be analyzed got can be filtered, the legitimacy of question and answer information is determined with this.
The present invention also provides a kind of optimization device of conversational system knowledge base, Fig. 4 shows the device in the present embodiment
Structural schematic diagram.
As shown in figure 4, the optimization device of conversational system knowledge base provided by the present embodiment preferably includes: target question and answer letter
Breath obtains module 401 and updates answer set generation module 402.Wherein, 401 user of target question and answer data obtaining module judge to
Whether the assessment parameter of analysis question and answer information meets default optimisation criteria.If met, if target question and answer data obtaining module 401
Using the question and answer information as target question and answer information, wherein target question and answer information is question and answer information to be optimized.
In the present embodiment, assessment parameter accessed by target question and answer data obtaining module 401 is preferably question and answer evaluation
Scoring of the system to question and answer information to be analyzed.For the problems in question and answer information information, the scoring of answer information is higher,
That is to say, bright user is higher to the satisfaction of the answer information；, whereas if the scoring of answer information is lower, that is to say, bright use
Family is lower to the satisfaction of the answer information, and the information that this kind of answer information also exactly needs to optimize.
Target question and answer data obtaining module 401 is if it is judged that the value of the assessment parameter of question and answer information to be analyzed is less than in advance
If testing and assessing threshold value, then target question and answer data obtaining module 401 will determine that assessment parameter meets default optimisation criteria, therefore target
Question and answer data obtaining module 401 will also determine the question and answer information to be analyzed for target question and answer information.
After determining target question and answer information, target question and answer information can be transferred to by target question and answer data obtaining module 401
Answer set generation module 402 is updated, to be generated according to target question and answer information for more by update answer set generation module 402
The update answer set of answer corresponding to problem information in fresh target question and answer information.
Specifically, in the present embodiment, target question and answer can be believed under specific scene by updating answer set generation module 402
The problems in breath information is pushed to the user of conversational system, and records user to the feedback of the problem information, to obtain problem
Answer information aggregate corresponding to information.
Update the mesh that answer set generation module 402 preferably determines problem information push by the user property of user
User is marked, to obtain more accurate, reliable answer information.Therefore, in the present embodiment, answer set generation module is updated
402 before carrying out problem push, can obtain the attribute of different user in conversational system, and judges that accessed user property is
It is no to meet default problem push request.If the user property of user meets default problem push request, answer set is updated
It closes generation module 402 and corresponding user is then determined as target user, and push asking in target question and answer information to target user
Inscribe information.
In the present embodiment, for determining whether user is that the user property of target user preferably includes: user identity letter
Breath, customer position information, age of user information and user gender information etc..Certainly, in different embodiments of the invention, more
New answer set generation module 402 both can only be used only a certain in item listed above when choosing to target user
Or a few items carry out the differentiation of target user, also can use other reasonable items or other reasonable items not listed above with
Above-mentioned a certain item or a few combinations carry out the differentiation of target user, and the invention is not limited thereto.
It should be pointed out that in other embodiments of the invention, updating answer set generation module 402 can also basis
Other reasonable parameters (such as interaction scenarios and/or interaction topic), which choose target from multiple users of conversational system, to be used
Family, the invention is not limited thereto.
Such as in one embodiment of the invention, it updates acquired in answer set generation module 402 for choosing mesh
Mark the parameter of the user interaction topic current for the different user of conversational system.Method provided by the present embodiment is exactly based on friendship
Mutual topic chooses target user from a large number of users of conversational system, when the current interactive topic and target question and answer information of user
Involved in middle problem information when topic correlation, it problem information be pushed to the user not only at this time not will use family and feel prominent
It is towering, additionally it is possible to user be allowed to feel that conversational system has " thought ".In this case, user is fed back for problem information
The obvious quality of answer information is higher, more can accurately send out and mirror user and really answer the problem information.
It should be pointed out that in other embodiments of the invention, the optimization method of the conversational system knowledge base can incite somebody to action
Multiple and different parameters (such as attribute, interaction topic and interaction scenarios of user), which combines, carrys out more accurate selection target use
Family, the present invention are similarly not so limited to.
It is also desirable to, it is noted that in the foregoing description, before the assessment parameter for obtaining question and answer information to be analyzed, go back
The question and answer information to be analyzed got can be filtered, the legitimacy of question and answer information is determined with this.
It is mostly by multiple and different as answering answer information included in information aggregate after obtaining answer information set
User fed back, therefore wherein both may be comprising the answer information that can enable the user more satisfied, it is also possible to comprising can not
Customer satisfaction system answer information is enabled, also just needs to screen answer information aggregate in this way, is asked using being determined to as target
Answer the information of the alternative answer of problem information in information.
Therefore, it after obtaining answer information set, updates answer set generation module 402 and calculates separately answer information aggregate
In each answer information degree of correlation parameter, and parameter is generated for problem information in target question and answer information more according to the degree of correlation
New answer information set.
It should be pointed out that updating the correlation that answer set generation module 402 calculates each answer information in the present embodiment
It is identical to spend the principle illustrated in the principle and the optimization method of above-mentioned conversational system knowledge base of parameter, therefore details are not described herein.
It should be understood that disclosed embodiment of this invention is not limited to specific structure disclosed herein, processing step
Or material, and the equivalent substitute for these features that those of ordinary skill in the related art are understood should be extended to.It should also manage
Solution, term as used herein is used only for the purpose of describing specific embodiments, and is not intended to limit.
" one embodiment " or " embodiment " mentioned in specification means the special characteristic described in conjunction with the embodiments, structure
Or characteristic is included at least one embodiment of the present invention.Therefore, the phrase " reality that specification various places throughout occurs
Apply example " or " embodiment " the same embodiment might not be referred both to.
Although above-mentioned example is used to illustrate principle of the present invention in one or more application, for the technology of this field
For personnel, without departing from the principles and ideas of the present invention, hence it is evident that can in form, the details of usage and implementation
It is upper that various modifications may be made and does not have to make the creative labor.Therefore, the present invention is defined by the appended claims.
Claims (8)
1. a kind of optimization method of conversational system knowledge base characterized by comprising
Target question and answer information acquiring step, judges whether the assessment parameter of question and answer information to be analyzed meets optimisation criteria, such as meets,
Then using the question and answer information to be analyzed as target question and answer information, wherein the assessment parameter characterization question and answer evaluation system is treated
The scoring of question and answer information is analyzed, for the problems in question and answer information information, the scoring of answer information and user answer this
The satisfaction of case information is positively correlated；
The problem of updating answer set generation step, the being based on the target question and answer information corresponding answer information collection of acquisition of information
It closes, calculates separately the degree of correlation parameter of each answer information in the answer information set, and raw according to the degree of correlation parameter
At the update answer set for problem information in the target question and answer information；
Wherein, when obtaining the answer information set, target user is chosen according to preset rules, and push away to the target user
The problems in target question and answer information information is sent, the answer letter that the target user is fed back for described problem information is obtained
Breath, to obtain the answer information set.
2. the method as described in claim 1, which is characterized in that in the target question and answer information acquiring step, described in judgement
Whether the assessment parameter of question and answer information to be analyzed is less than default assessment threshold value, if it is less, by the question and answer information to be analyzed
As target question and answer information.
3. method according to claim 1 or 2, which is characterized in that in the update answer set generation step, according to asking
It answers problem information in information and determines the degree of correlation parameter of answer information with whether answer information has semantic identical centre word,
Wherein, the quantity of semantic identical centre word is more, and the degree of correlation parameter of answer information and problem information is bigger.
4. method according to claim 1 or 2, which is characterized in that when choosing the target user,
The user property for obtaining conversational system different user, judges whether user property meets default problem push request, if
Meet, then corresponding user is determined as the target user, wherein the user property includes any in item set forth below
Or several:
Subscriber identity information, customer position information, age of user information, user gender information；
Or, obtaining the interaction scenarios and/or interaction topic of conversational system different user, the interaction scenarios and/or interaction are judged
Whether topic meets default problem push request, if met, corresponding user is determined as the target user.
5. a kind of optimization device of conversational system knowledge base characterized by comprising
Target question and answer data obtaining module, is used to judge whether the assessment parameter of question and answer information to be analyzed to meet optimisation criteria,
Such as meet, then using the question and answer information to be analyzed as target question and answer information, wherein assessment parameter characterization question and answer evaluation system
The scoring united to question and answer information to be analyzed, for the problems in question and answer information information, the scoring of answer information and user
It is positively correlated to the satisfaction of the answer information；
Answer set generation module is updated, the corresponding answer letter of acquisition of information the problem of based on the target question and answer information is used for
Breath set calculates separately the degree of correlation parameter of each answer information in the answer information set, and is joined according to the degree of correlation
Number generates the update answer set for problem information in the target question and answer information；
Wherein, the update answer set generation module is configured to when obtaining the answer information set, according to preset rules
Target user is chosen, and pushes the problems in target question and answer information information to the target user, the target is obtained and uses
Family is directed to the answer information that described problem information is fed back, to obtain the answer information set.
6. device as claimed in claim 5, which is characterized in that the target question and answer data obtaining module is configured to described in judgement
Whether the assessment parameter of question and answer information to be analyzed is less than default assessment threshold value, if it is less, by the question and answer information to be analyzed
As target question and answer information.
7. such as device described in claim 5 or 6, which is characterized in that the update answer set generation module is configured to basis
Whether problem information and answer information have semantic identical centre word to determine that the degree of correlation of answer information is joined in question and answer information
Number, wherein the quantity of semantic identical centre word is more, and the degree of correlation parameter of answer information and problem information is bigger.
8. such as device described in claim 5 or 6, which is characterized in that the update answer set generation module is configured to selecting
When taking the target user,
The user property for obtaining conversational system different user, judges whether user property meets default problem push request, if
Meet, then corresponding user is determined as the target user, wherein the user property includes any in item set forth below
Or several:
Subscriber identity information, customer position information, age of user information, user gender information；
Or, obtaining the interaction scenarios and/or interaction topic of conversational system different user, the interaction scenarios and/or interaction are judged
Whether topic meets default problem push request, if met, corresponding user is determined as the target user.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610262475.7A CN105912697B (en) | 2016-04-25 | 2016-04-25 | A kind of optimization method and device of conversational system knowledge base |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610262475.7A CN105912697B (en) | 2016-04-25 | 2016-04-25 | A kind of optimization method and device of conversational system knowledge base |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105912697A CN105912697A (en) | 2016-08-31 |
| CN105912697B true CN105912697B (en) | 2019-08-27 |
Family
ID=56751992
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610262475.7A Active CN105912697B (en) | 2016-04-25 | 2016-04-25 | A kind of optimization method and device of conversational system knowledge base |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105912697B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106446141B (en) * | 2016-09-21 | 2020-05-19 | 北京光年无限科技有限公司 | Interactive data processing method for intelligent robot system and robot system |
| CN106781790B (en) * | 2016-12-10 | 2022-10-21 | 杭州博世数据网络有限公司 | Online learning system with self-correcting function |
| CN106682137A (en) * | 2016-12-19 | 2017-05-17 | 武汉市灯塔互动文化传播有限公司 | Intelligent stock investment adviser questioning-answering interaction method and intelligent stock investment adviser questioning-answering interaction system |
| CN107729510B (en) * | 2017-10-23 | 2021-07-06 | 深圳市前海众兴科研有限公司 | Information interaction method, information interaction terminal and storage medium |
| CN109657038B (en) * | 2018-10-10 | 2023-04-18 | 创新先进技术有限公司 | Question and answer pair data mining method and device and electronic equipment |
| CN109460459B (en) * | 2018-10-31 | 2020-09-22 | 神思电子技术股份有限公司 | Log learning-based dialogue system automatic optimization method |
| CN111382239B (en) * | 2018-12-27 | 2023-06-23 | 上海智臻智能网络科技股份有限公司 | Interaction flow optimization method and device |
| CN110046235B (en) * | 2019-03-18 | 2023-06-02 | 创新先进技术有限公司 | Knowledge base assessment method, device and equipment |
| CN110263051A (en) * | 2019-06-11 | 2019-09-20 | 出门问问信息科技有限公司 | Question and answer for question answering system are to update method, device, equipment and storage medium |
| CN110674276B (en) * | 2019-09-23 | 2024-08-16 | 深圳前海微众银行股份有限公司 | Robot self-learning method, robot terminal, device and readable storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102637170A (en) * | 2011-02-10 | 2012-08-15 | 北京百度网讯科技有限公司 | Question pushing method and system |
| CN103365899A (en) * | 2012-04-01 | 2013-10-23 | 腾讯科技（深圳）有限公司 | Question recommending method and question recommending system both in questions-and-answers community |
| CN103425640A (en) * | 2012-05-14 | 2013-12-04 | 华为技术有限公司 | Multimedia questioning-answering system and method |
| CN103902652A (en) * | 2014-02-27 | 2014-07-02 | 深圳市智搜信息技术有限公司 | Automatic question-answering system |
| CN104050256A (en) * | 2014-06-13 | 2014-09-17 | 西安蒜泥电子科技有限责任公司 | Initiative study-based questioning and answering method and questioning and answering system adopting initiative study-based questioning and answering method |
| CN105512228A (en) * | 2015-11-30 | 2016-04-20 | 北京光年无限科技有限公司 | Bidirectional question-answer data processing method and system based on intelligent robot |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9892192B2 (en) * | 2014-09-30 | 2018-02-13 | International Business Machines Corporation | Information handling system and computer program product for dynamically assigning question priority based on question extraction and domain dictionary |
-
2016
- 2016-04-25 CN CN201610262475.7A patent/CN105912697B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102637170A (en) * | 2011-02-10 | 2012-08-15 | 北京百度网讯科技有限公司 | Question pushing method and system |
| CN103365899A (en) * | 2012-04-01 | 2013-10-23 | 腾讯科技（深圳）有限公司 | Question recommending method and question recommending system both in questions-and-answers community |
| CN103425640A (en) * | 2012-05-14 | 2013-12-04 | 华为技术有限公司 | Multimedia questioning-answering system and method |
| CN103902652A (en) * | 2014-02-27 | 2014-07-02 | 深圳市智搜信息技术有限公司 | Automatic question-answering system |
| CN104050256A (en) * | 2014-06-13 | 2014-09-17 | 西安蒜泥电子科技有限责任公司 | Initiative study-based questioning and answering method and questioning and answering system adopting initiative study-based questioning and answering method |
| CN105512228A (en) * | 2015-11-30 | 2016-04-20 | 北京光年无限科技有限公司 | Bidirectional question-answer data processing method and system based on intelligent robot |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105912697A (en) | 2016-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105912697B (en) | A kind of optimization method and device of conversational system knowledge base | |
| CN110279425B (en) | Psychological assessment method and system based on intelligent analysis | |
| CN109033229A (en) | Question and answer treating method and apparatus | |
| KR20180079391A (en) | Learning system, learning device, learning method, learning program, teacher data creation device, teacher data creation method, teacher data creation program, terminal device and threshold value change device | |
| Karimi et al. | Deep learning in a computational model for conceptual shifts in a co-creative design system | |
| CN107562863A (en) | Chat robots reply automatic generation method and system | |
| CN106845624A (en) | The multi-modal exchange method relevant with the application program of intelligent robot and system | |
| CN108052605A (en) | A kind of intelligent Answer System based on client feature library | |
| CN109726974A (en) | Intelligent recruitment methods, device, equipment and computer readable storage medium | |
| CN105808590A (en) | Search engine realization method as well as search method and apparatus | |
| CN109446387A (en) | A kind of Owners Committee's intelligent Answer System based on artificial intelligence | |
| CN113871020B (en) | Health management method and system after critical illness diagnosis based on AI machine learning | |
| CN108595609A (en) | Generation method, system, medium and equipment are replied by robot based on personage IP | |
| CN116414961A (en) | Question-answering method and system based on military domain knowledge graph | |
| CN108885637A (en) | Personage is mild-natured related | |
| Peixoto et al. | The issues of solving staffing and scheduling problems in software development projects | |
| Huang et al. | Applied to mobile multimedia intelligent speech system interactive topic guiding model | |
| Özerol et al. | A parallel between regret theory and outranking methods for multicriteria decision making under imprecise information | |
| CN113256100B (en) | Teaching method and system for indoor design based on virtual reality technology | |
| CN109693244A (en) | The method and device of optimization dialogue robot | |
| CN109299859A (en) | Evaluating method, device, equipment and the storage medium of data | |
| CN116126852B (en) | BIM-based intelligent management data storage method for fabricated building | |
| KR20200141863A (en) | System for providing personal recording data based ai chatting service | |
| CN109977217A (en) | A kind of method, system and the storage medium of intelligence augmentative communication | |
| WO2021192556A1 (en) | Report writing assistance system and report writing assistance method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |