CN106294618A - Searching method and device - Google Patents
Searching method and device Download PDFInfo
- Publication number
- CN106294618A CN106294618A CN201610621687.XA CN201610621687A CN106294618A CN 106294618 A CN106294618 A CN 106294618A CN 201610621687 A CN201610621687 A CN 201610621687A CN 106294618 A CN106294618 A CN 106294618A
- Authority
- CN
- China
- Prior art keywords
- key word
- search
- cluster
- cluster key
- user
- 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.)
- Pending
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
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention provides a kind of searching method and device.The embodiment of the present invention is by obtaining the search key word that user provides, and then according to described search key word, obtain at least one corresponding cluster key word, using as guiding key word, make it possible to described guiding key word, it is supplied to described user, owing to being no longer completely dependent on searching for key word execution search operation, but combine at least one the cluster key word obtained according to search key word and perform search operation, Search Results is made to substantially meet the real intention of user, therefore, it can be avoided that the problem of the data interaction increased between application and search engine caused owing to user is repeated search by application in prior art, thus reduce the processing load of search engine.
Description
[technical field]
The present invention relates to Internet technology, particularly relate to a kind of searching method and device.
[background technology]
Search engine refers to according to certain strategy, uses specific computer program to collect information from the Internet,
After information is organized and processed, providing the user search service, what user searched for relevant information shows user is
System.According to State Statistics Bureau, China's netizen's number has been over 400,000,000, and these data mean that China alreadys more than U.S.
State becomes the first big netizen state in the world, and the website total quantity of China has been over 2,000,000.Therefore, how search is utilized
Service meets user's request to greatest extent, for Internet enterprises, is an important problem all the time.User can will search
Rope key word is supplied to related application, application will search for key word, and be sent to search engine.Search engine then closes according to search
Keyword, scans in data base, to obtain and the Search Results of search Keywords matching, and returns to application and carries out defeated
Go out.
But, the search key word provided due to user is not likely to be very appropriate, and such as, grammer is not strict, key word
The situation such as imperfect, is completely dependent on searching for key word and performs search operation, may be such that Search Results cannot meet user's
Real intention so that user needs search to be repeated by application, so, can increase the data between application and search engine
Alternately, thus result in the increase of the processing load of search engine.
[summary of the invention]
The many aspects of the present invention provide a kind of searching method and device, in order to reduce the processing load of search engine.
An aspect of of the present present invention, it is provided that a kind of searching method, including:
Obtain the search key word that user provides；
According to described search key word, it is thus achieved that at least one corresponding cluster key word, using as guiding key word；
By described guiding key word, it is provided that to described user.
Aspect as above and arbitrary possible implementation, it is further provided a kind of implementation, described according to institute
State search key word, it is thus achieved that at least one corresponding cluster key word, using as guiding key word, including:
According to described search key word, it is thus achieved that with the cluster key word of described search Keywords matching；
The click situation data between cluster key word and other cluster key words according to described coupling, described in selection extremely
A few cluster key word.
Aspect as above and arbitrary possible implementation, it is further provided a kind of implementation, described according to institute
State search key word, it is thus achieved that before the cluster key word of described search Keywords matching, also include:
According to user's historical behavior data, it is thus achieved that co-occurrence search keyword sequence；
Described co-occurrence is searched for keyword sequence, carries out polymerization process, to obtain the crucial part of speech of at least one search；
According to the described crucial part of speech of at least one search, it is thus achieved that at least one each search of search key word apoplexy due to endogenous wind described is closed
The cluster key word of keyword class.
Aspect as above and arbitrary possible implementation, it is further provided a kind of implementation, described according to institute
State the click situation data between the cluster key word of coupling and other cluster key words, select at least one cluster key described
Before word, also include:
Obtain the semantic feature of the cluster key word of the described crucial part of speech of each search；
Semantic feature according to described cluster key word, it is thus achieved that the click situation data between cluster key word two-by-two.
Aspect as above and arbitrary possible implementation, it is further provided a kind of implementation, described click feelings
Condition data include at least one in CTR and expected revenus value.
Another aspect of the present invention, it is provided that a kind of searcher, including:
Acquiring unit, for obtaining the search key word that user provides；
Matching unit, for according to described search key word, it is thus achieved that at least one corresponding cluster key word, using as drawing
Lead key word；
Guidance unit, for by described guiding key word, it is provided that to described user.
Aspect as above and arbitrary possible implementation, it is further provided a kind of implementation, described coupling list
Unit, specifically for
According to described search key word, it is thus achieved that with the cluster key word of described search Keywords matching；And
The click situation data between cluster key word and other cluster key words according to described coupling, described in selection extremely
A few cluster key word.
Aspect as above and arbitrary possible implementation, it is further provided a kind of implementation, described coupling list
Unit, is additionally operable to
According to user's historical behavior data, it is thus achieved that co-occurrence search keyword sequence；
Described co-occurrence is searched for keyword sequence, carries out polymerization process, to obtain the crucial part of speech of at least one search；And
According to the described crucial part of speech of at least one search, it is thus achieved that at least one each search of search key word apoplexy due to endogenous wind described is closed
The cluster key word of keyword class.
Aspect as above and arbitrary possible implementation, it is further provided a kind of implementation, described coupling list
Unit, is additionally operable to
Obtain the semantic feature of the cluster key word of the described crucial part of speech of each search；And
Semantic feature according to described cluster key word, it is thus achieved that the click situation data between cluster key word two-by-two.
Aspect as above and arbitrary possible implementation, it is further provided a kind of implementation, described click feelings
Condition data include at least one in CTR and expected revenus value.
As shown from the above technical solution, the search key word that the embodiment of the present invention is provided by acquisition user, and then according to
Described search key word, it is thus achieved that at least one corresponding cluster key word, using as guiding key word, enabling draw described
Lead key word, it is provided that to described user, owing to being no longer completely dependent on searching for key word execution search operation, but combine according to searching
At least one cluster key word that rope key word is obtained performs search operation so that Search Results substantially meets the real of user
It is intended to, therefore, it is possible to avoid the increase caused owing to user is repeated search by application in prior art application and search
Index hold up between the problem of data interaction, thus reduce the processing load of search engine.
It addition, use technical scheme provided by the present invention, owing to being no longer completely dependent on searching for key word execution search behaviour
Make, but combine at least one the cluster key word obtained according to search key word and perform search operation so that Search Results
Substantially meet the real intention of user, thus improve the effectiveness of Search Results.
It addition, use technical scheme provided by the present invention, owing to being no longer completely dependent on searching for key word execution search behaviour
Make, but combine at least one the cluster key word obtained according to search key word and perform search operation so that Search Results
Substantially meet the real intention of user, thus improve the efficiency of search.
It addition, use technical scheme provided by the present invention, it is possible to be effectively improved the experience of user.
[accompanying drawing explanation]
For the technical scheme being illustrated more clearly that in the embodiment of the present invention, below will be to embodiment or description of the prior art
The accompanying drawing used required in is briefly described, it should be apparent that, the accompanying drawing in describing below is some realities of the present invention
Execute example, for those of ordinary skill in the art, on the premise of not paying creative work, it is also possible to attached according to these
Figure obtains other accompanying drawing.
The schematic flow sheet of the searching method that Fig. 1 provides for one embodiment of the invention；
The structural representation of the searcher that Fig. 2 provides for another embodiment of the present invention.
[detailed description of the invention]
For making the purpose of the embodiment of the present invention, technical scheme and advantage clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
The a part of embodiment of the present invention rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art
Other embodiments whole obtained under not making creative work premise, broadly fall into the scope of protection of the invention.
It should be noted that terminal involved in the embodiment of the present invention can include but not limited to mobile phone, individual digital
Assistant (Personal Digital Assistant, PDA), radio hand-held equipment, panel computer (Tablet Computer),
PC (Personal Computer, PC), MP3 player, MP4 player, wearable device (such as, intelligent glasses,
Intelligent watch, Intelligent bracelet etc.) etc..
It addition, the terms "and/or", a kind of incidence relation describing affiliated partner, expression can exist
Three kinds of relations, such as, A and/or B, can represent: individualism A, there is A and B, individualism B these three situation simultaneously.Separately
Outward, character "/" herein, typically represent the forward-backward correlation relation to liking a kind of "or".
The schematic flow sheet of the searching method that Fig. 1 provides for one embodiment of the invention, as shown in Figure 1.
101, the search key word that user provides is obtained.
102, according to described search key word, it is thus achieved that at least one corresponding cluster key word, using as guiding key word.
103, by described guiding key word, it is provided that to described user.
It should be noted that the executive agent of 101～103 can be partly or entirely the application being located locally terminal,
Or can also be to be arranged in the plug-in unit in the application of local terminal or SDK (Software
Development Kit, SDK) etc. functional unit, or can also for the search engine that is positioned in network side server, or
Can also be the distributed system being positioned at network side, this be particularly limited by the present embodiment.
It is understood that the local program (nativeApp) that described application can be mounted in terminal, or also may be used
To be a web page program (webApp) of browser in terminal, this is not particularly limited by the present embodiment.
So, by obtaining the search key word that user provides, and then according to described search key word, it is thus achieved that correspondence is extremely
A few cluster key word, using as guiding key word, enabling by described guiding key word, it is provided that to described user, by
In be no longer completely dependent on search for key word perform search operation, but combine according to search key word obtained at least one gather
Class keywords performs search operation so that Search Results substantially meets the real intention of user, therefore, it is possible to avoid prior art
In the problem of data interaction increased between application and search engine that causes owing to user is repeated search by application,
Thus reduce the processing load of search engine.
Alternatively, in a possible implementation of the present embodiment, in 101, specifically can gather user and be carried
The described search key word of confession.Specifically, specifically can be realized by the search command that user is triggered.Specifically can use
But it is not limited to following several ways triggering search command:
Mode one:
User can be inputted described search key word on the page that current application is represented, and then, is somebody's turn to do by clicking on
Search button on the page such as, using Baidu.com, to trigger search command, comprises described search key word in this search command.
Wherein, user inputs the order of described search key word can be random order.So, after receiving this search command,
Then can parse the described search key word included in it.
Mode two:
Using Asynchronous loading technology such as, Ajax Asynchronous loading or Jsonp Asynchronous loading etc., user in real is currently
The input content inputted on the page that application is represented, in order to make a distinction with search key word, input content now can
To be known as inputting key word.Wherein, user inputs the order of described search key word can be random order.Specifically, tool
Body can provide the interface such as Ajax interface or Jsonp interface, and these interfaces can use Java, supertext pretreatment
The language such as (Hypertext Preprocessor, PHP) language are write, and what it was concrete call can use Jquery, or
The language such as the JavaScript that person is primary are write.
Mode three: user can press the phonetic search button on the page that current application is represented by long, says and wants
The voice content of input, then, unclamps phonetic search button, to trigger search command, comprises according to described in this search command
The search key word of the textual form of the voice content conversion gone out.So, after receiving this search command, then can resolve
Go out the described search key word included in it.
Mode four: user can say want by clicking on the phonetic search button on the page that current application is represented
The voice content of input, treats that end says voice content a period of time such as, after 2 seconds, then triggers search command, this search
Order comprises the search key word of the textual form changed according to the voice content said.So, this search is being received
After order, then can parse the described search key word included in it.
After getting described input key word, then can perform subsequent operation that is 102～103.
Alternatively, in a possible implementation of the present embodiment, in 102, specifically can be according to described search
Key word, it is thus achieved that with the cluster key word of described search Keywords matching.And then, then can be crucial according to the cluster of described coupling
Click situation data between word and other cluster key words, select at least one cluster key word described.
Wherein, described click situation data can include but not limited to click on arrival rate (Click-Through-Rate,
CTR) i.e. clicking rate and click in expected revenus value i.e. Q-value at least one, this is not particularly limited by the present embodiment.
During a concrete implementation, it is also possible to further according to user's historical behavior data, it is thus achieved that co-occurrence is searched for
Keyword sequence.And then, then described co-occurrence can be searched for keyword sequence, carry out polymerization process, search obtaining at least one
Suo Guanjian part of speech.It is then possible to according to the described crucial part of speech of at least one search, it is thus achieved that the described crucial part of speech of at least one search
In the cluster key word of each crucial part of speech of search.
Here, the collection of user's historical behavior data, may rely on user view (Session) segment data of user.
Wherein, Session section is the retrieval behavior within certain period of time of a logical meaning, i.e. user, and it represents a user
A behavior within certain period is intended to, and from the point of view of the navigation patterns of user, Session section specifically can become at semanteme with stipulations
On there is the consecutive retrieval behavior of identical association.
Firstly, it is necessary to from Session segment data, find out the search key word of co-occurrence, composition includes that several co-occurrences are closed
The co-occurrence search keyword sequence of keyword.Based on these co-occurrences search for keyword sequence, can build one directive initially
Graph structure, the node of its figure is that these co-occurrences search for key word, and it is then all that the initial weight on limit i.e. clicks on situation data
The total degree that in Session segment data, two co-occurrence search key words occur.
After constructing initial graph structure, polymerization can be proceeded by and process.The method that concrete polymerization processes can use
Following method:
1), by each node in initial graph structure, it is initialized as a crucial part of speech of search；
2), for each search key part of speech, find out the search key part of speech that it is adjacent, calculate two crucial parts of speech of search
Between similarity；
Specifically, the calculating of the similarities between two crucial parts of speech of search can include two parts: first portion
Dividing is the semantic similarities between two crucial parts of speech of search；The second part be two crucial parts of speech of search out-degree node it
Between structural similarity, and the structural similarity between the in-degree nodes of two crucial parts of speech of search.To the two part
Similarity carries out calculation process, such as, ranking operation etc., it is thus achieved that a computing numerical value, between two crucial parts of speech of search
Similarity.
Wherein, the computational methods of the semantic similarity of first part are:
Calculate search key word apoplexy due to endogenous wind each search key word to close with another each search of search key word apoplexy due to endogenous wind
Semantic similarity between keyword, then, average such as weighted mean or arithmetic mean of instantaneous value, as two search keys
Semantic similarity between part of speech.
In the second part, the computational methods of structural similarity between in-degree node are:
Find out the in-degree node of the crucial part of speech of each search, according to the characteristic of these in-degree nodes, calculate two and search
Structural similarity between the in-degree node of Suo Guanjian part of speech.In-degree node, refers in directed graph as the terminal on limit in figure
Node.
In the second part, the computational methods of structural similarity between out-degree node are:
Find out the out-degree node of the crucial part of speech of each search, according to the characteristic of these out-degree nodes, calculate two and search
Structural similarity between the out-degree node of Suo Guanjian part of speech.Out-degree node, refers in directed graph as the starting point on limit in figure
Node.
3) if the similarity between two crucial parts of speech of search is less than or equal to the similarity threshold pre-set, then will
The crucial part of speech of the two search merges, it is thus achieved that a new search key part of speech.
4), above procedure continuously carry out many wheels, finally, result tends to stable and then stops, and i.e. this is taken turns produced search and closes
Difference between the number of the number of keyword class and the crucial part of speech of the produced search of previous round, less than quantity threshold, or reaches
Specify wheel number, then stop.
So far, then a graph structure being made up of the crucial part of speech of several search is built.
This polymerization that the present embodiment is used processes and is similar to bottom-up hierarchical clustering (Hierarchical
Clustering), being in place of difference, what Hierarchical Clustering found every time is the crucial part of speech of all search
In immediate two crucial parts of speech of search merge, it is then parts of speech crucial to all search that above-mentioned polymerization processes, and finds out
The search key part of speech that its adjacent search key word apoplexy due to endogenous wind can merge merges, and treatment effeciency is higher.
After building above-mentioned graph structure based on the crucial part of speech of search, in addition it is also necessary to further between node and node
Limit carry out one and estimate process, i.e. estimate score such as obtaining a weight, CTR score, represent cluster key word with
Click situation data between cluster key word.As such, it is possible to remove estimating the relatively low limit of score, it is possible to be prevented effectively from coupling
Cluster key word too many, it is also possible to as the result of a pre-sequence.User can according to sequence result, from
In other cluster key words corresponding to the cluster key word that described search key word is mated, select part cluster key word,
Using as guiding key word.
Traditional way is to promote decision tree (Gradient Boosting Decision Tree, GBDT) by gradient
Model calculates, and by calculating the statistical nature of two crucial parts of speech of search, then the training sample by clicking on carries out mould
Type training, finally carries out candidate data prediction by the model trained, and obtains final estimating scores.This method, its
Generalization ability is more weak and can only consider statistical information, and the new search key word much not occurred i.e. clusters key word,
Statistical information may be lacked and can not get preferable one and estimate score, cause not representing chance.
During another concrete implementation, it is also possible to the cluster obtaining the described crucial part of speech of each search further is closed
The semantic feature of keyword, and then, then can be according to the semantic feature of described cluster key word, it is thus achieved that two-by-two between cluster key word
Click situation data.
Specifically, semantic vector (Embedding) deep layer can be utilized neural according to the semantic feature of cluster key word
Network (Deep Neural Networks, DNN) or GBDT model, it is thus achieved that estimate score.As such, it is possible to take into full account semantic letter
Breath, has preferable generalization ability simultaneously, and the new search key word for not occurring i.e. clusters key word, also will not be because of
For can not find statistical nature, cause estimating score abnormal.Such as, two cluster key words are through a vector (Embedding)
Layer, so that cluster key word is mapped in a hyperspace, becomes a term vector.Then, by vector (Embedding)
Term vector input word bag (the Bag of words) model that layer is exported, carries out hidden layer conversion process, the knot that will obtain the most again
Fruit carries out quantity (DOT) product.That DOT product obtains as a result, it is possible to pass through loss function (logloss function) as target letter
Number, carries out the study of this semanteme Embedding DNN, obtains final estimating score.
Alternatively, DOT product obtain as a result, it is possible to as the feature of GBDT model, it is also possible to convert it through some
After, obtain final estimating score.
So far, the weight on the limit between node and node i.e. being estimated after score calculated, complete graph structure is just
Structure completes.So, then can be according to described search key word, the graph structure constructed by utilization, it is thus achieved that close with described search
The cluster key word of keyword coupling, and further according to described indicated by the weight on the limit between node with node mate poly-
Click situation data between class keywords and other cluster key words, can select the click feelings indicated by click situation data
The cluster key word of the specified quantity that condition is best, as at least one cluster key word described, or can also select selected element
Hit the situation data cluster key word more than or equal to predetermined threshold value, as at least one cluster key word described, the present embodiment
This is not particularly limited.
Specifically, specifically constructed graph structure can be stored in the storage device of terminal.
Such as, the storage device of described terminal can be slow storage device, is specifically as follows the hard disk of computer system,
Or can also be the inoperative internal memory i.e. physical memory of mobile phone, such as, read only memory (Read-Only Memory, ROM)
With RAM (random access memory) card etc., this is not particularly limited by the present embodiment.
Or, more such as, the storage device of described terminal can also be speedy storage equipment, is specifically as follows department of computer science
The internal memory of system, or can also be the running memory i.e. Installed System Memory of mobile phone, such as, random access memory (Random Access
Memory, RAM) etc., this is not particularly limited by the present embodiment.
CTR owing to being used estimates the click situation data that mode is obtained, and only considered with to search for key word corresponding
The click situation of cluster key word of the i.e. single-wheel of one click, and the global maximum of income to be reached, in addition it is also necessary to further
Consider repeatedly to click on the click situation of the cluster key word of the most wheels, such as, although currently the CTR of cluster key word A is than cluster
The CTR of key word B is high, but after clicking in the past, the CTR of the cluster key word corresponding to cluster key word B is but than cluster key
The CTR of the cluster key word corresponding to word A is much higher, and therefore the overall income of cluster key word B is the biggest.So, if pressed
According to CTR, if selecting cluster key word A, then, overall income is not the most maximum.Therefore, the present invention can introduce enhancing study
Algorithm (Reinforcement Learning), dynamically adjusts and learns the sequence of the cluster key word every time obtained, from
And reach global optimum as far as possible.
Specifically, specifically can use Q Learning algorithm, wherein, term state (State) is defined as current poly-
Class keywords, term behavior (Action) is adjacent cluster key word.In this case, each under each State
Action, can learn the i.e. Q of a Q-value (state, action), and this value is under this State, takes this Action to obtain
One approximate evaluation of the expected revenus taken.Then, by constantly adjusting this Q-value so that this estimates to become closer to very
Real expected revenus.
In the present invention, after user obtains guiding key word, it is possible to use this guiding key word, perform search behaviour
Make.Such as, utilize and guide key word, the key word searched for as this, perform search operation；Or, more such as, utilize search
Key word and guiding key word, the key word searched for collectively as this, perform search operation, and the present embodiment does not carry out spy to this
Do not limit.
Compared to existing technical scheme, technical scheme provided by the present invention has a techniques below advantage:
1), building a graph structure from the angle being intended to (intent), each intention i.e. node is no longer single one
Individual search key word, but a search key part of speech (query cluster) being made up of similarity key word, each
The crucial part of speech of search represents one intention the clearest and the most definite of user, and the process of guiding is the process that an intention redirects.This do
The scene that method is suitable for is more extensive, and the especially universal guiding under session operational scenarios is recommended.
2), semantic vector (Embedding) deep-neural-network (Deep Neural Networks, DNN) is used to estimate
Meter be intended between redirect probability, it is contemplated that the semantic information between intention, substantially increase the generalization ability of model.This language
Justice model, it is possible to alleviate statistical nature disappearance, or the most sparse brought the estimating of statistical nature of new search key word
Inaccurate problem.
3), considering many wheel incomes rather than single-wheel income, be the process of wheel more than owing to guiding, provide the user with is every
Secondary guiding, and each click of user, all will have influence on the income of follow-up entirety.Accordingly, it is considered to the long-term gain of many wheels, and
The short-term yield of non-single-wheel, is a more preferable reasonable manner.
In the present embodiment, by obtaining the search key word that user provides, and then according to described search key word, it is thus achieved that right
At least one the cluster key word answered, using as guiding key word, enabling by described guiding key word, it is provided that to described use
Family, owing to being no longer completely dependent on searching for key word execution search operation, but combines and is obtained at least according to search key word
One cluster key word performs search operation so that Search Results substantially meets the real intention of user, therefore, it is possible to avoid existing
There is the data interaction increased between application and search engine caused in technology owing to user is repeated search by application
Problem, thus reduce the processing load of search engine.
It addition, use technical scheme provided by the present invention, owing to being no longer completely dependent on searching for key word execution search behaviour
Make, but combine at least one the cluster key word obtained according to search key word and perform search operation so that Search Results
Substantially meet the real intention of user, thus improve the effectiveness of Search Results.
It addition, use technical scheme provided by the present invention, owing to being no longer completely dependent on searching for key word execution search behaviour
Make, but combine at least one the cluster key word obtained according to search key word and perform search operation so that Search Results
Substantially meet the real intention of user, thus improve the efficiency of search.
It addition, use technical scheme provided by the present invention, it is possible to be effectively improved the experience of user.
It should be noted that for aforesaid each method embodiment, in order to be briefly described, therefore it is all expressed as a series of
Combination of actions, but those skilled in the art should know, the present invention is not limited by described sequence of movement because
According to the present invention, some step can use other orders or carry out simultaneously.Secondly, those skilled in the art also should know
Knowing, embodiment described in this description belongs to preferred embodiment, involved action and the module not necessarily present invention
Necessary.
In the above-described embodiments, the description to each embodiment all emphasizes particularly on different fields, and does not has the portion described in detail in certain embodiment
Point, may refer to the associated description of other embodiments.
The structural representation of the searcher that Fig. 2 provides for another embodiment of the present invention, as shown in Figure 2.The present embodiment
Searcher can include acquiring unit 21, matching unit 22 and guidance unit 23.Wherein, acquiring unit 21, it is used for obtaining use
The search key word that family provides；Matching unit 22, for according to described search key word, it is thus achieved that at least one corresponding cluster is closed
Keyword, using as guiding key word；Guidance unit 23, for by described guiding key word, it is provided that to described user.
It should be noted that the searcher that provided of the present embodiment can be partly or entirely to be located locally terminal
Application, or can also be to be arranged in the plug-in unit in the application of local terminal or SDK (Software
Development Kit, SDK) etc. functional unit, or can also for the search engine that is positioned in network side server, or
Can also be the distributed system being positioned at network side, this be particularly limited by the present embodiment.
It is understood that the local program (nativeApp) that described application can be mounted in terminal, or also may be used
To be a web page program (webApp) of browser in terminal, this is not particularly limited by the present embodiment.
Alternatively, in a possible implementation of the present embodiment, described matching unit 22, specifically may be used for root
According to described search key word, it is thus achieved that with the cluster key word of described search Keywords matching；And the cluster according to described coupling
Click situation data between key word and other cluster key words, select at least one cluster key word described.
Wherein, described click situation data can include but not limited to click on arrival rate (Click-Through-Rate,
CTR) i.e. clicking rate and click in expected revenus value i.e. Q-value at least one, this is not particularly limited by the present embodiment.
During a concrete implementation, described matching unit 22, it is also possible to be further used for according to user's history row
For data, it is thus achieved that co-occurrence search keyword sequence；Described co-occurrence is searched for keyword sequence, carries out polymerization process, to obtain extremely
Few crucial part of speech of a search；And according to the described crucial part of speech of at least one search, it is thus achieved that at least one search key described
The cluster key word of the crucial part of speech of each search in part of speech.
During another concrete implementation, described matching unit 22, it is also possible to be further used for obtaining described each
The semantic feature of the cluster key word of the crucial part of speech of search；And the semantic feature according to described cluster key word, it is thus achieved that two-by-two
Click situation data between cluster key word.
It should be noted that method in embodiment corresponding to Fig. 1, the searcher that can be provided by the present embodiment realizes.
Describing the related content that may refer in embodiment corresponding to Fig. 1 in detail, here is omitted.
In the present embodiment, obtain, by acquiring unit, the search key word that user provides, and then by matching unit according to institute
State search key word, it is thus achieved that at least one corresponding cluster key word, using as guiding key word so that described guidance unit energy
Enough by described guiding key word, it is provided that to described user, owing to being no longer completely dependent on searching for key word execution search operation, but
Search operation is performed so that Search Results substantially meets in conjunction with at least one the cluster key word obtained according to search key word
The real intention of user, therefore, it is possible to avoid the increasing caused in prior art owing to user is repeated search by application
Add the problem applying the data interaction between search engine, thus reduce the processing load of search engine.
It addition, use technical scheme provided by the present invention, owing to being no longer completely dependent on searching for key word execution search behaviour
Make, but combine at least one the cluster key word obtained according to search key word and perform search operation so that Search Results
Substantially meet the real intention of user, thus improve the effectiveness of Search Results.
It addition, use technical scheme provided by the present invention, owing to being no longer completely dependent on searching for key word execution search behaviour
Make, but combine at least one the cluster key word obtained according to search key word and perform search operation so that Search Results
Substantially meet the real intention of user, thus improve the efficiency of search.
It addition, use technical scheme provided by the present invention, it is possible to be effectively improved the experience of user.
Those skilled in the art is it can be understood that arrive, for convenience and simplicity of description, and the system of foregoing description,
The specific works process of device and unit, is referred to the corresponding process in preceding method embodiment, does not repeats them here.
In several embodiments provided by the present invention, it should be understood that disclosed system, apparatus and method are permissible
Realize by another way.Such as, device embodiment described above is only schematically, such as, and described unit
Dividing, be only a kind of logic function and divide, actual can have other dividing mode, such as, multiple unit or group when realizing
Part can in conjunction with or be desirably integrated into another system, or some features can be ignored, or does not performs.Another point, shown
Or the coupling each other discussed or direct-coupling or communication connection can be indirect by some interfaces, device or unit
Coupling or communication connection, can be electrical, machinery or other form.
The described unit illustrated as separating component can be or may not be physically separate, shows as unit
The parts shown can be or may not be physical location, i.e. may be located at a place, or can also be distributed to multiple
On NE.Some or all of unit therein can be selected according to the actual needs to realize the mesh of the present embodiment scheme
's.
It addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it is also possible to
It is that unit is individually physically present, it is also possible to two or more unit are integrated in a unit.Above-mentioned integrated list
Unit both can realize to use the form of hardware, it would however also be possible to employ hardware adds the form of SFU software functional unit and realizes.
The above-mentioned integrated unit realized with the form of SFU software functional unit, can be stored in an embodied on computer readable and deposit
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions with so that a computer
Device (can be personal computer, server, or network equipment etc.) or processor (processor) perform the present invention each
The part steps of method described in embodiment.And aforesaid storage medium includes: USB flash disk, portable hard drive, read only memory (Read-
Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disc or CD etc. various
The medium of program code can be stored.
Last it is noted that above example is only in order to illustrate technical scheme, it is not intended to limit；Although
With reference to previous embodiment, the present invention is described in detail, it will be understood by those within the art that: it still may be used
So that the technical scheme described in foregoing embodiments to be modified, or wherein portion of techniques feature is carried out equivalent；
And these amendment or replace, do not make appropriate technical solution essence depart from various embodiments of the present invention technical scheme spirit and
Scope.
Claims (10)
1. a searching method, it is characterised in that including:
Obtain the search key word that user provides；
According to described search key word, it is thus achieved that at least one corresponding cluster key word, using as guiding key word；
By described guiding key word, it is provided that to described user.
Method the most according to claim 1, it is characterised in that described according to described search key word, it is thus achieved that correspondence is extremely
A few cluster key word, using as guiding key word, including:
According to described search key word, it is thus achieved that with the cluster key word of described search Keywords matching；
The click situation data between cluster key word and other cluster key words according to described coupling, described in selection at least one
Individual cluster key word.
Method the most according to claim 2, it is characterised in that described according to described search key word, it is thus achieved that to search with described
Before the cluster key word of rope Keywords matching, also include:
According to user's historical behavior data, it is thus achieved that co-occurrence search keyword sequence；
Described co-occurrence is searched for keyword sequence, carries out polymerization process, to obtain the crucial part of speech of at least one search；
According to the described crucial part of speech of at least one search, it is thus achieved that at least one search key word apoplexy due to endogenous wind each search key word described
The cluster key word of class.
Method the most according to claim 2, it is characterised in that the described cluster key word according to described coupling gathers with other
Click situation data between class keywords, before selecting at least one cluster key word described, also include:
Obtain the semantic feature of the cluster key word of the described crucial part of speech of each search；
Semantic feature according to described cluster key word, it is thus achieved that the click situation data between cluster key word two-by-two.
5. according to the method described in claim 2～4 any claim, it is characterised in that described click situation data include
At least one in CTR and expected revenus value.
6. a searcher, it is characterised in that including:
Acquiring unit, for obtaining the search key word that user provides；
Matching unit, for according to described search key word, it is thus achieved that at least one corresponding cluster key word, using as guiding pass
Keyword；
Guidance unit, for by described guiding key word, it is provided that to described user.
Device the most according to claim 6, it is characterised in that described matching unit, specifically for
According to described search key word, it is thus achieved that with the cluster key word of described search Keywords matching；And
The click situation data between cluster key word and other cluster key words according to described coupling, described in selection at least one
Individual cluster key word.
Device the most according to claim 7, it is characterised in that described matching unit, is additionally operable to
According to user's historical behavior data, it is thus achieved that co-occurrence search keyword sequence；
Described co-occurrence is searched for keyword sequence, carries out polymerization process, to obtain the crucial part of speech of at least one search；And
According to the described crucial part of speech of at least one search, it is thus achieved that at least one search key word apoplexy due to endogenous wind each search key word described
The cluster key word of class.
Device the most according to claim 7, it is characterised in that described matching unit, is additionally operable to
Obtain the semantic feature of the cluster key word of the described crucial part of speech of each search；And
Semantic feature according to described cluster key word, it is thus achieved that the click situation data between cluster key word two-by-two.
10. according to the device described in claim 7～9 any claim, it is characterised in that described click situation data include
At least one in CTR and expected revenus value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610621687.XA CN106294618A (en) | 2016-08-01 | 2016-08-01 | Searching method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610621687.XA CN106294618A (en) | 2016-08-01 | 2016-08-01 | Searching method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106294618A true CN106294618A (en) | 2017-01-04 |
Family
ID=57663900
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610621687.XA Pending CN106294618A (en) | 2016-08-01 | 2016-08-01 | Searching method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106294618A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107066558A (en) * | 2017-03-28 | 2017-08-18 | 北京百度网讯科技有限公司 | Boot entry based on artificial intelligence recommends method and device, equipment and computer-readable recording medium |
| CN107748801A (en) * | 2017-11-16 | 2018-03-02 | 北京百度网讯科技有限公司 | News recommends method, apparatus, terminal device and computer-readable recording medium |
| CN107832439A (en) * | 2017-11-16 | 2018-03-23 | 百度在线网络技术（北京）有限公司 | Method, system and the terminal device of more wheel state trackings |
| CN109313638A (en) * | 2017-03-09 | 2019-02-05 | 微软技术许可有限责任公司 | Using recommendation |
| CN109547847A (en) * | 2018-11-22 | 2019-03-29 | 广州酷狗计算机科技有限公司 | Add the method, apparatus and computer readable storage medium of video information |
| CN109547840A (en) * | 2018-12-03 | 2019-03-29 | 深圳创维数字技术有限公司 | Films and television programs search index method, TV and computer readable storage medium |
| CN111242752A (en) * | 2020-04-24 | 2020-06-05 | 支付宝(杭州)信息技术有限公司 | Method and system for determining recommended object based on multi-task prediction |
| CN112650907A (en) * | 2020-12-25 | 2021-04-13 | 百度在线网络技术（北京）有限公司 | Search word recommendation method, target model training method, device and equipment |
| CN113486071A (en) * | 2021-07-27 | 2021-10-08 | 掌阅科技股份有限公司 | Searching method, server, client and system based on electronic book |
| CN113516491A (en) * | 2020-04-09 | 2021-10-19 | 百度在线网络技术（北京）有限公司 | Promotion information display method and device, electronic equipment and storage medium |
| CN113609372A (en) * | 2021-08-26 | 2021-11-05 | 中国银行股份有限公司 | Search method, apparatus, server, medium, and product |
| CN114238778A (en) * | 2022-02-23 | 2022-03-25 | 深圳市云初信息科技有限公司 | Scientific and technological information recommendation method, device, medium and electronic equipment based on big data |
| US20240061894A1 (en) * | 2022-08-17 | 2024-02-22 | Ascent Korea Co., Ltd. | Service providing apparatus and method for providing search path |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102902671A (en) * | 2011-07-25 | 2013-01-30 | 腾讯科技（深圳）有限公司 | Search method and device for advertising system |
| CN103902597A (en) * | 2012-12-27 | 2014-07-02 | 百度在线网络技术（北京）有限公司 | Method and device for determining search relevant categories corresponding to target keywords |
| CN105404680A (en) * | 2015-11-25 | 2016-03-16 | 百度在线网络技术（北京）有限公司 | Searching recommendation method and apparatus |
-
2016
- 2016-08-01 CN CN201610621687.XA patent/CN106294618A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102902671A (en) * | 2011-07-25 | 2013-01-30 | 腾讯科技（深圳）有限公司 | Search method and device for advertising system |
| CN103902597A (en) * | 2012-12-27 | 2014-07-02 | 百度在线网络技术（北京）有限公司 | Method and device for determining search relevant categories corresponding to target keywords |
| CN105404680A (en) * | 2015-11-25 | 2016-03-16 | 百度在线网络技术（北京）有限公司 | Searching recommendation method and apparatus |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109313638A (en) * | 2017-03-09 | 2019-02-05 | 微软技术许可有限责任公司 | Using recommendation |
| CN109313638B (en) * | 2017-03-09 | 2023-09-01 | 微软技术许可有限责任公司 | Application recommendation |
| CN107066558A (en) * | 2017-03-28 | 2017-08-18 | 北京百度网讯科技有限公司 | Boot entry based on artificial intelligence recommends method and device, equipment and computer-readable recording medium |
| CN107748801B (en) * | 2017-11-16 | 2022-04-29 | 北京百度网讯科技有限公司 | News recommendation method and device, terminal equipment and computer readable storage medium |
| CN107748801A (en) * | 2017-11-16 | 2018-03-02 | 北京百度网讯科技有限公司 | News recommends method, apparatus, terminal device and computer-readable recording medium |
| CN107832439A (en) * | 2017-11-16 | 2018-03-23 | 百度在线网络技术（北京）有限公司 | Method, system and the terminal device of more wheel state trackings |
| CN107832439B (en) * | 2017-11-16 | 2019-03-08 | 百度在线网络技术（北京）有限公司 | Method, system and the terminal device of more wheel state trackings |
| US10664755B2 (en) | 2017-11-16 | 2020-05-26 | Baidu Online Network Technology (Beijing) Co., Ltd. | Searching method and system based on multi-round inputs, and terminal |
| CN109547847A (en) * | 2018-11-22 | 2019-03-29 | 广州酷狗计算机科技有限公司 | Add the method, apparatus and computer readable storage medium of video information |
| CN109547840A (en) * | 2018-12-03 | 2019-03-29 | 深圳创维数字技术有限公司 | Films and television programs search index method, TV and computer readable storage medium |
| CN113516491B (en) * | 2020-04-09 | 2024-04-30 | 百度在线网络技术（北京）有限公司 | Popularization information display method and device, electronic equipment and storage medium |
| CN113516491A (en) * | 2020-04-09 | 2021-10-19 | 百度在线网络技术（北京）有限公司 | Promotion information display method and device, electronic equipment and storage medium |
| CN111242752A (en) * | 2020-04-24 | 2020-06-05 | 支付宝(杭州)信息技术有限公司 | Method and system for determining recommended object based on multi-task prediction |
| CN112650907B (en) * | 2020-12-25 | 2023-07-14 | 百度在线网络技术（北京）有限公司 | Search word recommendation method, target model training method, device and equipment |
| CN112650907A (en) * | 2020-12-25 | 2021-04-13 | 百度在线网络技术（北京）有限公司 | Search word recommendation method, target model training method, device and equipment |
| CN113486071A (en) * | 2021-07-27 | 2021-10-08 | 掌阅科技股份有限公司 | Searching method, server, client and system based on electronic book |
| CN113609372A (en) * | 2021-08-26 | 2021-11-05 | 中国银行股份有限公司 | Search method, apparatus, server, medium, and product |
| CN114238778A (en) * | 2022-02-23 | 2022-03-25 | 深圳市云初信息科技有限公司 | Scientific and technological information recommendation method, device, medium and electronic equipment based on big data |
| US20240061894A1 (en) * | 2022-08-17 | 2024-02-22 | Ascent Korea Co., Ltd. | Service providing apparatus and method for providing search path |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106294618A (en) | Searching method and device | |
| CN110598206B (en) | Text semantic recognition method and device, computer equipment and storage medium | |
| CN109635273B (en) | Text keyword extraction method, device, equipment and storage medium | |
| CN110032632A (en) | Intelligent customer service answering method, device and storage medium based on text similarity | |
| US11704500B2 (en) | Techniques to add smart device information to machine learning for increased context | |
| CN109299237B (en) | Cyclic network man-machine conversation method based on actor critic reinforcement learning algorithm | |
| CN111274365B (en) | Intelligent inquiry method and device based on semantic understanding, storage medium and server | |
| CN110083705A (en) | A kind of multi-hop attention depth model, method, storage medium and terminal for target emotional semantic classification | |
| CN104834747A (en) | Short text classification method based on convolution neutral network | |
| JP2021508866A (en) | Promote area- and client-specific application program interface recommendations | |
| CN107844533A (en) | A kind of intelligent Answer System and analysis method | |
| CN109597995A (en) | A kind of document representation method based on BM25 weighted combination term vector | |
| JP2021508391A (en) | Promote area- and client-specific application program interface recommendations | |
| CN113569578B (en) | User intention recognition method and device and computer equipment | |
| CN116401345A (en) | Intelligent question-answering method, device, storage medium and equipment | |
| CN112215629B (en) | Multi-target advertisement generating system and method based on construction countermeasure sample | |
| CN113569118B (en) | Self-media pushing method, device, computer equipment and storage medium | |
| CN116467461A (en) | Data processing method, device, equipment and medium applied to power distribution network | |
| US7895206B2 (en) | Search query categrization into verticals | |
| CN109472424B (en) | Method and device for predicting actual criminal period of crime, storage medium and server | |
| CN117131273A (en) | Resource searching method, device, computer equipment, medium and product | |
| WO2024007119A1 (en) | Training method for text processing model, and text processing method and device | |
| Bajpai et al. | CEEBERT: Cross-Domain Inference in Early Exit BERT | |
| Ruan et al. | Chinese news text classification method based on attention mechanism | |
| Prajapati et al. | Automatic Question Tagging using Machine Learning and Deep learning Algorithms |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170104 |