CN109885651A - A kind of question pushing method and device - Google Patents
A kind of question pushing method and device Download PDFInfo
- Publication number
- CN109885651A CN109885651A CN201910038458.9A CN201910038458A CN109885651A CN 109885651 A CN109885651 A CN 109885651A CN 201910038458 A CN201910038458 A CN 201910038458A CN 109885651 A CN109885651 A CN 109885651A
- Authority
- CN
- China
- Prior art keywords
- checked
- typical
- target
- similarity
- word
- 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
Abstract
The embodiment of the invention provides a kind of question pushing method and devices.The present invention relates to artificial intelligence fields, this method comprises: obtaining the behavioral data of target user;According to default mapping relations, typical problem corresponding with the behavioral data of target user is searched, the first typical problem is obtained, wherein presets the mapping relations between the typical problem stored in the behavioral data and target database that mapping relations are user;The first typical problem is pushed to target user.Therefore, technical solution provided in an embodiment of the present invention is able to solve the problem of prior art intelligent customer service system intelligently can not recommend problem to user.
Description
[technical field]
The present invention relates to artificial intelligence field more particularly to a kind of question pushing methods and device.
[background technique]
Intelligent customer service system is the technology to grow up on the basis of extensive knowledge processing, is that enterprise and magnanimity are used
Communication between family provides a kind of efficiently and effectively technological means based on natural language.
In order to make intelligent customer service system provide service convenient for user, need to store a set of mark in intelligent customer service system
Quasi- problem and the corresponding answer of typical problem, when user puts question to, the typical problem and the standard for finding user's input are asked
Inscribe corresponding answer, for example, user inputs " how changing Mobile Phone Short Message Service? " in intelligent customer service system this standard is asked
Topic, intelligent customer service system push the answer of this typical problem to user.
But in actual application, such as user is met difficulty during using certain a software and wants to put question to,
But may user can not accurate description problem, user input the problem of, intelligent customer service system not quite identical with typical problem
Keyword is extracted in the problem of system is inputted from user, according to keyword search typical problem, more than one standard can be searched and asked
The whole typical problems searched are returned to user by topic, intelligent customer service system, and user needs to consume more time from intelligent visitor
It is found in multiple typical problems that dress system returns and oneself really wants the problem of puing question to and confirm, then intelligent customer service system is returned
The answer of the problem is returned, in this process, the problem of intelligent customer service system unpredictable user, thus can not be intelligently to user
Recommendation problem, can only be according to Keywords matching typical problem, and the multiple typical problems matched also need user's selection real
Want the problem of puing question to, the time of more wasteful user.
[summary of the invention]
In view of this, the embodiment of the invention provides a kind of question pushing method and device, to solve prior art intelligence
The problem of energy customer service system intelligently can not recommend problem to user.
On the one hand, the embodiment of the invention provides a kind of question pushing methods, which comprises obtains target user's
Behavioral data;According to default mapping relations, typical problem corresponding with the behavioral data of the target user is searched, obtains first
Typical problem, wherein the default mapping relations are asked for the standard stored in the behavioral data of user and the target database
Mapping relations between topic, the default mapping relations first pass through following steps in advance and determine: receiving the problem of multiple users send
Inquiry request, and the behavioral data that the multiple user generates when sending described problem inquiry request is acquired, described problem is looked into
Asking includes problem to be checked in request;Extract described problem inquiry request in include problem to be checked, filter out with it is described to
The highest typical problem of inquiry problem similarity;Establish the mapping between collected behavioral data and the typical problem filtered out
Relationship;First typical problem is pushed to the target user.
Further, described to filter out and the highest typical problem of problem similarity to be checked, comprising: described in calculating
The distance between the central point that clusters for the classification that each clusters in the corresponding sentence vector of problem to be checked and N number of classification that clusters, obtains N
A distance, wherein N number of classification that clusters is determined previously according to following steps: obtaining whole marks in the target database
Quasi- problem;Whole typical problems are respectively converted into corresponding sentence vector;To the corresponding sentence of the whole typical problem to
Amount is clustered, and N number of classification that clusters is obtained, and N is the natural number more than or equal to 2;Filter out in N number of distance it is the smallest away from
From the corresponding classification that clusters, obtains target and cluster classification;It calculates separately the problem to be checked and the target clusters in classification
Similarity between each typical problem, obtains M similarity, and M is that the target clusters the quantity of classification Plays problem;It will
In the M similarity the corresponding typical problem of maximum similarity as with the highest standard of problem similarity to be checked
Problem.
Further, it is described calculate separately the problem to be checked and the target cluster in classification each typical problem it
Between similarity, comprising: determine first object word and the second target word, wherein the first object word be appear in
In the problem to be checked and the word in the second typical problem is not appeared in, the second target word is appears in
The word in the problem to be checked is stated in the second typical problem and is not appeared in, second typical problem is the mesh
Mark any standard problem that the classification that clusters includes;According to the problem to be checked and the corresponding sentence of second typical problem
Vector and the first object word, the second target word calculate the problem to be checked and second typical problem
Between similarity.
Further, it is described according to the problem to be checked and the corresponding sentence vector of second typical problem and
The first object word, the second target word calculate the phase between the problem to be checked and second typical problem
Like degree, comprising: calculate the similarity degree parameter between the first object word and the second target word;According to it is described to
Inquiry problem and the corresponding sentence vector of second typical problem and the similarity degree parameter calculate described to be checked
Similarity between problem and second typical problem.
Further, it is described according to the problem to be checked and the corresponding sentence vector of second typical problem and
The similarity degree parameter calculates the similarity between the problem to be checked and second typical problem, comprising: according to public affairs
FormulaThe similarity between the problem to be checked and second typical problem is calculated, wherein
SIM (A, B) indicates the similarity between the problem to be checked and second typical problem,Indicate the problem to be checked
Corresponding sentence vector,Indicate the corresponding sentence vector of second typical problem,Indicate that the problem to be checked is corresponding
Sentence vector field homoemorphism,Indicate that the corresponding sentence vector field homoemorphism of second typical problem, K indicate the first object word and institute
State the similarity degree parameter between the second target word.
On the one hand, the embodiment of the invention provides a kind of problem driving means, described device includes: acquiring unit, is used for
Obtain the behavioral data of target user;Searching unit, for searching the behavior with the target user according to mapping relations are preset
The corresponding typical problem of data, obtains the first typical problem, wherein behavioral data and institute of the default mapping relations for user
State the mapping relations between the typical problem stored in target database;Push unit, for pushing institute to the target user
State the first typical problem, wherein the default mapping relations are that the pre- determination unit that first passes through determines, the determination unit packet
It includes: receiving subelement, for receiving the problem of multiple users send inquiry request;Subelement is acquired, it is the multiple for acquiring
The behavioral data that user generates when sending described problem inquiry request includes problem to be checked in described problem inquiry request;
Subelement is extracted, for extracting the problem to be checked for including in described problem inquiry request;Screen subelement, for filter out with
The highest typical problem of problem similarity to be checked;Subelement is established, for establishing collected behavioral data and screening
The mapping relations between typical problem out.
Further, the screening subelement includes: the first computing module, corresponding for calculating the problem to be checked
The distance between the central point that clusters for the classification that each clusters in sentence vector and N number of classification that clusters, obtains N number of distance, wherein described
N number of classification that clusters is determined previously according to following steps: obtaining whole typical problems in the target database;By the whole
Typical problem is respectively converted into corresponding sentence vector;The corresponding sentence vector of the whole typical problem is clustered, is obtained N number of
Cluster classification, and N is the natural number more than or equal to 2;Screening module, it is the smallest apart from right in N number of distance for filtering out
The classification that clusters answered obtains target and clusters classification;Second computing module, for calculating separately the problem to be checked and the mesh
The similarity that mark clusters in classification between each typical problem, obtains M similarity, and M clusters classification Plays for the target
The quantity of problem;Determining module, for using the corresponding typical problem of similarity maximum in the M similarity as with it is described
The highest typical problem of problem similarity to be checked.
Further, second computing module comprises determining that submodule, for determining first object word and the second mesh
Mark word, wherein the first object word is to appear in the problem to be checked and do not appear in the second standard to ask
Word in topic, the second target word are to appear in second typical problem and do not appear in described to be checked
Word in problem, second typical problem are that the target clusters any standard problem that classification includes;Computational submodule,
For according to the problem to be checked and the corresponding sentence vector of second typical problem and the first object word,
The second target word calculates the similarity between the problem to be checked and second typical problem.
Further, the computational submodule includes: the big module of the first calculating, for calculate the first object word with
Similarity degree parameter between the second target word;Second calculates big module, for according to the problem to be checked and institute
It states the corresponding sentence vector of the second typical problem and the similarity degree parameter and calculates the problem to be checked and described the
Similarity between two typical problems.
Further, the big module of second calculating includes: calculating little module, for according to formulaThe similarity between the problem to be checked and second typical problem is calculated, wherein
SIM (A, B) indicates the similarity between the problem to be checked and second typical problem,Indicate the problem to be checked
Corresponding sentence vector,Indicate the corresponding sentence vector of second typical problem,Indicate that the problem to be checked is corresponding
Sentence vector field homoemorphism,Indicate that the corresponding sentence vector field homoemorphism of second typical problem, K indicate the first object word and institute
State the similarity degree parameter between the second target word.
On the one hand, the embodiment of the invention provides a kind of storage medium, the storage medium includes the program of storage,
In, equipment where controlling the storage medium in described program operation executes above-mentioned problem method for pushing.
On the one hand, the embodiment of the invention provides a kind of computer equipment, including memory and processor, the memories
For storing the information including program instruction, the processor is used to control the execution of program instruction, and described program instruction is located
The step of reason device loads and realizes above-mentioned problem method for pushing when executing.
In the embodiment of the present invention, the behavioral data and typical problem of user are determined previously according to the behavioral data of historical user
Between mapping relations, obtain the behavioral data of target user;And it is stored according in the behavioral data of user and target database
Typical problem between mapping relations, search corresponding with the behavioral data of target user typical problem, obtain the first standard
Problem;The first typical problem is pushed to target user, by predicting that target user may mention according to the behavioral data of target user
The problem of asking, achieved the effect that intelligence to user recommend problem, solve prior art intelligent customer service system can not intelligently to
User recommends the problem of problem.
[Detailed description of the invention]
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this field
For those of ordinary skill, without any creative labor, it can also be obtained according to these attached drawings other attached
Figure.
Fig. 1 is a kind of flow chart of optional question pushing method according to embodiments of the present invention;
Fig. 2 is a kind of schematic diagram of optional problem driving means according to embodiments of the present invention;
Fig. 3 is a kind of schematic diagram of optional computer equipment provided in an embodiment of the present invention.
[specific embodiment]
For a better understanding of the technical solution of the present invention, being retouched in detail to the embodiment of the present invention with reference to the accompanying drawing
It states.
It will be appreciated that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Base
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts it is all its
Its embodiment, shall fall within the protection scope of the present invention.
The term used in embodiments of the present invention is only to be not intended to be limiting merely for for the purpose of describing particular embodiments
The present invention.In the embodiment of the present invention and the "an" of singular used in the attached claims, " described " and "the"
It is also intended to including most forms, unless the context clearly indicates other meaning.
It should be appreciated that term "and/or" used herein is only a kind of incidence relation for describing affiliated partner, indicate
There may be three kinds of relationships, for example, A and/or B, can indicate: individualism A, exist simultaneously A and B, individualism B these three
Situation.In addition, character "/" herein, typicallys represent the relationship that forward-backward correlation object is a kind of "or".
Fig. 1 is a kind of flow chart of optional question pushing method according to embodiments of the present invention, as shown in Figure 1, this method
Include:
Step S102 obtains the behavioral data of target user.
The behavioral data of target user includes operation data, the state that target user is generated during using target software
Data, at least one of data of seeking help, such as following data can be used as the behavioral data of target user: target user's touching
The behavioral data of a certain control in target software is sent out, target user triggers the behavioral data of a certain chained address in target software,
Target user uses the temporal information status data of target software, and target user sends customer service during using target software and consults
The data etc. of seeking help ask.
Step S104 searches typical problem corresponding with the behavioral data of target user, obtains according to default mapping relations
First typical problem, wherein default mapping relations be in the behavioral data and target database of user the typical problem that stores it
Between mapping relations, default mapping relations first pass through following steps in advance and determine: receive the problem of multiple users send inquiry request,
And the behavioral data that multiple users generate when sending problem inquiry request is acquired, it include inquiry to be checked in problem inquiry request
Topic;The problem to be checked for including in extraction problem inquiry request filters out and the highest typical problem of problem similarity to be checked;
Establish the mapping relations between collected behavioral data and the typical problem filtered out.
A set of typical problem, row of the embodiment of the present invention based on historical user are stored in the database of intelligent customer service system
The problem of sending for data and historical user inquiry request establishes the standard stored in the behavioral data and database of user and asks
Mapping relations between topic can match corresponding typical problem according to the behavioral data of target user, and push away for target user
Send typical problem.After target user sees typical problem, typical problem can be confirmed, to confirm that the problem is strictly
The problem of oneself desired search, pushes the answer of the typical problem to target user after target user's confirmation.
The mapping relations between typical problem stored in the behavioral data and target database of user pre-establish,
The mapping relations how established between the typical problem stored in the behavioral data and target database of user are specifically described below.
Such as: historical user A is encountered a difficulty when using target software, and historical user A can be to target software at this time
Customer service sends problem inquiry request, which can be telephony modalities, be also possible to on-line consulting, intelligent customer service system
System determines behavior number of historical user A when using target software when receiving the problem of historical user A is sent inquiry request
According to D1.At this point, the problem to be checked for including in the problem of intelligent customer service system is sent according to historical user A inquiry request, in number
According to filtered out in the typical problem stored in library with the highest typical problem Q1 of problem similarity to be checked, then behavioral data D1 with
There may be mapping relations between typical problem Q1, but are not can determine that one between behavioral data D1 and typical problem Q1 at this time
Surely there are mapping relations because the behavior of single user and do not have generality.
According to aforesaid way, the problem of acquiring a large amount of historical users, and is sending the same of problem inquiry request at inquiry request
When or behavioral data before, according to a large amount of historical users send the problem of inquiry request in include problem to be checked, in number
According to filtered out in the typical problem stored in library with the highest typical problem of problem similarity to be checked, if it find that most of
In the case of, a certain behavioral data is all corresponding with a certain typical problem, then it is assumed that deposits between behavior data and the typical problem
In mapping relations.For example, having the problem of acquiring 10000 historical users inquiry request altogether and sending problem inquiry request
While or behavioral data before, discovery have the inquiry to be checked for including in the problem of 9000 historical users send inquiry request
Inscribe corresponding typical problem all and be typical problem Q1 (typical problem Q1 be in all typical problems stored in database with it is to be checked
Inquiry inscribe the highest typical problem of similarity), and this 9000 historical users while sending problem inquiry request or it
Preceding behavioral data is behavioral data D1, then illustrates to be implicitly present in mapping relations between behavioral data D1 and typical problem Q1.
Step S106 pushes the first typical problem to target user.
In the embodiment of the present invention, the behavioral data and typical problem of user are determined previously according to the behavioral data of historical user
Between mapping relations, obtain the behavioral data of target user;And it is stored according in the behavioral data of user and target database
Typical problem between mapping relations, search corresponding with the behavioral data of target user typical problem, obtain the first standard
Problem;The first typical problem is pushed to target user, by predicting that target user may mention according to the behavioral data of target user
The problem of asking, achieved the effect that intelligence to user recommend problem, solve prior art intelligent customer service system can not intelligently to
User recommends the problem of problem.
Optionally, it filters out and the highest typical problem of problem similarity to be checked, comprising: it is corresponding to calculate problem to be checked
Sentence vector and N number of classification that clusters in each cluster the distance between the central point that clusters of classification, obtain N number of distance, wherein N
A classification that clusters is determined previously according to following steps: the whole typical problems obtained in target database;By whole typical problems
It is respectively converted into corresponding sentence vector;The corresponding sentence vector of whole typical problems is clustered, N number of classification that clusters is obtained, N is
Natural number more than or equal to 2;It filters out the smallest apart from the corresponding classification that clusters in N number of distance, obtains target and cluster classification;
The similarity that problem and target to be checked cluster in classification between each typical problem is calculated separately, obtains M similarity, M is
Target clusters the quantity of classification Plays problem;Using the corresponding typical problem of similarity maximum in M similarity as with to
The highest typical problem of inquiry problem similarity.
As an alternative embodiment, clustering to the corresponding sentence vector of whole typical problems, N number of cluster is obtained
The realization step of classification includes: S1, determines N value according to priori, N is the number of clusters of cluster;S2 randomly chooses N number of standard and asks
Inscribe cluster central point of the corresponding sentence vector as N number of classification that clusters;S3 calculates first vector and N for first vector
Each cluster the distance between central point in a central point that clusters, and first vector is referred to first vector distance most
In the close corresponding classification of the central point that clusters, first vector is any one typical problem in remaining L-N typical problem
Corresponding sentence vector, L are the quantity of typical problem;S4, after the corresponding sentence vector of whole typical problems, which is sorted out, to be completed, according to
The corresponding sentence vector of each classification Plays problem recalculates the central point that clusters of N number of classification, and updates clustering for N number of classification
Central point, circulation execute S3 and S4, until classification each in N number of classification the adjacent central point that clusters twice distance it is default away from
From within.
N value can determine that priori refers to the classification of whole typical problems in target database according to priori
Situation, therefore N value can be determined according to existing classification.For example, it is assumed that N=30, calculates the corresponding sentence vector of problem to be checked
With the distance between the central point that clusters for the classification that each clusters in 30 classifications that cluster, obtain 30 distances, it is assumed that it is the smallest away from
From it is corresponding be the 10th classification that clusters the central point that clusters, then cluster classification using the 10th classification that clusters as target, it is assumed that
It includes 50 typical problems (M=50) that 10th classification that clusters, which has altogether, then calculates separately problem to be checked and target clusters classification
In similarity between this 50 typical problems, obtain 50 similarities, 50 similarities be subjected to descending arrangements, sequence is the
One similarity is maximum similarity in 50 similarities, using sequence first the corresponding typical problem of similarity as with
The highest typical problem of problem similarity to be checked.
It is to be noted that the central point that clusters for the classification that each clusters in N number of classification that clusters calculates in advance, only need
It precalculates primary, calculates in the clustering of the classification that each clusters in the corresponding sentence vector of problem to be checked and N number of classification that clusters
It only needs directly to bring use when the distance between heart point.Calculate similarity when use clustering algorithm, avoid calculating to
Similarity in inquiry problem and target database between whole typical problems, greatly reduces calculation amount, improves calculating effect
Rate.
Optionally, the similarity that problem and target to be checked cluster in classification between each typical problem is calculated separately, packet
It includes: determining first object word and the second target word, wherein first object word is to appear in problem to be checked and do not have
The word in the second typical problem is occurred, the second target word is to appear in the second typical problem and do not appear in
Word in problem to be checked, the second typical problem are that target clusters any standard problem that classification includes;According to inquiry to be checked
Topic and the corresponding sentence vector of the second typical problem and first object word, the second target word calculate problem to be checked and
Similarity between second typical problem.Optionally, according to problem to be checked and the corresponding sentence vector of the second typical problem
And first object word, the second target word calculate the similarity between problem to be checked and the second typical problem, comprising: meter
Calculate the similarity degree parameter between first object word and the second target word;According to problem to be checked and the second typical problem point
Not corresponding sentence vector and similarity degree parameter calculate the similarity between problem to be checked and the second typical problem.
Calculate problem to be checked and target cluster similarity in classification between each typical problem when, since user is defeated
Enter the inaccurate reason of the description of problem to be checked, causes to have differences between typical problem, but problem to be checked and mark
The semanteme of quasi- problem is substantially consistent, so not only need to consider to repeat between the two when calculating the two similarity
Word, it is also necessary to consider that orphan deposits word between the two and carry out calculating similarity, first object word and the second target word are orphan
Deposit word, such as: problem to be checked is " this can wrap postal evil spirit all? ", the second typical problem is " this Beijing and Shanghai freight free
? ", the orphan in first object word, that is, problem to be checked deposit word be " can with ", " packet postal ", " evil spirit is all ", " ", the second target word
It is " Beijing ", " Shanghai ", " freight free ", " " that orphan in language i.e. the second typical problem, which deposits word,.Calculating problem to be checked and mesh
When marking the similarity in the classification that clusters between each typical problem, it is contemplated that orphan between the two deposits contribution of the word to similarity,
Calculated result accuracy is high.
Optionally, according to problem to be checked and the corresponding sentence vector of the second typical problem and similarity degree parameter meter
Calculate the similarity between problem to be checked and the second typical problem, comprising: according to formula S IM (A, B)=Calculate to
Similarity between inquiry problem and the second typical problem, wherein SIM (A, B) indicate problem to be checked and the second typical problem it
Between similarity,Indicate the corresponding sentence vector of problem to be checked,Indicate the corresponding sentence vector of the second typical problem,Table
Show the corresponding sentence vector field homoemorphism of problem to be checked,Indicate that the corresponding sentence vector field homoemorphism of the second typical problem, K indicate the first mesh
Mark the similarity degree parameter between word and the second target word.
Calculate the formula of the similarity degree parameter K of first object word and the second target word are as follows:N is
The quantity of first object word, semantic similar parameter of the ki for i-th of first object word, ki=W1 × W2 × S (1,2),
In, W1 is the word weight of i-th of first object word, and W2 is and highest second target of i-th of first object Words similarity
The word weight of word, S (1,2) are i-th of first object word and highest second mesh of i-th of first object Words similarity
Mark the semantic degree of closeness parameter of word.
Such as: problem to be checked is " this can wrap postal evil spirit all? ", the second typical problem is that " this Beijing and Shanghai is exempted from
Freight charges? ", the orphan in first object word, that is, problem to be checked deposit word be " can with ", " packet postal ", " evil spirit is all ", " ", the second mesh
It is " Beijing ", " Shanghai ", " freight free ", " " that orphan in mark word i.e. the second typical problem, which deposits word, is searched from corpus each
Word weight of the TF-IDF value of a word as the word, is denoted as W, as shown in table 1.TF-IDF(term frequency–inverse
Document frequency) it is a kind of common weighting technique for information retrieval and data mining.TF means word frequency
(Term Frequency), IDF mean inverse document frequency (Inverse Document Frequency).If two
The word of words is more similar, their content just should be more similar.Therefore, it can start with from word frequency, calculate their similarity degree.
Table 1
Based on word frequency, problem to be checked and the corresponding sentence vector of the second typical problem are as follows:
The corresponding sentence vector of problem to be checked
The corresponding sentence vector of second typical problem
First object word has 4, be respectively as follows: can with, packet postal, evil spirit all, word weight is respectively as follows: WIt can be with=0.6, WPacket postal
=7, WEvil spirit is all=5, W?=0.3.
Second target word has 4, be respectively as follows: Beijing, Shanghai, freight free, word weight is respectively as follows: WBeijing=4.3,
WShanghai=4.8, WFreight free=6, W?=1.
Inquiry corpus obtains the semantic degree of closeness parameter between first object word and the second target word, such as table 2
Shown, the semantic degree of closeness parameter between first object word " evil spirit is all " and the second target word " Shanghai " is 1, first object
Semantic degree of closeness parameter between word " evil spirit is all " and the second target word " Beijing " is 0.2.
Table 2
Beijing | Shanghai | Freight free | ? | |
It can be with | 0 | 0 | 0 | 0 |
Packet postal | 0 | 0 | 1 | 0 |
Evil spirit is all | 0.2 | 1 | 0 | 0 |
? | 0 | 0 | 0 | 0.7 |
For first object word " can with ", i.e. the 1st first object word, k1=0.
For first object word " packet postal ", i.e. the 2nd first object word, k2=7 × 6 × 1=42.
For first object word " evil spirit is all ", i.e. the 3rd first object word, k3=5 × 4.8 × 1=24.
For first object word " ", i.e. the 4th first object word, k4=0.3 × 1 × 0.7=0.21.
Similarity degree parameter K=k1+k2+k3+k4=0+42+24+ between first object word and the second target word
0.21=66.21.
According to formulaIt calculates similar between problem to be checked and the second typical problem
Degree,It says
The similarity of bright problem to be checked and the second typical problem is higher.
Fig. 2 is a kind of schematic diagram of optional problem driving means according to embodiments of the present invention, and the device is for executing
Question pushing method is stated, as shown in Fig. 2, the device includes: acquiring unit 10, searching unit 20, push unit 30.
Acquiring unit 10, for obtaining the behavioral data of target user.
Searching unit 20, for searching standard corresponding with the behavioral data of target user and asking according to mapping relations are preset
Topic, obtains the first typical problem, wherein presets the standard stored in the behavioral data and target database that mapping relations are user
Mapping relations between problem.
Push unit 30, for pushing the first typical problem to target user.
Optionally, default mapping relations are that the pre- determination unit 40 that first passes through determines, determination unit 40 includes: to receive son list
Member acquires subelement, extracts subelement, screen subelement, establish subelement.
Receiving subelement, for receiving the problem of multiple users send inquiry request.
Subelement, the behavioral data generated for acquiring multiple users when sending problem inquiry request are acquired, problem is looked into
Asking includes problem to be checked in request.
Subelement is extracted, for extracting the problem to be checked for including in problem inquiry request.
Subelement is screened, for filtering out and the highest typical problem of problem similarity to be checked.
Subelement is established, the mapping relations for establishing between collected behavioral data and the typical problem filtered out.
In the embodiment of the present invention, the behavioral data and typical problem of user are determined previously according to the behavioral data of historical user
Between mapping relations, obtain the behavioral data of target user;And it is stored according in the behavioral data of user and target database
Typical problem between mapping relations, search corresponding with the behavioral data of target user typical problem, obtain the first standard
Problem;The first typical problem is pushed to target user, by predicting that target user may mention according to the behavioral data of target user
The problem of asking, achieved the effect that intelligence to user recommend problem, solve prior art intelligent customer service system can not intelligently to
User recommends the problem of problem.
Optionally, screening subelement includes: the first computing module, screening module, the second computing module, determining module.The
One computing module, for calculating in the clustering of the classification that each clusters in the corresponding sentence vector of problem to be checked and N number of classification that clusters
The distance between heart point, obtains N number of distance, wherein N number of classification that clusters is determined previously according to following steps: obtaining target data
Whole typical problems in library;Whole typical problems are respectively converted into corresponding sentence vector;It is corresponding to whole typical problems
Sentence vector is clustered, and N number of classification that clusters is obtained, and N is the natural number more than or equal to 2.Screening module, it is N number of for filtering out
It is the smallest apart from the corresponding classification that clusters in distance, it obtains target and clusters classification.Second computing module, it is to be checked for calculating separately
The similarity that inquiry topic and target cluster in classification between each typical problem, obtains M similarity, and M is that target clusters classification
The quantity of Plays problem.Determining module, for using the corresponding typical problem of similarity maximum in M similarity as with to
The highest typical problem of inquiry problem similarity.
Optionally, the second computing module comprises determining that submodule, computational submodule.Submodule is determined, for determining first
Target word and the second target word, wherein first object word is to appear in problem to be checked and do not appear in the
Word in two typical problems, the second target word are to appear in the second typical problem and do not appear in problem to be checked
In word, the second typical problem is that target clusters any standard problem that classification includes.Computational submodule, for according to be checked
Sentence vector corresponding with the second typical problem is inscribed in inquiry and first object word, the second target word calculate inquiry to be checked
Similarity between topic and the second typical problem.
Optionally, computational submodule includes: the big module of the first calculating, the second big module of calculating.First calculates big module, uses
Similarity degree parameter between calculating first object word and the second target word.Second calculates big module, for according to
Inquiry problem and the corresponding sentence vector of the second typical problem and similarity degree parameter calculate problem to be checked and the second mark
Similarity between quasi- problem.
Optionally, the second big module of calculating includes: calculating little module.Little module is calculated, for according to formulaThe similarity between problem to be checked and the second typical problem is calculated, wherein SIM (A, B) table
Show the similarity between problem to be checked and the second typical problem,Indicate the corresponding sentence vector of problem to be checked,Indicate the
The corresponding sentence vector of two typical problems,Indicate the corresponding sentence vector field homoemorphism of problem to be checked,Indicate the second typical problem
Corresponding sentence vector field homoemorphism, K indicate the similarity degree parameter between first object word and the second target word.
On the one hand, the embodiment of the invention provides a kind of storage medium, storage medium includes the program of storage, wherein
Equipment where control storage medium executes following steps when program is run: obtaining the behavioral data of target user;It is reflected according to default
Relationship is penetrated, typical problem corresponding with the behavioral data of target user is searched, obtains the first typical problem, wherein default mapping
Relationship is the mapping relations in the behavioral data and target database of user between the typical problem that stores, and it is pre- to preset mapping relations
It first passes through following steps to determine: receiving the problem of multiple users send inquiry request, and acquire multiple users and looked into transmission problem
The behavioral data generated when request is ask, includes problem to be checked in problem inquiry request;Include in extraction problem inquiry request
Problem to be checked filters out and the highest typical problem of problem similarity to be checked;Establish collected behavioral data and screening
The mapping relations between typical problem out;The first typical problem is pushed to target user.
Optionally, when program is run, equipment where control storage medium also executes following steps: calculating problem to be checked
The distance between the central point that clusters for the classification that each clusters in corresponding sentence vector and N number of classification that clusters, obtains N number of distance,
In, N number of classification that clusters is determined previously according to following steps: obtaining whole typical problems in target database;By whole standards
Problem is respectively converted into corresponding sentence vector;The corresponding sentence vector of whole typical problems is clustered, N number of class that clusters is obtained
Not, N is the natural number more than or equal to 2;It filters out the smallest apart from the corresponding classification that clusters in N number of distance, it is poly- to obtain target
Cluster classification;Calculate separately the similarity that problem and target to be checked cluster in classification between each typical problem, obtain M it is similar
Degree, M are that target clusters the quantity of classification Plays problem;The corresponding typical problem of similarity maximum in M similarity is made
For with the highest typical problem of problem similarity to be checked.
Optionally, when program is run, equipment where control storage medium also executes following steps: determining first object word
Language and the second target word, wherein first object word is to appear in problem to be checked and do not appear in the second standard
Word in problem, the second target word are the word for appearing in the second typical problem and not appearing in problem to be checked
Language, the second typical problem are that target clusters any standard problem that classification includes;According to problem to be checked and the second typical problem
Corresponding sentence vector and first object word, the second target word calculate between problem and the second typical problem to be checked
Similarity.
Optionally, when program is run, equipment where control storage medium also executes following steps: calculating first object word
Similarity degree parameter between language and the second target word;According to problem to be checked and the corresponding sentence of the second typical problem to
Amount and similarity degree parameter calculate the similarity between problem to be checked and the second typical problem.
Optionally, when program is run, equipment where control storage medium also executes following steps: according to formulaThe similarity between problem to be checked and the second typical problem is calculated, wherein SIM (A, B) table
Show the similarity between problem to be checked and the second typical problem,Indicate the corresponding sentence vector of problem to be checked,Indicate the
The corresponding sentence vector of two typical problems,Indicate the corresponding sentence vector field homoemorphism of problem to be checked,Indicate the second typical problem
Corresponding sentence vector field homoemorphism, K indicate the similarity degree parameter between first object word and the second target word.
On the one hand, the embodiment of the invention provides a kind of computer equipments, including memory and processor, memory to be used for
Storage includes the information of program instruction, and processor is used to control the execution of program instruction, and program instruction is loaded and held by processor
The behavioral data for obtaining target user is performed the steps of when row;According to default mapping relations, the behavior with target user is searched
The corresponding typical problem of data, obtains the first typical problem, wherein default mapping relations are the behavioral data and number of targets of user
According to the mapping relations between the typical problem stored in library, default mapping relations first pass through following steps in advance and determine: receiving multiple
The problem of user sends inquiry request, and acquire the behavioral data that multiple users generate when sending problem inquiry request, problem
It include problem to be checked in inquiry request;The problem to be checked for including in extraction problem inquiry request filters out and inquiry to be checked
Inscribe the highest typical problem of similarity;Establish the mapping relations between collected behavioral data and the typical problem filtered out;
The first typical problem is pushed to target user.
Optionally, it is corresponding that calculating problem to be checked is also performed the steps of when program instruction is loaded and executed by processor
Sentence vector and N number of classification that clusters in each cluster the distance between the central point that clusters of classification, obtain N number of distance, wherein N
A classification that clusters is determined previously according to following steps: the whole typical problems obtained in target database;By whole typical problems
It is respectively converted into corresponding sentence vector;The corresponding sentence vector of whole typical problems is clustered, N number of classification that clusters is obtained, N is
Natural number more than or equal to 2;It filters out the smallest apart from the corresponding classification that clusters in N number of distance, obtains target and cluster classification;
The similarity that problem and target to be checked cluster in classification between each typical problem is calculated separately, obtains M similarity, M is
Target clusters the quantity of classification Plays problem;Using the corresponding typical problem of similarity maximum in M similarity as with to
The highest typical problem of inquiry problem similarity.
Optionally, also performed the steps of when program instruction is loaded and executed by processor determining first object word and
Second target word, wherein first object word is to appear in problem to be checked and do not appear in the second typical problem
In word, the second target word is the word for appearing in the second typical problem and not appearing in problem to be checked,
Second typical problem is that target clusters any standard problem that classification includes;Distinguished according to problem to be checked and the second typical problem
Corresponding sentence vector and first object word, the second target word calculate the phase between problem to be checked and the second typical problem
Like degree.
Optionally, when program instruction is loaded and is executed by processor also perform the steps of calculate first object word with
Similarity degree parameter between second target word;According to problem to be checked and the corresponding sentence vector of the second typical problem with
And similarity degree parameter calculates the similarity between problem to be checked and the second typical problem.
Optionally, it also performs the steps of when program instruction is loaded and executed by processor according to formulaThe similarity between problem to be checked and the second typical problem is calculated, wherein SIM (A, B) table
Show the similarity between problem to be checked and the second typical problem,Indicate the corresponding sentence vector of problem to be checked,Indicate the
The corresponding sentence vector of two typical problems,Indicate the corresponding sentence vector field homoemorphism of problem to be checked,Indicate the second typical problem
Corresponding sentence vector field homoemorphism, K indicate the similarity degree parameter between first object word and the second target word.
Fig. 3 is a kind of schematic diagram of computer equipment provided in an embodiment of the present invention.As shown in figure 3, the meter of the embodiment
Machine equipment 50 is calculated to include: processor 51, memory 52 and be stored in the meter that can be run in memory 52 and on processor 51
Calculation machine program 53 realizes the problems in embodiment method for pushing when the computer program 53 is executed by processor 51, to avoid weight
It is multiple, it does not repeat one by one herein.Alternatively, being realized when the computer program is executed by processor 51 in embodiment in problem driving means
The function of each model/unit does not repeat one by one herein to avoid repeating.
Computer equipment 50 can be desktop PC, notebook, palm PC and cloud server etc. and calculate equipment.
Computer equipment may include, but be not limited only to, processor 51, memory 52.It will be understood by those skilled in the art that Fig. 3 is only
It is the example of computer equipment 50, does not constitute the restriction to computer equipment 50, may include more more or fewer than illustrating
Component perhaps combines certain components or different components, such as computer equipment can also include input-output equipment, net
Network access device, bus etc..
Alleged processor 51 can be central processing unit (Central Processing Unit, CPU), can also be
Other general processors, digital signal processor (Digital Signal Processor, DSP), specific integrated circuit
(Application Specific Integrated Circuit, ASIC), field programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components etc..General processor can be microprocessor or the processor is also possible to any conventional processor
Deng.
Memory 52 can be the internal storage unit of computer equipment 50, such as the hard disk or interior of computer equipment 50
It deposits.Memory 52 is also possible to the plug-in type being equipped on the External memory equipment of computer equipment 50, such as computer equipment 50
Hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash card
(Flash Card) etc..Further, memory 52 can also both including computer equipment 50 internal storage unit and also including
External memory equipment.Memory 52 is for storing other programs and data needed for computer program and computer equipment.It deposits
Reservoir 52 can be also used for temporarily storing the data that has exported or will export.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
In several embodiments provided by the present invention, it should be understood that disclosed system, device and method can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit
It divides, only a kind of logical function partition, there may be another division manner in actual implementation, for example, multiple units or group
Part can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, it is shown
Or the mutual coupling, direct-coupling or communication connection discussed can be through some interfaces, device or unit it is indirect
Coupling or communication connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.
The above-mentioned integrated unit being realized in the form of SFU software functional unit can store and computer-readable deposit at one
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer
It is each that device (can be personal computer, server or network equipment etc.) or processor (Processor) execute the present invention
The part steps of embodiment the method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (Read-
Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic or disk etc. it is various
It can store the medium of program code.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, any modification, equivalent substitution, improvement and etc. done be should be included within the scope of the present invention.
Claims (10)
1. a kind of question pushing method, which is characterized in that the described method includes:
Obtain the behavioral data of target user;
According to default mapping relations, typical problem corresponding with the behavioral data of the target user is searched, the first standard is obtained
Problem, wherein the default mapping relations are in the behavioral data and target database of user between the typical problem that stores
Mapping relations, the default mapping relations first pass through following steps in advance and determine: the problem of multiple users send inquiry request is received,
And the behavioral data that the multiple user generates when sending described problem inquiry request is acquired, it is wrapped in described problem inquiry request
Containing problem to be checked;The problem to be checked for including in described problem inquiry request is extracted, is filtered out and the problem phase to be checked
Like the highest typical problem of degree;Establish the mapping relations between collected behavioral data and the typical problem filtered out;
First typical problem is pushed to the target user.
2. the method according to claim 1, wherein described filter out and the problem similarity highest to be checked
Typical problem, comprising:
Between the central point that clusters for calculating the classification that each clusters in the corresponding sentence vector of the problem to be checked and N number of classification that clusters
Distance, obtain N number of distance, wherein it is described it is N number of cluster classification previously according to following steps determine: obtain the target data
Whole typical problems in library;Whole typical problems are respectively converted into corresponding sentence vector;Whole standards are asked
It inscribes corresponding sentence vector to be clustered, obtains N number of classification that clusters, N is the natural number more than or equal to 2;
It filters out the smallest apart from the corresponding classification that clusters in N number of distance, obtains target and cluster classification;
It calculates separately the problem to be checked and similarity that the target clusters in classification between each typical problem, obtains M
A similarity, M are that the target clusters the quantity of classification Plays problem;
Using the corresponding typical problem of similarity maximum in the M similarity as with the problem similarity highest to be checked
Typical problem.
3. method according to claim 2, which is characterized in that described to calculate separately the problem to be checked and the mesh
Mark the similarity in the classification that clusters between each typical problem, comprising:
Determine first object word and the second target word, wherein the first object word is to appear in the inquiry to be checked
In topic and the word in the second typical problem is not appeared in, the second target word is to appear in second standard to ask
In topic and the word in the problem to be checked is not appeared in, second typical problem is that the target clusters classification packet
Any standard problem included;
According to the problem to be checked and the corresponding sentence vector of second typical problem and the first object word,
The second target word calculates the similarity between the problem to be checked and second typical problem.
4. method according to claim 3, which is characterized in that described to be marked according to the problem to be checked with described second
The quasi- corresponding sentence vector of problem and the first object word, the second target word calculate the problem to be checked
With the similarity between second typical problem, comprising:
Calculate the similarity degree parameter between the first object word and the second target word;
According to the problem to be checked and the corresponding sentence vector of second typical problem and the similarity degree parameter
Calculate the similarity between the problem to be checked and second typical problem.
5. method according to claim 4, which is characterized in that described to be marked according to the problem to be checked with described second
The quasi- corresponding sentence vector of problem and the similarity degree parameter calculate the problem to be checked and ask with second standard
Similarity between topic, comprising:
According to formulaCalculate the phase between the problem to be checked and second typical problem
Like degree, wherein SIM (A, B) indicates the similarity between the problem to be checked and second typical problem,Described in expression
The corresponding sentence vector of problem to be checked,Indicate the corresponding sentence vector of second typical problem,Indicate described to be checked
The corresponding sentence vector field homoemorphism of problem,Indicate that the corresponding sentence vector field homoemorphism of second typical problem, K indicate first mesh
Mark the similarity degree parameter between word and the second target word.
6. a kind of problem driving means, which is characterized in that described device includes:
Acquiring unit, for obtaining the behavioral data of target user;
Searching unit, for searching typical problem corresponding with the behavioral data of the target user according to mapping relations are preset,
Obtain the first typical problem, wherein the default mapping relations are the mark that stores in the behavioral data and target database of user
Mapping relations between quasi- problem;
Push unit, for pushing first typical problem to the target user,
Wherein, the default mapping relations are that the pre- determination unit that first passes through determines that the determination unit includes:
Receiving subelement, for receiving the problem of multiple users send inquiry request;
Acquire subelement, the behavioral data generated for acquiring the multiple user when sending described problem inquiry request, institute
It states in problem inquiry request comprising problem to be checked;
Subelement is extracted, for extracting the problem to be checked for including in described problem inquiry request;
Subelement is screened, for filtering out and the highest typical problem of problem similarity to be checked;
Subelement is established, the mapping relations for establishing between collected behavioral data and the typical problem filtered out.
7. device according to claim 6, which is characterized in that the screening subelement includes:
First computing module, for calculating the class that each clusters in the corresponding sentence vector of the problem to be checked and N number of classification that clusters
Other the distance between central point that clusters, obtains N number of distance, wherein N number of classification that clusters is true previously according to following steps
It is fixed: to obtain whole typical problems in the target database;By whole typical problems be respectively converted into corresponding sentence to
Amount;The corresponding sentence vector of the whole typical problem is clustered, N number of classification that clusters is obtained, N is oneself more than or equal to 2
So number;
Screening module, it is the smallest apart from the corresponding classification that clusters in N number of distance for filtering out, it obtains target and clusters class
Not;
Second computing module, for calculate separately the problem to be checked and the target cluster in classification each typical problem it
Between similarity, obtain M similarity, M is that the target clusters the quantity of classification Plays problem;
Determining module, for using the corresponding typical problem of similarity maximum in the M similarity as with it is described to be checked
The highest typical problem of problem similarity.
8. device according to claim 7, which is characterized in that second computing module includes:
Submodule is determined, for determining first object word and the second target word, wherein the first object word is to occur
In the problem to be checked and the word in the second typical problem is not appeared in, the second target word is to appear in
In second typical problem and the word in the problem to be checked is not appeared in, second typical problem is described
Target clusters any standard problem that classification includes;
Computational submodule, for according to the problem to be checked and the corresponding sentence vector of second typical problem and institute
State first object word, the second target word calculate it is similar between the problem to be checked and second typical problem
Degree.
9. a kind of storage medium, which is characterized in that the storage medium includes the program of storage, wherein run in described program
When control the storage medium where equipment perform claim require any one of 1 to 5 described in problem method for pushing.
10. a kind of computer equipment, including memory and processor, the memory is for storing the letter including program instruction
Breath, the processor are used to control the execution of program instruction, it is characterised in that: described program instruction is loaded and executed by processor
The step of problem method for pushing described in Shi Shixian claim 1 to 5 any one.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910038458.9A CN109885651A (en) | 2019-01-16 | 2019-01-16 | A kind of question pushing method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910038458.9A CN109885651A (en) | 2019-01-16 | 2019-01-16 | A kind of question pushing method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109885651A true CN109885651A (en) | 2019-06-14 |
Family
ID=66926094
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910038458.9A Pending CN109885651A (en) | 2019-01-16 | 2019-01-16 | A kind of question pushing method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109885651A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111143530A (en) * | 2019-12-24 | 2020-05-12 | 平安健康保险股份有限公司 | Intelligent answering method and device |
CN113505293A (en) * | 2021-06-15 | 2021-10-15 | 深圳追一科技有限公司 | Information pushing method and device, electronic equipment and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106649742A (en) * | 2016-12-26 | 2017-05-10 | 上海智臻智能网络科技股份有限公司 | Database maintenance method and device |
CN106776751A (en) * | 2016-11-22 | 2017-05-31 | 上海智臻智能网络科技股份有限公司 | The clustering method and clustering apparatus of a kind of data |
CN106897334A (en) * | 2016-06-24 | 2017-06-27 | 阿里巴巴集团控股有限公司 | A kind of question pushing method and equipment |
CN108268877A (en) * | 2016-12-30 | 2018-07-10 | 中国移动通信集团黑龙江有限公司 | A kind of method and apparatus for identifying target terminal |
CN108804567A (en) * | 2018-05-22 | 2018-11-13 | 平安科技(深圳)有限公司 | Improve method, equipment, storage medium and the device of intelligent customer service response rate |
CN109033156A (en) * | 2018-06-13 | 2018-12-18 | 腾讯科技(深圳)有限公司 | A kind of information processing method, device and terminal |
-
2019
- 2019-01-16 CN CN201910038458.9A patent/CN109885651A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106897334A (en) * | 2016-06-24 | 2017-06-27 | 阿里巴巴集团控股有限公司 | A kind of question pushing method and equipment |
CN106776751A (en) * | 2016-11-22 | 2017-05-31 | 上海智臻智能网络科技股份有限公司 | The clustering method and clustering apparatus of a kind of data |
CN106649742A (en) * | 2016-12-26 | 2017-05-10 | 上海智臻智能网络科技股份有限公司 | Database maintenance method and device |
CN108268877A (en) * | 2016-12-30 | 2018-07-10 | 中国移动通信集团黑龙江有限公司 | A kind of method and apparatus for identifying target terminal |
CN108804567A (en) * | 2018-05-22 | 2018-11-13 | 平安科技(深圳)有限公司 | Improve method, equipment, storage medium and the device of intelligent customer service response rate |
CN109033156A (en) * | 2018-06-13 | 2018-12-18 | 腾讯科技(深圳)有限公司 | A kind of information processing method, device and terminal |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111143530A (en) * | 2019-12-24 | 2020-05-12 | 平安健康保险股份有限公司 | Intelligent answering method and device |
CN111143530B (en) * | 2019-12-24 | 2024-04-05 | 平安健康保险股份有限公司 | Intelligent reply method and device |
CN113505293A (en) * | 2021-06-15 | 2021-10-15 | 深圳追一科技有限公司 | Information pushing method and device, electronic equipment and storage medium |
CN113505293B (en) * | 2021-06-15 | 2024-03-19 | 深圳追一科技有限公司 | Information pushing method and device, electronic equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108804641B (en) | Text similarity calculation method, device, equipment and storage medium | |
JP7343568B2 (en) | Identifying and applying hyperparameters for machine learning | |
CN110162695B (en) | Information pushing method and equipment | |
US9870408B2 (en) | Search method and system | |
CN101055585B (en) | System and method for clustering documents | |
CN108304444B (en) | Information query method and device | |
CN110941698B (en) | Service discovery method based on convolutional neural network under BERT | |
CN110209808A (en) | A kind of event generation method and relevant apparatus based on text information | |
CN110019876B (en) | Data query method, electronic device and storage medium | |
JP2013504118A (en) | Information retrieval based on query semantic patterns | |
CN105653701B (en) | Model generating method and device, word assign power method and device | |
CN107729492A (en) | A kind of method for pushing of exercise, system and terminal device | |
CN112035599A (en) | Query method and device based on vertical search, computer equipment and storage medium | |
CN109241392A (en) | Recognition methods, device, system and the storage medium of target word | |
CN112632261A (en) | Intelligent question and answer method, device, equipment and storage medium | |
CN104462347B (en) | The sorting technique and device of keyword | |
CN105786810B (en) | The method for building up and device of classification mapping relations | |
CN108537291A (en) | A kind of sample classification method and apparatus | |
CN109885651A (en) | A kind of question pushing method and device | |
CN114780606A (en) | Big data mining method and system | |
CN111667018B (en) | Object clustering method and device, computer readable medium and electronic equipment | |
CN109635072A (en) | Public sentiment data distributed storage method, device, storage medium and terminal device | |
CN104778202B (en) | The analysis method and system of event evolutionary process based on keyword | |
CN111191065A (en) | Homologous image determining method and device | |
CN115841144A (en) | Training method and device for text retrieval model |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |