CN103365899B - The problem of in a kind of Ask-Answer Community, recommends method and system - Google Patents
The problem of in a kind of Ask-Answer Community, recommends method and system Download PDFInfo
- Publication number
- CN103365899B CN103365899B CN201210095917.5A CN201210095917A CN103365899B CN 103365899 B CN103365899 B CN 103365899B CN 201210095917 A CN201210095917 A CN 201210095917A CN 103365899 B CN103365899 B CN 103365899B
- Authority
- CN
- China
- Prior art keywords
- weight
- user
- keywords
- keyword
- lists
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The present invention suitable for Internet technical field there is provided a kind of Ask-Answer Community the problem of recommend method and system, methods described comprises the steps：Lists of keywords the problem of problem to be recommended is matched with the Weight key to the issue word list of default each user, obtain the weight sum of keyword the problem of the match is successful corresponding with each user, the key to the issue word list is matched with all Weight fields lists of keywords of default each user respectively, obtain the maximum in the weight sum of the corresponding all field keywords that the match is successful of each user, with reference to it is corresponding with same user the match is successful the problem of the weight sum of keyword and the weight sum of field keyword in maximum, obtain the degree of correlation of the problem to be recommended and the user, the problem to be recommended is pushed to degree of correlation highest one or more user, so that user is improved to the response rate of problem in Ask-Answer Community, and Consumer's Experience effect gets a promotion.
Description
Technical field
The problem of the invention belongs in Internet technical field, more particularly to a kind of Ask-Answer Community, recommends method and system.
Background technology
At present, in internet Ask-Answer Community, it is often necessary to recommend problem automatically to user, existing recommendation method is main
It is divided into two classes, i.e. collaborative recommendation method and the recommendation method based on attribute.In collaborative recommendation method, system searching with it is to be recommended
The similar user of user, and according to these users once selected project, the user is given by suitable project recommendation；Or ginseng
The project similar to project to be recommended is examined, and is that suitable user recommends this according to the user for once selecting these projects
Mesh.As electronic emporium can find out him according to the history buying behavior of user may commodity interested and be that he recommends.And
In method based on attribute, it can be recommended according to some attributes of user, such as, can be according to the height and body weight of user
User's recommended dietary, or according to the financial situation of user for its recommend finance product.
However, in Ask-Answer Community, a user proposes the problem of oneself is interested, and by the use of the problem can be answered
Family is answered for it, because problem is no longer needed to understand that the user of the problem continues to answer after answer, therefore is seldom deposited
A problem is answered jointly in multiple users, seldom has a problem that answered by multiple users in other words, therefore based on collaboration
The method of recommendation is not suitable for this situation.And the method recommended based on user property is more suitable for question recommending.Asking
Answer in community, the attribute of user generally comprises the problem of he answered quantity, his active degree, and problem length etc..
But, current the problem of, recommends method or lays particular emphasis on to push the problem of he answered to user, to ensure that he can answer this
Problem；Or lay particular emphasis on and recommend the problem of he did not answer to user as far as possible so that he will not be fed up with.The former can lead
The problem of causing same is always received by same user so that system can not obtain other people answer, and easily makes to be pushed
User is sick of, and the latter then causes user to receive his excessive unanswerable problem.
The content of the invention
The purpose of the embodiment of the present invention is to recommend method and system the problem of offer in a kind of Ask-Answer Community, it is intended to solve
In Ask-Answer Community to user carry out question recommending when, exist to user recommend the problem of it is single and push user lose interest in
The problem of situation, the problem of causing not high to the response rate of problem in Ask-Answer Community and not good Consumer's Experience effect.
The problem of embodiment of the present invention is achieved in that in a kind of Ask-Answer Community is recommended under method, methods described include
State step：
The problem of obtaining problem to be recommended lists of keywords；
Described problem lists of keywords is matched with the Weight key to the issue word list of default each user, obtained
Take the weight sum of keyword the problem of the match is successful corresponding with each user；
All Weight fields lists of keywords by described problem lists of keywords respectively with default each user is entered
Row matching, obtains the maximum in the weight sum of all field keywords that the match is successful corresponding with each user
Value；
With reference to it is corresponding with same user the match is successful the problem of keyword weight sum and the power of field keyword
Maximum in weight sum, obtains the problem to be recommended and the degree of correlation of the user；
According to the problem to be recommended and the degree of correlation of each user, the problem to be recommended is pushed to degree of correlation highest
One or more user.
The another object of the embodiment of the present invention is to provide commending system the problem of in a kind of Ask-Answer Community, the system bag
Include：
Lists of keywords obtains master unit, lists of keywords the problem of for obtaining problem to be recommended；
Weight sum acquiring unit, for by the Weight problem of described problem lists of keywords and default each user
Lists of keywords is matched, and obtains the weight sum of keyword the problem of the match is successful corresponding with each user；
Maximum acquiring unit, for all cum rights by described problem lists of keywords respectively with default each user
Weight field lists of keywords is matched, acquisition all field keywords that the match is successful corresponding with each user
Maximum in weight sum；
Degree of correlation acquiring unit, for combine it is corresponding with same user the match is successful the problem of keyword weight sum
And the maximum in the weight sum of field keyword, obtain the problem to be recommended and the degree of correlation of the user；And
Problem push unit, for the degree of correlation according to the problem to be recommended and each user, to be recommended is asked described
Topic is pushed to degree of correlation highest one or more user.
The embodiment of the present invention is by by the Weight of lists of keywords the problem of problem to be recommended and default each user
Key to the issue word list is matched, and obtains the weight sum of keyword the problem of the match is successful corresponding with each user,
The key to the issue word list is matched with all Weight fields lists of keywords of default each user respectively, obtained
Maximum in the weight sum of the corresponding all field keywords that the match is successful of each user, and combine and same use
Maximum in the weight sum of keyword that family is corresponding the problem of the match is successful and the weight sum of field keyword, is obtained
The problem to be recommended and the degree of correlation of the user, the problem to be recommended is pushed into degree of correlation highest, and one or more is used
Family, solve in Ask-Answer Community to user carry out question recommending when, exist to user recommend the problem of it is single and push use
The situation of the uninterested content in family, causes not high to the response rate of problem in Ask-Answer Community and Consumer's Experience effect is not good to ask
Topic, it is to avoid the problem of recommending to user is single and belongs to the problem of user loses interest in, so as to improve user to question and answer society
The response rate of problem in area, also improves experience effect of the user in Ask-Answer Community.
Brief description of the drawings
Fig. 1 be first embodiment of the invention provide Ask-Answer Community in the problem of recommend method implementation process figure；
Fig. 2 is the realization stream of the Weight key to the issue word list method for the acquisition user that second embodiment of the invention is provided
Cheng Tu；
Fig. 3 be third embodiment of the invention provide Ask-Answer Community in the problem of commending system structure chart；
Fig. 4 is the concrete structure diagram for the maximum acquiring unit that third embodiment of the invention is provided.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
The embodiment of the present invention is by using the Weight key to the issue word list of default each user and default each
All Weight fields lists of keywords of user, obtains the power of keyword the problem of the match is successful corresponding with each user
Maximum in weight sum and the weight sum of the corresponding all field keywords that the match is successful of each user, with reference to
Maximum in the weight sum of keyword that same user is corresponding the problem of the match is successful and the weight sum of field keyword
Value, obtains the degree of correlation of the problem to be recommended and the user, the degree of correlation is sorted from high to low, default high to the degree of correlation
Several users carries out the recommendation of problem, so as to when carrying out question recommending to user in Ask-Answer Community, can accurately recommend
To user it is interested the problem of etc. so that the experience of user is more preferably.
Implementing for the present invention is described in detail below in conjunction with specific embodiment：
Embodiment one：
Fig. 1 shows the implementation process for recommending method the problem of in the Ask-Answer Community that first embodiment of the invention is provided, in detail
State as follows：
In step S101, the problem of obtaining problem to be recommended lists of keywords.
In specific implementation process, after having new problem proposition, it is desirable to be able to answer the user of the problem to the new problem
Answered, yet with not knowing which field and/or which user etc. can be answered the new problem, or due to
The user that the new problem can be answered can not learn that the problem of it can be answered is suggested, then using provided in an embodiment of the present invention
In the lists of keywords for the problem of method for recommending problem can first obtain the proposition, the problem of in particular extracting the proposition, energy
Enough reflect keyword of the problem content information etc., the new problem may finally be recommended suitably according to the lists of keywords
User answers.Wherein, the key to the issue word list is the set of the keyword related to the content of the problem to be recommended, namely should
Related keyword is from reflecting the content information that the problem to be recommended is included.
In step s 102, by the key to the issue word list and the Weight key to the issue word list of default each user
Matched, obtain the weight sum of keyword the problem of the match is successful corresponding with each user.
Before step S102 is performed, method is recommended also to include the problem of in the Ask-Answer Community：
The problem of user once answers and refuses to answer list is obtained, sequence is extracted by the problem list according to time sequence
The problem of occurrence number is more than 1 in the list of afterwards the problem of keyword, and lists of keywords the problem of generate the user；
According to the Weight Acquisition relation of default key to the issue word, the Weight key to the issue word list of the user is obtained.
Wherein, the specific implementation process of the step of Weight key to the issue word list of each user of above-mentioned acquisition is for example following
Embodiment two, will not be repeated here.
Step S102 is specifically included：
The lists of keywords the problem of problem to be recommended and the Weight of a user in default all users are asked
Topic lists of keywords is matched, and obtains being present in simultaneously in the lists of keywords and the Weight key to the issue word list
With keyword the problem of success；
Based on the Weight key to the issue word list, weight corresponding with being somebody's turn to do the problem of the match is successful keyword is obtained；
Should with this match is successful the problem of the corresponding weight of keyword be added, obtain matches corresponding with a user
The weight sum of successful key to the issue word；
It is next user in default all users to update a user, repeats above-mentioned steps, until obtaining
After the weight sum for taking keyword the problem of the match is successful corresponding with each user in default all users, exit.
It is used as one embodiment of the invention, it is assumed that lists of keywords is { a1, a2, a3 } the problem of problem to be recommended, this is preset
All users be 3 users, respectively U1, U2, U3, and the corresponding Weight key to the issue word lists of U1 for U11, T11,
U12, T12 ... }, the corresponding Weight key to the issue word lists of U2 are { U21, T21, U22, T22 ... }, the corresponding cum rights of U3
Weight key to the issue word list is { U31, T31, U32, T32 ... }, and wherein U11 etc. is keyword, and T11 is keyword U11 power
Weight, after above-mentioned matching：
For U1, by lists of keywords the problem of the problem to be recommended and the progress of U1 Weight key to the issue word list
Match somebody with somebody, it is assumed that in the Weight key to the issue word list of a1 and U1 in the problem of obtaining the problem to be recommended lists of keywords
U11 is matched, then the weight sum of keyword the problem of the match is successful corresponding with U1 is T11；
For U2, by lists of keywords the problem of the problem to be recommended and the progress of U2 Weight key to the issue word list
Match somebody with somebody, it is assumed that a1, a2 in the problem of obtaining the problem to be recommended lists of keywords correspond to the Weight key to the issue with U2 respectively
U23, U24 matching in word list, then the weight sum of keyword the problem of the match is successful corresponding with U2 is (T23+
T24)；
For U3, by lists of keywords the problem of the problem to be recommended and the progress of U3 Weight key to the issue word list
Match somebody with somebody, it is assumed that the a1 in the problem of obtaining the problem to be recommended lists of keywords, a2, a3 respectively correspond to U3 in U33, U35,
U36 is matched, then the weight sum of keyword the problem of the match is successful corresponding with U2 is (T33+T35+T36).
In embodiments of the present invention, the Weight key to the issue word list of the user reflects user and answers or refuse back
Answer with the keyword the problem of frequency, thus recommend the problem of using in the present invention method to calculate to obtain that the match is successful
The problem of keyword weight sum, weighed the wish that user answers the problem.When user is repeatedly answered with a certain keyword
The problem of when, this keyword can have high weight in the lists of keywords of the user；Conversely, when user repeatedly refuses back
When answering with a certain crucial word problem, the weight of this word will decrease up to -1, obtain problem to be recommended and each user
The problem of lists of keywords matching degree or after claiming weight, question recommending can be had uncommon in other words to weight sum is larger
Hope the user answered.
In step s 103, all Weight fields by the key to the issue word list respectively with default each user are closed
Keyword list is matched, in the weight sum for obtaining all field keywords that the match is successful corresponding with each user
Maximum.
Before step S103 is performed, method is recommended also to include the problem of in the Ask-Answer Community：
Obtain the set that occurrence number in all problems of designated field is more than the field keyword of default value；
The first frequency that each keyword in the set occurs in all problems of the designated field is obtained, and is obtained
The second frequency for taking each keyword to occur in all problems of all spectra；
According to each keyword first frequency and the second frequency, each keyword of the designated field is obtained
Weight, and generate the Weight field lists of keywords of the designated field.
In specific implementation process, if for designated field c, being extracted in occurrence number in c and being more than or equal to a certain numerical value
The set W={ w1, w2 ... wn } of field keyword, the numerical value can be chosen according to actual conditions, such as can take 10 etc. whole
Number, the field keyword refers to the set of the keyword related to content of all the problems in c in the designated field, namely the phase
The keyword of the pass content information that all problems are included from reflection field c, to each keyword wi, i=1 in W,
2 ..., n, is calculated as below：
Note sum the problem of all spectra was delivered is N, and all spectra includes field a, b, c etc., remembers in N number of problem
The problem of comprising keyword wi, number was di, and the problem of remembering in a certain field c sum is Nc, remembers and keyword wi is included in the c of the field
The problem of number be Nci, then be keyword wi distribute a weight parameter cwi, according to following formula (1) obtain cwi, wherein,
Nci/Nc represents the first frequency that each keyword occurs in all problems of the designated field, and di/N represents this each
The second frequency that keyword occurs in all problems of all spectra,
Cwi=log ((Nci/Nc) × (N/di)) × log (N/di), (1)
If cwi≤0, remove wi from W, otherwise, retain the wi.
Afterwards, the weight wi of each keyword can also be normalized, obtains the specified neck after updating
Domain c Weight field lists of keywords, namely after each keyword in set W is calculated as above, take in set W
The cwi of all words maximum, is designated as max_cw, and computing is normalized to all cwi：Normalizing is represented with old_cwi
Cwi before change, the then cwi=old-_cwi/max_cw after normalizing.All word wi and associated weight cwi in set W
Composition field c Weight lists of keywords.
Summary step, then can obtain the Weight lists of keywords in the field of each in all spectra, can also
Referred to as Weight field lists of keywords, the Weight field lists of keywords is unrelated with sole user, but with all users
It is related.
Wherein, the weight of each keyword in the Weight field lists of keywords be scope (0,1] in reality
Number, represents some keyword degree relevant with this field.
Step S103 is specifically included：
A. the field involved by all problems that a user in default all users answered is obtained；
B. according to the Weight field lists of keywords in each field in default all spectra, obtain involved with this
All Weight fields lists of keywords of the corresponding user in field；
C. by lists of keywords the problem of the problem to be recommended and all Weight fields lists of keywords of the user
A Weight field lists of keywords matched, obtain simultaneously be present in the key to the issue word list and a cum rights
The field keyword that the match is successful in weight field lists of keywords；
D. the Weight field lists of keywords is based on, obtains with this that Weight field lists of keywords is corresponding matches into
The weight sum of the field keyword of work(；
E. the Weight field lists of keywords updated in all Weight fields lists of keywords of the user is
Next Weight field lists of keywords in all Weight fields lists of keywords of the user, repeat the above steps c
To d, until obtaining the field keyword that the match is successful corresponding with each Weight field lists of keywords of the user
Weight sum after, perform step f；
F. the maximum in the weight sum is obtained；
G. it is next user in default all users to update a user, repeats above-mentioned steps a to f,
Until in the weight sum of the acquisition field keyword that the match is successful corresponding with each user in default all users
Maximum.
It is used as one embodiment of the invention, it is assumed that lists of keywords is { a1, a2, a3 } the problem of problem to be recommended, default
All spectra is { q1, q2, q3, q4 }, and the corresponding field lists of keywords of q1 is { U11, T11, U12, T12 ... }, q2 correspondences
Field lists of keywords be { U21, T21, U22, T22 ... }, the corresponding field lists of keywords of q3 for U31, T31, U32,
T32 ... }, the corresponding field lists of keywords of q4 is { U41, T41, U42, T42 ... }, and default all users are 3
User, respectively U1, U2, U3, and assume that the field involved by all problems answered of U1 is { q2 }, and the institute that U2 was answered
Problematic involved field is { q1, q3 }, and the field involved by all problems answered of U3 is { q2, q3, q4 }, wherein
U11 etc. is keyword, and T11 etc. is keyword U11 weight, then after above-mentioned matching,
For U1, the field involved by all problems that U1 was answered be { q2 } with default all spectra for q1,
Q2, q3, q4 } matched, then corresponding Weight field lists of keywords is the Weight field keyword row corresponding to q2
Table U21, T21, U22, T22 ..., it is assumed that the cum rights of a1 and U1 in the problem of obtaining the problem to be recommended lists of keywords
U11 matchings in weight key to the issue word list, then the weight sum of keyword the problem of the match is successful corresponding with U1 is
T11, then the weight sum of the corresponding field keyword that the match is successful of U1 Weight field lists of keywords maximum
For T11；
For U2, the field involved by all problems that U2 was answered is that { q1, q3 } is with default all spectra
{ q1, q2, q3, q4 } is matched, then corresponding matching field is respectively q1, q3, q1, and the Weight field corresponding to q3 is crucial
Word list difference U11, T11, U12, T12 ..., U31, T31, U32, T32 ..., it is assumed that obtain the problem to be recommended
A1, a2 in key to the issue word list correspond to respectively with U2 Weight key to the issue word list U11, T11, U12,
T12 ... } in U13, U14 matching, and a1 the problem of the problem to be recommended in lists of keywords corresponds to the band with U2 respectively
U33, U34 matching in Weight lists of keywords { U31, T31, U32, T32 ... }, then corresponding with U2 the match is successful
The problem of keyword weight sum have two, be respectively (T13+T14), (T33+T34), if determining whether to learn (T13+
T14) it is more than (T33+T34), the then corresponding field keyword that the match is successful of U2 Weight field lists of keywords weight
The maximum of sum is also (T13+T14)；
For U3, the field involved by all problems that U3 was answered is that { q2, q3, q4 } is with default all spectra
{ q1, q2, q3, q4 } is matched, then corresponding matching field is respectively q2, q3, q4, q2, q3, the Weight neck corresponding to q4
Domain lists of keywords difference U21, T21, U22, T22 ..., U31, T31, U32, T32 ..., U41, T41, U42,
T42 ... }, the lists of keywords the problem of problem to be recommended is matched with U3 Weight key to the issue word list, it is false
If the a2 in the problem of obtaining the problem to be recommended lists of keywords corresponds to the Weight key to the issue word list with U3 respectively
U13 matchings in { U11, T11, U12, T12 ... }, and a3 the problem of the problem to be recommended in lists of keywords corresponds to respectively
With U23, U24 matching in U3 Weight key to the issue word list { U21, T21, U22, T22 ... }, and the problem to be recommended
The problem of lists of keywords in a1, a3 correspond to respectively with U3 Weight key to the issue word list U31, T31, U32,
T32 ... } in U33, U35, U36 matching, then the weight sum of keyword the problem of the match is successful corresponding with U2 is
(T13), (T23+T24), (T33+T35+T36), if determining whether to learn that (T33+T35+T36) is more than more than (T23+T24)
(T13), the then maximum of the weight sum of the corresponding field keyword that the match is successful of U3 Weight field lists of keywords
Also it is (T33+T35+T36).
So as to by above-mentioned steps, obtain that corresponding with each user in default all users the match is successful
Field keyword weight sum in maximum.
In embodiments of the present invention, the Weight field lists of keywords reflects the temperature feelings of keyword in a certain field
Condition, after step S104 is performed, for a certain user, if obtained corresponding in art the problem of involved by the user
A field weight sum it is maximum, even if then illustrating the problem of user did not answer the field, be likely to
The problem of field can be answered, so allow for recommend with recommendation problem answer in history the most people of the problem it
Outer people, namely the user, also cause a problem to have more Candidate Recommendation users, user has more select permeabilities.
In step S104, with reference to it is corresponding with same user the match is successful the problem of keyword weight sum and neck
Maximum in the weight sum of domain keyword, obtains the degree of correlation of the problem to be recommended and the user.
Wherein, step S104 is specifically included：
By same user it is corresponding the match is successful the problem of keyword weight sum and field keyword weight it
Maximum with is added；
The numerical value of rear gained be will add up as the problem to be recommended and the degree of correlation of the user.
In step S105, according to the problem to be recommended and the degree of correlation of each user, the problem to be recommended is pushed to
One or more user of degree of correlation highest.
In specific implementation process, after the degree of correlation for obtaining the problem to be recommended and each user, if there are multiple use
Family, then can obtain multiple degrees of correlation corresponding with the plurality of user, and the plurality of degree of correlation can be sorted from high to low,
The corresponding several users of the degree of correlation ranked in the top are obtained, will treat that the problem of band is recommended recommends this several users, the reception
User's number of question recommending can be set according to actual needs.
In embodiments of the present invention, when to be recommended based on this problem of, lists of keywords score the problem of the user of acquisition
Or weight sum is relatively low, and the field lists of keywords score of user or claim weight sum maximum it is higher when, explanation
The user answered with this it is to be recommended the problem of it is similar the problem of, and problem art as such to be recommended asks with this
Inscribe art related, represent the user there is a strong possibility that the problem of this is to be recommended can be answered, so allowing for problem may push away
Recommend to the people answered in history outside the most people of the problem so that a problem there are more Candidate Recommendation users, also uses
There are more select permeabilities at family.
In embodiments of the present invention, the degree of correlation of each to be recommended the problem of and user are closed by the Weight problem of the user
The field keyword in the involved field of the problem of keyword list recommends the matching degree of problem and the user to answer with the band
List and the band recommend the matching degree of problem to be combined into so that when recommending problem to user, it is to avoid same user
The problem of receiving same, and can not obtain the answer answer of other users, it also avoid user and receives that excessive he is unanswerable
Problem and be fed up with, such as, and the negative weight of the keyword due to introducing user, this method can pushed to user it is new
While fresh problem, it is to avoid push the problem of he loses interest in.If user is to certain class problem without answer, key therein
Word weight will be reduced gradually, so that system will not recommend such problem for user, because the method for the recommendation problem is utilized
To field antistop list, it can also attempt to recommend the other problems in the field for user.
Embodiment two：
Fig. 2 shows the method for the Weight key to the issue word list for the acquisition user that second embodiment of the invention is provided
Implementation process, it is necessary to obtain this before recommendation method recommends new problem to a certain user the problem of in using the Ask-Answer Community
The Weight key to the issue word list of user, details are as follows：
In step s 201, obtain user and once answer and refuse all problems answered, by all problems by generation
Time order and function sorts.
In step S202, the problem of occurrence number is more than 1 in the problem of extracting after sequence keyword, and form the user
The problem of lists of keywords.
In step S203, the initial weight for setting each keyword in the key to the issue word list is 0.
In specific implementation process, the problem of a certain user U is once answered and refused to answer list Q is collected first, by Q
The problem of by time of origin successively sort, extract Q in occurrence number be more than 1 the problem of keyword set, be used as key to the issue
Word list, it is W={ w1, w2 ..., wn } to set the key to the issue word list, and for each keyword wi, wherein i=1,
2 ..., n, it is 0 to set a real parameters or initial weight Ulrwi, wherein i=1,2 ..., n, Ulrwi initial value, and
One learning rate constant RATE is set, can learn that RATE span is (0,1) according to practical experience, such as, can be with
RATE=0.02 is set.
In step S204, the problem of obtaining first problem in all problems after sequence label and the problem are closed
All crucial word problem parameters of keyword list.
In step S205, label, the initial weight of all keywords, problem parameter with reference to the problem of the first problem
And according to default Weight Acquisition relation, obtain the first weight of the first keyword in the key to the issue word list.
In step S206, the initial weight for updating first keyword is the first weight of first keyword.
In specific implementation process, to the first problem in Q, it is calculated as below：
The label label of the first problem is write down, if user U answers the problem, problem label label=1,
Otherwise label=0.If keyword wi is appeared in the first problem, problem parameter xi=1, otherwise xi=0.It is then right
Each keyword wi in set W, can obtain corresponding problem parameter, according to the following equation or can claim default power
Recapture and take relation (2), (3), (4) calculating to obtain delta_wi：
Delta_wi=(label-p) × RATE × xi, (4)
For keyword wi, according to assignment method old_Ulrwi=Ulrwi, Ulrwi=old_Ulrwi+delta_wi is more
Newly initial weight corresponding with keyword wi is the first weight Ulrwi.
In step S207, judge whether to have traveled through all keywords in the key to the issue word list, be then to go to and hold
Row step S209, it is no, then perform step S208.
In step S208, next keyword of the key to the issue word list is regard as the first keyword.
In specific implementation process, when have updated first keyword of the key to the issue word list in the first problem
, it is necessary to continue to update weight of other keywords in the first problem, until updating the key to the issue after shared weight
The weight of all keywords in word list, afterwards, all keywords for continuing to update in the lists of keywords are asked next
Shared weight in topic.
In step S209, judge whether to have traveled through all problems, be then execution step S211, backed off after random, otherwise hold
Row step S210.
In step S210, next problem in all problems after this is sorted is used as first problem.
In step S211, the weight of all keywords in lists of keywords the problem of after updating is obtained.
Specifically, weights of each keyword wi in the first problem in W can be obtained using above-mentioned steps.
It has updated after the corresponding parameter Ulrwi of each keyword wi, in the title of the key words asked as next using the Ulrwi after renewal
Wi initial parameter, that is, to next problem in Q, if occurring in that the wi in next problem, with wi correspondences
Renewal after Ulrwi values as initial value re-execute above-mentioned steps, continue to update the Ulrwi, until keyword wi is in Q
In some problem last time when occurring, Ulrwi values after renewal as keyword wi final real parameters value.If certainly
To next problem in Q, wi no longer occurs, then after the corresponding final real parameters value Ulrwi of wi then update for last time
Value.So as to the problem of may finally obtaining the user U using above-mentioned steps keyword W Weight key to the issue word list
ULRW={ Ulrw1, Ulrw2 ..., Ulrwn }, Ulrwi represent user U and keyword wi weight, i=1,2 ..., n.
In step S212, according to what is occurred comprising one in the list crucial word problem number in all problems
Frequency, and the byte number shared by one of keyword, continue to update the weight of all keywords in the list.
In step S213, to being exported after the key to the issue word list normalization after renewal.
In embodiments of the present invention, also continue to be calculated as below then to each keyword wi in W：
The problem of whole users of note delivered sum is asking comprising keyword wi in N, the problem of whole users delivered
Topic number is di, then distributes a parameter idfi=ln (N/di) for keyword wi, and wherein ln is natural logrithm function, is also key
Word wi distribute a parameter bytei, be designated as the byte number that the keyword takes in calculator memory, the bytei be, it is known that after
The continuous associated weight for calculating user U and keyword wi is Uwi, Uwi=Ulrwi*idfi*ln (bytei), then according to above-mentioned calculating
Can obtain in all problems, user U and each keyword wi associated weight Uwi, be designated as list UW=Uw1,
Uw2 ..., Uwn }, the maximum of all Uwi, i=1,2 ..., n absolute value in set UW is taken, max_Uw is designated as, and to institute
Computing is normalized in some Uwi, and old_Uwi represents the Uwi before normalization, then the Uuwi=old_Uwi after normalizing
/max_Uw。
To sum up it is somebody's turn to do, all keyword wi associated weight Uwis corresponding with the wi in set W constitutes user U cum rights
Weight key to the issue word list, wherein, the weight of key to the issue word is real number of the scope in [- 1,1], is received on the occasion of expression user
The degree of the keyword, negative value then represents that user refuses the degree of the keyword.
In embodiments of the present invention, the problem of user answers and refuses it can be seen from situation, if a certain key to the issue word pair
The weighted value answered is higher, then illustrates that the user has the ability or had a mind to answer and include the key to the issue word problem etc., otherwise,
The user may be less interested in the problem or even refuses to answer comprising the key to the issue word problem etc., so that the band of user
Weight lists of keywords reflects the situation of user's process problem in community's question and answer to a certain extent, then the question and answer society
The administrative staff in area wait then targetedly can recommend problem so that Yong Huneng according to the question and answer situation of the user to user
The problem of enough obtaining interested in time, it helps usage experience of lifting user etc..
Can be with one of ordinary skill in the art will appreciate that realizing that all or part of step in above-described embodiment method is
The hardware of correlation is instructed to complete by program, the program being somebody's turn to do can be stored in a computer read/write memory medium, should
Storage medium, such as ROM/RAM, disk, CD.
Embodiment three：
Fig. 3 shows the structure of commending system the problem of in the Ask-Answer Community that third embodiment of the invention is provided, in order to just
In explanation, the part related to the embodiment of the present invention illustrate only.
The system that the problem of in the Ask-Answer Community recommends includes lists of keywords and obtains master unit 31, weight sum acquisition list
Member 32, maximum acquiring unit 33, degree of correlation acquiring unit 34 and problem push unit 35, wherein：
Lists of keywords obtains master unit 31, lists of keywords the problem of for obtaining problem to be recommended.
In embodiments of the present invention, after having new problem proposition, it is desirable to be able to answer the user of the problem to the new problem
Answered, yet with not knowing which field and/or which user etc. can be answered the new problem, or due to
The user that the new problem can be answered can not learn that the problem of it can be answered is suggested, then using provided in an embodiment of the present invention
Method for recommending problem can obtain the lists of keywords for the problem of master unit 31 first obtains the proposition using lists of keywords, according to
The lists of keywords may finally recommend the new problem suitable user's answer.Wherein, the key to the issue word list be with
The set of the related keyword of the content of the problem to be recommended, namely the related keyword from reflection, wrapped by the problem to be recommended
The content information contained.
Weight sum acquiring unit 32, for by the Weight problem of the key to the issue word list and default each user
Lists of keywords is matched, and obtains the weight sum of keyword the problem of the match is successful corresponding with each user.
In embodiments of the present invention, before triggering weight sum acquiring unit 32, recommend the problem of in the Ask-Answer Community
System also includes：
List generation unit, for obtaining the problem of each user once answers and refuses to answer list, the problem is arranged
Table according to time sequence, extracts the problem of occurrence number is more than 1 in list the problem of after sequence keyword, and generate asking for the user
Inscribe lists of keywords；And
Weight list acquiring unit, for the Weight Acquisition relation according to default key to the issue word, obtains the user
Weight key to the issue word list.
Wherein, the weight of key to the issue word is real number of the scope in [- 1,1], and the keyword is received on the occasion of expression user
Degree, negative value then represents that user refuses the degree of the keyword.And obtain the Weight key to the issue word list of each user
The step of for example above-mentioned embodiment two of specific implementation process, will not be repeated here.
In addition, the weight sum acquiring unit 32 is specifically included：
Matching keywords acquiring unit 321, for lists of keywords the problem of the problem to be recommended to be owned with default
The Weight key to the issue word list of a user in user is matched, and is obtained and is present in the lists of keywords simultaneously and is somebody's turn to do
The problem of the match is successful in Weight key to the issue word list keyword；
Weight Acquisition unit 322, for based on the Weight key to the issue word list, obtaining with being somebody's turn to do the problem of the match is successful
The corresponding weight of keyword；
Weight sum obtains subelement 323, for should with this match is successful the problem of keyword corresponding weight addition,
Obtain the weight sum of keyword the problem of the match is successful corresponding with a user；And
Updating block 324, is next user in default all users for updating a user, triggers this
With keyword acquiring unit, corresponding with each user in default all users closed until obtaining the problem of the match is successful
After the weight sum of keyword, exit.
In embodiments of the present invention, the Weight key to the issue word list of the user reflects user and answers or refuse back
Answer with the keyword the problem of frequency, thus recommend the problem of using in the present invention method to calculate to obtain that the match is successful
The problem of keyword weight sum, weighed the wish that user answers the problem.When user is repeatedly answered with a certain keyword
The problem of when, this keyword can have high weight in the lists of keywords of the user；Conversely, when user repeatedly refuses back
When answering with a certain crucial word problem, the weight of this word will decrease up to -1, obtain problem to be recommended and each user
The problem of lists of keywords matching degree or after claiming weight, question recommending is had to most weight sum is larger can in other words
Wish the user answered.
Maximum acquiring unit 33, for all cum rights by the key to the issue word list respectively with default each user
Weight field lists of keywords is matched, the power of acquisition all field keywords that the match is successful corresponding with each user
Maximum in weight sum.
In embodiments of the present invention, before triggering maximum acquiring unit 33, system is recommended the problem of in the Ask-Answer Community
System includes：
Gather the field pass that occurrence number in acquiring unit, all problems for obtaining designated field is more than default value
The set of keyword；
Frequency acquisition unit, each keyword for obtaining in the set goes out in all problems of the designated field
Existing first frequency, and obtain the second frequency that each keyword occurs in all problems of all spectra；And
Weight field list generation unit, for according to each keyword first frequency and the second frequency, obtaining
The weight of each keyword of the designated field is taken, and generates the Weight field lists of keywords of the designated field.
In embodiments of the present invention, generated using gathering acquiring unit, frequency acquisition unit and the list of Weight field
Unit can obtain the Weight lists of keywords in the field of each in all spectra, be referred to as Weight field keyword
List, the Weight field lists of keywords is unrelated with sole user, but related to all users.The Weight field is crucial
The weight of each keyword in word list be scope (0,1] in real number, indicate some keyword relevant with this field
Degree.
As shown in figure 4, the maximum acquiring unit 33, which is specifically included, is related to field acquiring unit 41, the acquisition of field keyword
It is single that unit 42, matching keywords acquiring unit 43, weight sum acquiring unit 44, the first updating block 45, maximum obtain son
The updating block 47 of member 46 and second, wherein：
It is related to field acquiring unit 41, for obtaining all problems that a user in default all users answered
Involved field；
Field keyword acquiring unit 42, it is crucial for the Weight field according to each field in default all spectra
Word list, obtains all Weight fields lists of keywords of the user corresponding with the involved field；
Matching keywords acquiring unit 43, for by lists of keywords the problem of the problem to be recommended respectively with the user's
A Weight field lists of keywords in the lists of keywords of all Weight fields is matched, and obtains exist simultaneously respectively
The field keyword that the match is successful in the key to the issue word list and a Weight field lists of keywords；
Weight sum acquiring unit 44, for based on the Weight field lists of keywords, obtaining and the Weight field
The weight sum of the corresponding field keyword that the match is successful of lists of keywords；
First updating block 45, the Weight in the lists of keywords of all Weight fields for updating the user
Field lists of keywords is next Weight field keyword row in all Weight fields lists of keywords of the user
Table, triggers the matching keywords acquiring unit 43, until obtaining each Weight field lists of keywords with the user
After the weight sum of the corresponding field keyword that the match is successful, triggering maximum obtains subelement 46；
Maximum obtains subelement 46, for obtaining the maximum in the weight sum；And
Second updating block 47, is next user in default all users, triggering for updating a user
This is related to field acquiring unit 41, until obtaining the neck that the match is successful corresponding with each user in default all users
Maximum in the weight sum of domain keyword.
In embodiments of the present invention, the Weight field lists of keywords reflects the temperature feelings of keyword in a certain field
Condition, for a certain user, if obtain a corresponding field in art the problem of involved by the user weight it
And maximum, even if then illustrating the problem of user did not answer the field, it is likely to that asking for the field can be answered
Topic, so allows for that the people answered in history outside the most people of the problem may be recommended with recommendation problem, namely the user,
Also so that a problem has more Candidate Recommendation users, user has more select permeabilities.
Degree of correlation acquiring unit 34, for combine it is corresponding with same user the match is successful the problem of keyword weight it
With and the weight sum of field keyword in maximum, obtain the degree of correlation of the problem to be recommended and the user.
Problem push unit 35, for the degree of correlation according to the problem to be recommended and each user, by the problem to be recommended
It is pushed to degree of correlation highest one or more user.
In embodiments of the present invention, degree of correlation acquiring unit 34 by same user it is corresponding the match is successful the problem of keyword
Weight sum and field keyword weight sum in maximum be added, the numerical value that will add up rear gained is used as this
Problem to be recommended and the degree of correlation of the user.After the degree of correlation for obtaining the problem to be recommended and each user, then Utilizing question
Push unit 35 can obtain multiple degrees of correlation corresponding with the plurality of user, and the plurality of degree of correlation is arranged from high to low
Sequence, obtains the corresponding several users of the degree of correlation ranked in the top, will treat that the problem of band is recommended recommends this several users, this connects
Receiving user's number of question recommending can set according to actual needs.
In embodiments of the present invention, in the Ask-Answer Community the problem of commending system by user recommend problem when, energy
Enough Weight key to the issue word lists according to the user and asking that the matching degree of band recommendation problem and the user were answered
The field lists of keywords in the involved field of topic recommends the matching degree of problem with the band, to obtain the problem and the phase of user
Pass degree, so that by the question recommending user higher to degree of correlation so that user obtains to the response rate of problem in Ask-Answer Community
To raising, experience effect of the user in Ask-Answer Community gets a promotion.
The embodiment of the present invention is by by the Weight of lists of keywords the problem of problem to be recommended and default each user
Key to the issue word list is matched, and obtains the weight sum of keyword the problem of the match is successful corresponding with each user,
The key to the issue word list is matched with all Weight fields lists of keywords of default each user respectively, obtained
Maximum in the weight sum of the corresponding all field keywords that the match is successful of each user, and combine and same use
Maximum in the weight sum of keyword that family is corresponding the problem of the match is successful and the weight sum of field keyword, is obtained
The problem to be recommended and the degree of correlation of the user, the problem to be recommended is pushed into degree of correlation highest, and one or more is used
Family, the problem of solving not high to the response rate of problem in Ask-Answer Community and not good Consumer's Experience effect, improves user to asking
The response rate of problem in community is answered, experience effect of the user in Ask-Answer Community is also improved.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
Any modifications, equivalent substitutions and improvements made within refreshing and principle etc., should be included in the scope of the protection.
Claims (10)
1. the problem of in a kind of Ask-Answer Community, recommends method, it is characterised in that methods described comprises the steps：
The problem of obtaining problem to be recommended lists of keywords；
Described problem lists of keywords is matched with the Weight key to the issue word list of default each user, obtain with
The weight sum of each user is corresponding the problem of the match is successful keyword, wherein, the Weight key to the issue word of user
List is used to reflect that user answers or refusal answers the frequency with any crucial word problem in list；
By all Weight fields lists of keywords progress of described problem lists of keywords respectively with default each user
Match somebody with somebody, obtain the maximum in the weight sum of all field keywords that the match is successful corresponding with each user；
With reference to it is corresponding with same user the match is successful the problem of keyword weight sum and field keyword weight it
Maximum with, obtains the problem to be recommended and the degree of correlation of the user；
According to the problem to be recommended and the degree of correlation of each user, the problem to be recommended is pushed to degree of correlation highest one
Individual or multiple users.
2. the method as described in claim 1, it is characterised in that perform it is described by described problem lists of keywords with it is default
Before the step of Weight key to the issue word list of each user is matched, methods described also includes：
Obtain the problem of each user once answers and refuses to answer list, by described problem list according to time sequence, the row of extraction
The problem of occurrence number is more than 1 in the list of the problem of after sequence keyword, and lists of keywords the problem of generate the user；
According to the Weight Acquisition relation of default key to the issue word, the Weight key to the issue word list of the user is obtained.
3. the method as described in claim 1, it is characterised in that perform it is described by described problem lists of keywords respectively with advance
If all Weight fields lists of keywords of each user the step of matched before, methods described also includes：
Obtain the set that occurrence number in all problems of designated field is more than the field keyword of default value；
The first frequency that each keyword in the set occurs in all problems of the designated field is obtained, and is obtained
The second frequency for taking each described keyword to occur in all problems of all spectra；
According to each described keyword first frequency and the second frequency, each keyword of the designated field is obtained
Weight, and form the Weight field lists of keywords of the designated field.
4. the method as described in claim 1, it is characterised in that described by described problem lists of keywords and default each use
The Weight key to the issue word list at family is matched, and obtains keyword the problem of the match is successful corresponding with each user
Weight sum the step of specifically include：
By the Weight problem of a user in lists of keywords the problem of the problem to be recommended and default all users
Lists of keywords is matched, and is obtained and is present in simultaneously in the lists of keywords and the Weight key to the issue word list
The problem of the match is successful keyword；
Based on the Weight key to the issue word list, obtain with it is described the match is successful the problem of the corresponding weight of keyword；
Will it is described with it is described the match is successful the problem of the corresponding weight of keyword be added, corresponding with one user of acquisition
Weight sum with keyword the problem of success；
It is next user in default all users to update one user, repeats above-mentioned steps, until obtaining
After the weight sum of keyword the problem of the match is successful corresponding with each user in default all users, exit.
5. the method as described in claim 1, it is characterised in that it is described by described problem lists of keywords respectively with it is default every
All Weight fields lists of keywords of one user is matched, obtain all matchings corresponding with each user into
The step of maximum in the weight sum of the field keyword of work(, specifically includes：
A. the field involved by all problems that a user in default all users answered is obtained；
B. according to the Weight field lists of keywords in each field in default all spectra, obtain and the involved neck
All Weight fields lists of keywords of the corresponding user in domain；
C. by lists of keywords the problem of the problem to be recommended and all Weight fields lists of keywords of the user
A Weight field lists of keywords matched, obtain simultaneously be present in described problem lists of keywords and one
The field keyword that the match is successful in the lists of keywords of Weight field；
D. based on each described Weight field lists of keywords, obtain and each described Weight field lists of keywords
The weight sum of the corresponding field keyword that the match is successful；
E. the Weight field lists of keywords updated in all Weight fields lists of keywords of the user is institute
Next Weight field lists of keywords in all Weight fields lists of keywords of user is stated, repeat the above steps c
To d, until it is crucial to obtain the field that the match is successful corresponding with each Weight field lists of keywords of the user
After the weight sum of word, step f is performed；
F. the maximum in the weight sum is obtained；
G. it is next user in default all users to update one user, repeats above-mentioned steps a to f, directly
Into the weight sum for obtaining the field keyword that the match is successful corresponding with each user in default all users
Maximum.
6. the method as described in claim 1, it is characterised in that the weight of described problem keyword is that scope belongs to [- 1,1]
Real number, the weight of the field keyword belong to for scope (0,1] real number.
7. the commending system of the problem of in a kind of Ask-Answer Community, it is characterised in that the system includes：
Lists of keywords obtains master unit, lists of keywords the problem of for obtaining problem to be recommended；
Weight sum acquiring unit, for by the Weight key to the issue of described problem lists of keywords and default each user
Word list is matched, and obtains the weight sum of keyword the problem of the match is successful corresponding with each user, wherein, use
The Weight key to the issue word list at family is used to reflecting that user to be answered or refusal is answered and asked with any keyword in list
The frequency of topic；
Maximum acquiring unit, leads for all Weights by described problem lists of keywords respectively with default each user
Domain lists of keywords is matched, and obtains the weight of all field keywords that the match is successful corresponding with each user
Maximum in sum；
Degree of correlation acquiring unit, for combine it is corresponding with same user the match is successful the problem of keyword weight sum and
Maximum in the weight sum of field keyword, obtains the problem to be recommended and the degree of correlation of the user；And
Problem push unit, for the degree of correlation according to the problem to be recommended and each user, the problem to be recommended is pushed away
Give degree of correlation highest one or more user.
8. system as claimed in claim 7, it is characterised in that the system also includes：
List generation unit, for obtaining the problem of each user once answers and refuses to answer list, by described problem list
According to time sequence, the problem of occurrence number is more than 1 in list the problem of after sequence keyword is extracted, and generates asking for the user
Inscribe lists of keywords；And
Weight list acquiring unit, for the Weight Acquisition relation according to default key to the issue word, obtains the user's
Weight key to the issue word list.
9. system as claimed in claim 7, it is characterised in that the system also includes：
Gather the field keyword that occurrence number in acquiring unit, all problems for obtaining designated field is more than default value
Set；
Frequency acquisition unit, each keyword for obtaining in the set goes out in all problems of the designated field
Existing first frequency, and obtain the second frequency that each described keyword occurs in all problems of all spectra；And
Weight field list generation unit, for according to each described keyword first frequency and the second frequency, obtaining
The weight of each keyword of the designated field is taken, and generates the Weight field lists of keywords of the designated field.
10. system as claimed in claim 7, it is characterised in that the weight sum acquiring unit is specifically included：
Matching keywords acquiring unit, for by lists of keywords the problem of the problem to be recommended and default all users
The Weight key to the issue word list of a user matched, obtain and be present in the lists of keywords and the band simultaneously
The problem of the match is successful in Weight lists of keywords keyword；
Weight Acquisition unit, for based on the Weight key to the issue word list, obtain with it is described the match is successful the problem of close
The corresponding weight of keyword；And
Weight sum obtains subelement, for will it is described with it is described the match is successful the problem of the corresponding weight addition of keyword, obtain
Take the weight sum of keyword the problem of the match is successful corresponding with one user；And
Updating block, is next user in default all users for updating one user, triggers the matching
Keyword acquiring unit, corresponding with each user in default all users is closed the problem of the match is successful until obtaining
After the weight sum of keyword, exit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210095917.5A CN103365899B (en) | 2012-04-01 | 2012-04-01 | The problem of in a kind of Ask-Answer Community, recommends method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210095917.5A CN103365899B (en) | 2012-04-01 | 2012-04-01 | The problem of in a kind of Ask-Answer Community, recommends method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103365899A CN103365899A (en) | 2013-10-23 |
| CN103365899B true CN103365899B (en) | 2017-10-20 |
Family
ID=49367263
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210095917.5A Active CN103365899B (en) | 2012-04-01 | 2012-04-01 | The problem of in a kind of Ask-Answer Community, recommends method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103365899B (en) |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104699696B (en) * | 2013-12-05 | 2018-12-28 | 深圳市腾讯计算机系统有限公司 | File recommendation method and device |
| CN103761314A (en) * | 2014-01-26 | 2014-04-30 | 句容云影响软件技术开发有限公司 | Multifunctional conversation information control method |
| WO2016101765A1 (en) * | 2014-12-25 | 2016-06-30 | 北京奇虎科技有限公司 | Question-and-answer page related question recommendation method and device |
| JP6604542B2 (en) * | 2015-04-02 | 2019-11-13 | パナソニックＩｐマネジメント株式会社 | Dialogue method, dialogue program and dialogue system |
| CN106294058A (en) * | 2015-05-28 | 2017-01-04 | 腾讯科技（深圳）有限公司 | The target strategy processing problems of operation document determines method and device |
| CN105205159B (en) * | 2015-09-29 | 2020-06-02 | 陈中和 | Device and method for automatically feeding back information |
| CN105912697B (en) * | 2016-04-25 | 2019-08-27 | 北京光年无限科技有限公司 | A kind of optimization method and device of conversational system knowledge base |
| CN107436900B (en) * | 2016-05-26 | 2022-08-09 | 北京搜狗科技发展有限公司 | Information processing method and device based on search engine |
| CN108090055A (en) * | 2016-11-21 | 2018-05-29 | 凌今 | The help information method for pushing and equipment of a kind of scientific research interaction platform |
| CN106997383A (en) * | 2017-03-22 | 2017-08-01 | 山东大学 | The problem of based on PageRank algorithms, recommends method and system |
| CN107451199B (en) * | 2017-07-05 | 2020-06-26 | 阿里巴巴集团控股有限公司 | Question recommendation method, device and equipment |
| CN107483595B (en) * | 2017-08-23 | 2023-11-21 | 北京百度网讯科技有限公司 | Information pushing method and device |
| CN107766536A (en) * | 2017-10-30 | 2018-03-06 | 江西博瑞彤芸科技有限公司 | The searching method of related information |
| CN107992604B (en) * | 2017-12-14 | 2020-08-28 | 北京搜狗科技发展有限公司 | Task item distribution method and related device |
| CN109460504A (en) * | 2018-09-21 | 2019-03-12 | 广州神马移动信息科技有限公司 | The answer main body recommended method and its device, electronic equipment, computer-readable medium of answer are reserved in Knowledge Community |
| CN109992602B (en) * | 2019-04-02 | 2023-05-16 | 海南颖川科技有限公司 | Digital reading guiding equipment for children |
| CN111538826B (en) * | 2020-07-13 | 2021-01-29 | 智者四海（北京）技术有限公司 | Routing invitation method and device based on problems |
| CN112765326B (en) * | 2021-01-27 | 2023-04-21 | 西安电子科技大学 | Question-answering community expert recommendation method, system and application |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6618722B1 (en) * | 2000-07-24 | 2003-09-09 | International Business Machines Corporation | Session-history-based recency-biased natural language document search |
| CN101257512A (en) * | 2008-02-02 | 2008-09-03 | 黄伟才 | Inquiry answer matching method used for inquiry answer system as well as inquiry answer method and system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120030126A1 (en) * | 2008-11-26 | 2012-02-02 | Mccagg Brin | Structured Job Search Engine with Guided Navigation |
-
2012
- 2012-04-01 CN CN201210095917.5A patent/CN103365899B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6618722B1 (en) * | 2000-07-24 | 2003-09-09 | International Business Machines Corporation | Session-history-based recency-biased natural language document search |
| CN101257512A (en) * | 2008-02-02 | 2008-09-03 | 黄伟才 | Inquiry answer matching method used for inquiry answer system as well as inquiry answer method and system |
Non-Patent Citations (1)
| Title |
|---|
| 面向问答系统的答案获取方法研究与实现;呼大为;《中国博士学位论文全文数据库 信息科技辑》;20090615(第06（2009）期);第5.3.1节第1-5段，图5.3 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103365899A (en) | 2013-10-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103365899B (en) | The problem of in a kind of Ask-Answer Community, recommends method and system | |
| CN110046236B (en) | Unstructured data retrieval method and device | |
| CN105144164B (en) | Scoring concept terms using a deep network | |
| CN106682169B (en) | Application label mining method and device, application searching method and server | |
| EP2137638B1 (en) | A method for the extraction of relation patterns from articles | |
| US11264006B2 (en) | Voice synthesis method, device and apparatus, as well as non-volatile storage medium | |
| CN105630938A (en) | Intelligent question-answering system | |
| CN106970991B (en) | Similar application identification method and device, application search recommendation method and server | |
| CN109819015B (en) | Information pushing method, device and equipment based on user portrait and storage medium | |
| WO2014210387A2 (en) | Concept extraction | |
| CN111522886B (en) | Information recommendation method, terminal and storage medium | |
| CN108986910A (en) | Answering method, device, computer equipment and storage medium on line | |
| Dethlefs et al. | Conditional random fields for responsive surface realisation using global features | |
| CN104978314A (en) | Media content recommendation method and device | |
| CN114064910A (en) | Knowledge graph construction method and system | |
| CN112650842A (en) | Human-computer interaction based customer service robot intention recognition method and related equipment | |
| CN104375977B (en) | The processing method and processing device of reply message in Ask-Answer Community | |
| CN114780861B (en) | Clustering technology-based user multi-interest recommendation method, device, equipment and medium | |
| Adeleke et al. | A two-step feature selection method for quranic text classification | |
| US20180276286A1 (en) | Metadata Extraction and Management | |
| CN110717038A (en) | Object classification method and device | |
| Stewart et al. | Cognitive redeployment in ACT-R: Salience, vision, and memory | |
| CN104376034B (en) | Information processing equipment, information processing method and program | |
| Erni et al. | Wild hopes: sourcing the political vocabulary of digital citizenship from the Lihkg forum | |
| CN110019768B (en) | Method and device for generating text abstract |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| ASS | Succession or assignment of patent right |
Owner name: SHENZHEN SHIJI LIGHT SPEED INFORMATION TECHNOLOGY Free format text: FORMER OWNER: TENGXUN SCI-TECH (SHENZHEN) CO., LTD. Effective date: 20131029 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| COR | Change of bibliographic data |
Free format text: CORRECT: ADDRESS; FROM: 518044 SHENZHEN, GUANGDONG PROVINCE TO: 518057 SHENZHEN, GUANGDONG PROVINCE |
|
| TA01 | Transfer of patent application right |
Effective date of registration: 20131029 Address after: A Tencent Building in Shenzhen Nanshan District City, Guangdong streets in Guangdong province science and technology 518057 16 Applicant after: Shenzhen Shiji Guangsu Information Technology Co., Ltd. Address before: Shenzhen Futian District City, Guangdong province 518044 Zhenxing Road, SEG Science Park 2 East Room 403 Applicant before: Tencent Technology (Shenzhen) Co., Ltd. |
|
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |