CN104572982A - Personalized recommendation method and system based on question guide - Google Patents
Personalized recommendation method and system based on question guide Download PDFInfo
- Publication number
- CN104572982A CN104572982A CN201410856318.XA CN201410856318A CN104572982A CN 104572982 A CN104572982 A CN 104572982A CN 201410856318 A CN201410856318 A CN 201410856318A CN 104572982 A CN104572982 A CN 104572982A
- Authority
- CN
- China
- Prior art keywords
- article
- user
- theme
- guiding
- behavior
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
- G06F16/337—Profile generation, learning or modification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Genetics & Genomics (AREA)
- Physiology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a personalized recommendation method and a personalized recommendation system based on question guide, wherein the method comprises the following steps: a first step: obtaining a semantic subject or a behavior subject and constructing a plurality of question guide trees, wherein the semantic subject is obtained according to a question input by a user, and the behavior subject is obtained according to viewed content of the user; performing data filtration on the user behavior data, the user grading data and the object element data to obtain a plurality of object subjects, and constructing question guide trees which are in one-to-one correspondence with the object subjects based on a genetic algorithm; a second step: determining question guide trees through subject matching, and obtaining liked subjects according to the semantic subject or the behavior subject; matching the liked subjects with the question guide trees, selecting the question guide trees matched with the liked subjects, performing question output to the user according to the selected question guide trees, and obtaining the data liked by the user to perform personalized recommendation. The personalized recommendation method can be used for solving a cold-starting problem of the user, and improving the user experience.
Description
Technical field
The present invention relates to article recommended technology field, more specifically, relate to a kind of personalized recommendation method based on problem guiding and system.
Background technology
At present, personalized recommendation system obtains and uses widely in internet, applications, wherein, personalized recommendation algorithm often all depends on the content of text data of click traffic data and the generation produced in user interaction process, excavates potential hobby and the demand of user according to these mass datas.
Due to this feature of personalized recommendation algorithm, this industry common problem of user's cold start-up all can be faced when recommending for new user, do not have due to new user or only have little interaction data, in actual application, be all often that new user more needs the auxiliary of commending system, old user not too depends on machine recommendation on the contrary to applying more to understand.
For the cold start-up problem of new user, the collective strategy that usual employing is traditional, the method utilizing rank to supply using article the highest for the most popular scoring as recommendations, the recommend method of the overall rank of this employing have lost personalized feature, content is all the same to cause everyone to see, is difficult to the demand meeting long-tail user.
Another cold start-up way to solve the problem for new user is: by the mode of booting problem, the firsthand data by forcing the mode of answering a question to gather user after user's registration, thus carries out personalized recommendation based on these data.Current is adopt quantity of information (Entropy) algorithm to the data processing method of booting problem, makes to obtain maximum quantity of information by the problem of minimum number.Its specific implementation thinking is by analyzing the evaluating data collection of known users to article, carrys out Construct question homing tree as by the two-layer problem guiding tree shown in Fig. 1 as choosing suitable article.
As shown in Figure 1, the article A-article M that each node is corresponding is best article of puing question to, the feedback to upper strata article (like, do not like and do not know) of the limit representative of consumer in homing tree.First such as problem guiding tree can inquire whether user likes article A, and do not know if user answers, homing tree can then put question to user whether to like article C, and do not like if user answers, then homing tree can then inquire article J, by that analogy.
The process that homing tree builds is actually the process of the grouping of hiving off to customer group, supposes that known users has n user to the evaluating data of article is concentrated, according to the marking of user to article A, is divided into the candidate child node of three groups of users as decision tree.By analyzing the similarity (variance) that in every group group, user evaluates, assessment article A is as the quality (often group group inner evaluation is more similar better) of grouping problem, by that analogy, for each child node, need to carry out iteration grouping for the customer group in current group, until homing tree level reaches certain threshold values.Carry out according to the node of homing tree puing question to the common hobby that just can find the customer group similar to active user's behavior fast, thus the cold start-up problem in commending system can be solved.Fig. 2 shows problem guiding flow process, and as shown in Figure 2, its concrete problem guiding process is:
After commending system builds, carry out user grouping to each article according to evaluation score according to the outer evaluating data of user's score data in station and station, in then calculating group, user marks variance, and then draws the poor sum of many prescriptions; Choose minimum variance and article as current decision tree node, and according to scoring build its child node.In the process that decision tree builds, judge whether the decision tree degree of depth exceedes pre-set threshold value, if the degree of depth of constructed decision tree exceedes pre-set threshold value, then complete the structure of decision tree; Otherwise further each child node data set is repeated to the processing procedure of aforementioned " grouping ~ build child node ", until complete the structure of decision tree.Build complete decision tree, just can inquire that user is to the evaluation of article according to the decision tree of this structure afterwards in user's Website login/application (step S202).
According to the description of above-mentioned flow process, mainly there are lower three problems in current problem guiding method:
1. bootmode is dumb.Such bootstrap technique is all often be bundled in the register flow path of user, user need according to flow process, each problem is answered, user on stream without any dominant right, selection cannot be carried out for a certain class problem and answer.
2. problem lacks relevance.On problem is selected, have employed overall article, talk about most owing to pursuing excessive information amount, lack relevance between problem, this mode causes similar algorithm can only be used on initial user, cannot continue to carry out alternately with user.Such as, if user only wants in store to buy suitable clothes, such algorithm the problem about clothes are relevant of only puing question to cannot carry out classification problem guiding.
3. sparse matrix support is low.All to carry out hiving off grouping (such as " like ", " not liking " and " not knowing " three classifications) based on the classification for the marking of article on problem is selected, all extremely sparse in most of application scenarios because user evaluates rating matrix, the article that user evaluates only account for the very fraction of total article number, so along with group's refining data minimizing, " do not know " that the number of users of classification can considerably beyond other categorical measures, cause the quantity of information of all optional article all less thus do not have suitable problem to put question to, algorithm premature termination.
In order to solve the above problem, need a kind of method that effective problem guiding is provided, thus solve user's cold start-up problem and improve Consumer's Experience.
Summary of the invention
In view of the above problems, the object of this invention is to provide a kind of personalized recommendation method based on problem guiding and system, to solve user's cold start-up problem, improve Consumer's Experience.
According to an aspect of the present invention, provide a kind of personalized recommendation method based on problem guiding, comprise two steps;
First step: obtain semantic topic or behavior theme and build multiple problem guiding tree; Wherein,
Problem according to user's input obtains semantic topic;
Browsing content according to user obtains behavior theme;
Data filtering is carried out to user behavior data, user's score data and article metadata, obtain multiple article themes of user, article and fancy grade ternary relation, and based on genetic algorithm build with the multiple article themes obtained one to one problem guiding set;
Second step: by theme coupling problem identificatioin homing tree; Wherein,
Hobby theme is obtained according to semantic topic or behavior theme;
Set with problem guiding by hobby theme and mate, determine that the problem guiding matched with hobby theme is set, the problem guiding tree according to determining carries out problem output to user, obtains user preference data thus carries out personalized recommendation.
In addition, preferred scheme is, obtains in the process of semantic topic in the problem inputted according to user,
The problem that user inputs is carried out Chinese word segmentation and semantic analysis, obtains semantic topic; Wherein,
In the process of Chinese word segmentation, utilize Chinese Word Automatic Segmentation that the paragraph sentence in natural language description text is split as word; Chinese Word Automatic Segmentation comprises maximum matching algorithm, the longest equal word algorithm and minimum variation algorithm;
In the process of semantic analysis, each word in the paragraph sentence of counting user input and the similarity degree of article theme;
In the process obtaining semantic topic, choose the semantic topic that maximum similarity themes as this statement.
In addition, preferred scheme is, is obtaining in the process of behavior theme according to the browsing content of user,
Obtain user behavior according to the browsing content of user, obtain user to the fancy grade of article theme according to user behavior, using maximum for user preferences degree and the article theme exceeding pre-set threshold value as behavior theme.
In addition, preferred scheme is, according to the selection of user to article each in the list containing multiple article, obtains the article that user likes most.
In addition, preferred scheme is, is building with the multiple article themes process that problem guiding is set one to one obtained based on genetic algorithm,
Population Selecting operation, crossing operation, mutation operator Generating Problems homing tree are carried out to the user's evaluating data collection corresponding to each article theme obtained; Wherein,
In the process of population Selecting operation, from colony, select winning individuality, eliminate worst individual; Wherein, the N number of article collection of stochastic generation is adopted to carry out crossing operation in the initial stage of population Selecting operation, after completing crossing operation and described mutation operator at every turn, from the N number of article collection of candidate item collection random selecting, and calculate the average fitness of the article collection population in population Selecting operation, wherein, N is greater than 1;
In the process of crossing operation, any two article collection of each random selecting, and carry out exchanging the new article collection of generation one to any two article, and meet newly-generated article and concentrate the article not have repetition;
In the process of mutation operator, random selecting replacing is carried out to any article that any article generated after crossing operation are concentrated, and the article met after variation concentrate the article not having to repeat, and obtain the average fitness of the article collection population after variation.
In addition, preferred scheme is, the average fitness of the article collection population in the average fitness of the article collection population generated according to crossing operation and described mutation operator, population Selecting operation and threshold values, carry out children User group allocation according to current item theme, Generating Problems homing tree.
According to another aspect of the present invention, a kind of personalized recommendation system based on problem guiding is provided, comprises:
Semantic topic acquiring unit, obtains semantic topic for the problem inputted according to user;
Behavior theme acquiring unit, obtains behavior theme for the browsing content according to user;
Theme acquiring unit, for carrying out data filtering to user behavior data, user's score data and article metadata, and obtains multiple article themes of user, article and fancy grade ternary relation,
Problem guiding tree generation unit, for build based on genetic algorithm with multiple article themes of obtaining one to one problem guiding set;
Hobby theme acquiring unit, for according to semantic topic or behavior theme, obtains hobby theme;
Theme matching unit, mates for being set with problem guiding by hobby theme, determines that the problem guiding matched with hobby theme is set;
Problem output unit, for carrying out problem output according to the problem guiding tree determined to user, obtaining user preference data thus carrying out personalized recommendation.
In addition, preferred scheme is, according to the selection of user to article each in the list containing multiple article, obtains the article that user likes most.
In addition, preferred scheme is, problem guiding tree generation unit comprises:
Population Selecting operation module, for selecting winning individuality from colony, eliminates worst individual; Wherein, the N number of article collection of stochastic generation is adopted to carry out crossing operation in the initial stage of population Selecting operation, after completing crossing operation and mutation operator at every turn, will from the N number of article collection of candidate item collection random selecting, and calculate the average fitness of the article collection population in population Selecting operation, wherein, N is greater than 1;
Crossing operation module, for any two article collection of random selecting, and carries out exchanging the new article collection of generation one to any two article in inside, and meets newly-generated article and concentrate the article not have repetition;
Mutation operator module, carries out random selecting replacing for any article concentrated any article generated after crossing operation, and the article met after variation concentrate the article not having to repeat, and obtains the average fitness of the article collection population after variation.
In addition, preferred scheme is, problem guiding tree generation unit also comprises:
Groups of users distributes module, for average fitness and the pre-set threshold value of the article collection population in the average fitness of the article collection population that generates according to crossing operation and mutation operator, population Selecting operation, children User group allocation is carried out, Generating Problems homing tree according to current item theme.
From technical scheme above, the personalized recommendation method based on problem guiding of the present invention and system, by obtaining semantic topic, solving the problem that classic method cannot mate guiding flexibly, can increase the dirigibility obtaining user preferences; By obtaining behavior theme, solution traditional problem boot flow fixedly can only be used in the problem in register flow path, can buy commodity by assisted user in real time; By adopting genetic algorithm Construct question to guide forest, problem of low quality of dividing into groups when solving Sparse, can improve the dirigibility of user grouping and the acquisition of Global Information amount.
In order to realize above-mentioned and relevant object, will describe in detail and the feature particularly pointed out in the claims after one or more aspect of the present invention comprises.Explanation below and accompanying drawing describe some illustrative aspects of the present invention in detail.But what these aspects indicated is only some modes that can use in the various modes of principle of the present invention.In addition, the present invention is intended to comprise all these aspects and their equivalent.
Accompanying drawing explanation
By reference to the content below in conjunction with the description of the drawings and claims, and understand more comprehensively along with to of the present invention, other object of the present invention and result will be understood and easy to understand more.In the accompanying drawings:
The two-layer problem guiding tree schematic diagram that Fig. 1 is;
Fig. 2 is problem guiding method flow schematic diagram;
Fig. 3 is the personalized recommendation method schematic flow sheet based on problem guiding according to the embodiment of the present invention;
Fig. 4 is the example flow schematic diagram of the method for the personalized recommendation based on problem guiding according to the embodiment of the present invention;
Fig. 5 is the personalized recommendation system structured flowchart based on problem guiding according to the embodiment of the present invention;
Fig. 6 is the structured flowchart of the semantic topic acquiring unit according to the embodiment of the present invention;
Fig. 7 is the structured flowchart of the behavior theme acquiring unit according to the embodiment of the present invention;
Fig. 8 is the problem guiding tree generation module structured flowchart according to the embodiment of the present invention.
Label identical in all of the figs indicates similar or corresponding feature or function.
Embodiment
In the following description, for purposes of illustration, in order to provide the complete understanding to one or more embodiment, many details have been set forth.But, clearly, also these embodiments can be realized when there is no these details.
From aforesaid prior art, problem guiding is only considered from the angle of maximum fault information, on Consumer's Experience without any optimization, user can lose patience premature termination question and answer.Simultaneously in the selection of problem, fixing scoring threshold values is adopted to classify, although be effectively controlled on computation complexity, because article Evaluations matrix has serious loose line, the consistance in group can be caused very low based on such user grouping, and affect follow-up recommendation results.
For these reasons, the present invention proposes to adopt and obtain semantic topic, obtain behavior theme and improve the openness of the dirigibility of bootmode and relevance and solving matrix based on genetic algorithm Construct question homing tree three aspects.
Below with reference to accompanying drawing, specific embodiments of the invention are described in detail.
Fig. 3 shows the personalized recommendation method flow process based on problem guiding according to the embodiment of the present invention.
As shown in Figure 3, the personalized recommendation method based on problem guiding provided by the invention mainly comprises two steps.
First step: S310: obtain semantic topic or behavior theme and build multiple problem guiding tree.
Particularly, according to the problem of user's input, obtain semantic topic; According to the browsing content of user, obtain behavior theme; Data filtering is carried out to user behavior data, user's score data and article metadata, and obtain multiple article themes of user, article and fancy grade ternary relation, based on genetic algorithm build with the multiple article themes obtained one to one problem guiding set.
Wherein, user, between article and fancy grade ternary, relation constitutes multiple article theme, such as, some user preferences action movies, some user preferences horror films, some user preferences comedies etc., these action movies, horror film, comedy are all article themes, and the slice, thin piece of different types constitutes multiple article theme.
Second step: S320: mated by theme, problem identificatioin homing tree.
Particularly, hobby theme is obtained according to semantic topic or behavior theme, hobby theme is set with problem guiding and mates, determine that the problem guiding matched with hobby theme is set, problem guiding tree according to determining carries out problem output to user, obtains user preference data thus carries out personalized recommendation.
Wherein, in step S310, the problem that user inputs is carried out Chinese word segmentation and semantic analysis, obtain semantic topic; Browsing content according to user obtains user behavior, user is obtained to the fancy grade of article theme according to user behavior, and judge whether its fancy grade exceedes pre-set threshold value, choose that user preferences degree is maximum and the article exceeding pre-set threshold value theme as behavior theme; Population Selecting operation, crossing operation, mutation operator Generating Problems homing tree are carried out to each article theme obtained.
In core of the present invention for obtaining semantic topic and behavior theme and building multiple problem guiding tree, how detailed description obtained semantic topic below, obtain behavior theme and build multiple problem guiding tree.
The first, semantic topic is obtained
The problem that user inputs is carried out Chinese word segmentation and semantic analysis, obtains semantic topic.
Wherein, Chinese word segmentation refers to and utilizes Chinese Word Automatic Segmentation that the paragraph sentence in natural language description text is split as word.Conventional Chinese Word Automatic Segmentation comprises maximum matching algorithm, the longest equal word algorithm and minimum variation algorithm.
Semantic analysis refers to the similarity degree of each word in the paragraph sentence that statistical summaries user inputs and article theme.If analyze paragraph by Chinese word segmentation in the paragraph of user's input to comprise n word, establish the article theme of pre-prefinishing (such as action movie, horror film etc.) to have m simultaneously, be respectively [T
1, T
2... T
m], set the correlativity of word i and article theme j as P in addition
ij,the similarity of word and article theme can obtain at the middle co-occurrence probabilities of language material training data (such as, the metadata story of a play or opera descriptor of film, film review text data of user etc.) with article theme by calculating, and its formula is expressed as:
Wherein, C
irepresent the number of times that word i occurs in language material; C
jrepresent the number of times that article theme j occurs in language material; C
ijfor the number of times that word i and article theme occur jointly.
The similarity degree of final paragraph statement s and article theme j is:
Further, choosing maximum similarity theme t is the semantic topic of this statement, and its formula is expressed as: t=argmax
jp
sj
Therefore, in the process obtaining semantic topic, by using pattern and the semantic analysis mode of dialogue chat, realm information can be extracted from user session content, thus recommendation guiding can be carried out for specific field, solve the defect that classic method cannot mate booting problem flexibly, increase the dirigibility obtaining user preferences.
The second, behavior theme is obtained
According to the browsing content of user, obtain user behavior, and whether exceed threshold values, finally using maximum for user preferences degree and fancy grade exceedes the theme of pre-set threshold value as behavior theme according to the user behavior statistics hobby theme obtained.
Wherein, obtain user behavior and refer to that collection storage user browses web sites, the operation behavior of Mobile solution, the operations such as such as, broadcasting in video website, download, comment, collection.
Whether statistics hobby theme exceedes threshold values refers to and calculates user to the fancy grade of article theme by the operation behavior of user, and judges whether fancy grade exceedes pre-set threshold value.
The calculating of fancy grade can carry out matching primitives by business rule, such as: play video and add 1 point, download and add 1.5 points, collection adds 2 and grades.A user can be added up to the fancy grade of article by gathering multiple vaild act, gathering a user to the multiple article fancy grades under an article theme simultaneously, and then obtain a user to the hobby value of an article theme.
In the process of the behavior of acquisition theme, the interested article theme of user is obtained by gathering user behavior data and adopting behavior to like digging technology, initiatively real-time interactive is carried out by the mode and user that play window, solution traditional problem boot flow fixedly can only be used in the problem in register flow path, can buy commodity by assisted user in real time.
Therefore, obtain semantic topic and can support to link up with user's direct dialogue text, acquisition behavior theme can be supported in Active participation boot flow in the process that user browses web sites, and these two kinds of modes can improve the dirigibility of guiding.
Three, Construct question homing tree
In the process of Construct question homing tree, first, data filtering is carried out to user behavior data, user's score data and article metadata, and obtain multiple article themes of user, article and fancy grade ternary relation; Then, the multiple article themes building based on genetic algorithm and obtain one to one problem guiding are set, and form the problem guiding forest of multiple article theme.
Wherein, it should be noted that, in the present invention, according to the selection of user to article each in the list containing multiple article, obtain the article that user likes most.
Traditional problem bootmode is that inquiry user is to the hobby suggestion (like, do not like and do not know) of single article, bootmode provided by the invention is inquire user to the fancy grade of multiple article (such as, like which portion in following 4 films), therefore, question guide mode of the present invention has larger level of coverage.
Particularly, user is arranged out to the fancy grade of article (such as: 1-5 divides according to information filterings such as user behavior data, user's score data and article metadata, 1 be divided into disagreeable, 5 be divided into like), and according to the relation of article and article theme, consumer articles score data to be classified according to article theme; The ternary relation matrix of the user under multiple article theme, article, fancy grade can be obtained.If user i is R to the evaluating data of article j
ij∈ R, separately sets the article under article theme t to integrate as S
t, the evaluating data set representations under article theme t is:
R(t)={R
ij∈R|j∈S
t}
To the data set under each article theme based on genetic algorithm Construct question homing tree respectively, thus the problem guiding forest of constructing multiple article theme solves the semantic association problem in problem guiding process.
That is, by adopting data filtering techniques, category filter is carried out to data, data set for different article theme builds different article problem of subject homing trees, and the semantic topic gone out by semantic analysis and behavior theme select hobby theme, and then select the problem guiding the most close with user preferences theme and set, thus the booting problem of relevance can be provided.
Building with the multiple article themes process that problem guiding is set one to one obtained based on genetic algorithm, population Selecting operation, crossing operation, mutation operator Generating Problems homing tree are carried out to each article theme obtained, thus constructs the problem guiding forest of multiple article theme.
Particularly, population Selecting operation refers to selects winning individuality in genetic algorithm from colony, eliminates worst individual.The structure of whole homing tree needs to utilize genetic algorithm to calculate each node, chooses best division article collection (such as: film " red sorghum ", " Farewell My Concubine ", " living ").
Further, the fitness of assessment division article collection can be obtained by the following method: establish article theme to comprise N number of article W
1,w
2,... W
n, then user can be divided into N+1 group (child node) G according to known users score data collection
1, G
2... G
n, G
n+1if the scoring of user i to article theme j is R
ij(setting numerical range as 1-5), grouping set 1 is defined as to n:
G
j={i∈U|R
ij≥3}
Wherein, U is the user complete or collected works of present node, G
jset representative be to article W
jthe user group liked; Any one user can belong to multiple grouping (child node) simultaneously, and grouping N+1 is defined as:
(child node) G in a group
ifitness analysis standard GR
ifor:
Wherein, μ
ikrepresent cohort G
iin all users to the average score of article K, if user's set of giving a mark to article K is for R (k), then:
So, any component is split to the comprehensive fitness degree evaluation criteria GR of article collection W
wfor the fitness sum of all groups, its formula is expressed as:
Wherein, it should be noted that, population Selecting operation initial stage adopts the N number of article collection of stochastic generation to carry out next step crossing operation, after completing Population breeding (that is: crossing operation and mutation operator after) at every turn, will from the N number of article collection of candidate item collection random selecting, the probability of choosing of each article collection is directly proportional to fitness and can be expressed as: P
w∝ GR
w, wherein, N is greater than 1; Further, its average fitness is obtained according to the fitness of the article collection population in population Selecting operation.
In the process of crossing operation, any two article collection of each random selecting, and carry out exchanging the new article collection of generation one to any two article in inside, and meet newly-generated article and concentrate the article not have repetition; This is that newly-generated offspring's article collection quantity is greater than the quantity (such as twice quantity) of parent article collection, for population Selecting operation module provides enough candidate's cluster in order to ensure the stability of population quantity in genetic iteration calculates.
In the process of mutation operator, random selecting replacing is carried out to any article that any article generated after crossing operation are concentrated, and the article met after variation concentrate the article not having to repeat, and obtain the average fitness of the article collection population after variation.
The average fitness of the article collection population in the average fitness of the article collection population then generated according to crossing operation and mutation operator, population Selecting operation and threshold values, carry out children User group allocation according to current topic, Generating Problems homing tree.
That is, whether the difference of the average fitness of article collection population newly-generated after completing crossing operation and mutation operator and the average fitness of former generation population (that is: the population after Selecting operation) is less than threshold values, or whether iterations exceedes pre-set threshold value.If so, then the article theme completing present node is chosen, and carries out the distribution of children User group according to current item theme, then completes whole the article collection setting all nodes according to the degree of depth preset and chooses, Generating Problems homing tree.If do not exceeded, then return population choose computing repeat calculate.
The detailed process of the above-mentioned problem guiding tree for generating, finally generates the problem guiding forest of multiple article theme.Because theme coupling needs the support of many problem guiding trees, the data set being exported multiple different article theme by data filtering sets up problem guiding tree respectively.In order to raise the efficiency, the problem guiding tree of different article theme can use distributive parallel computation framework computing simultaneously on multiple stage physical machine.
Based on genetic algorithm Construct question homing tree, it is the homing tree by topic distillation data construct many different themes, thus construct multi-threaded problem guiding forest, guide therefore, it is possible to mate by theme the problem guiding tree choosing different article theme, improve the correlativity of bootmode.The Question Classification of problem guiding tree adopts multiple different article instead of different scorings interval, thus can reduce the impact that Deta sparseness brings, and improves the consistance in different user group.
In step s 320, first, the hobby theme of user is determined according to the semantic topic obtained and behavior theme; Then the hobby theme of the user determined and problem guiding forest are carried out theme to mate, determine that the problem guiding matched with hobby theme is set, the problem guiding tree according to determining carries out problem output to user.
Particularly, choose according to the semantic topic obtained or behavior theme the problem guiding tree matched and carry out interaction with user, that is, put question to user according to the article collection in the problem guiding tree node selected.Such as: you like which article following? option one: article A, option 2: article B, option 3: article C, option 4: do not like or do not know
Genetic algorithm Construct question homing tree of the present invention, choose the different article of scoring interval (such as: scoring is greater than 3 representatives and likes scoring interval) as division option (that is: liking the user of identical items to be divided into a group), instead of the difference of article is marked, and alternatively (article are according to liking in interval, dislike and do not know to be divided into three groups of users), this mode causes when can solve Sparse dividing into groups problem of low quality (namely, solve most of user without scoring and be divided into one group, cause the problem that the customer group volume deviation of different group is very large), improve the dirigibility of user grouping and the acquisition of Global Information amount.
In order to further describe the method for the personalized recommendation based on problem guiding, Fig. 4 shows the example flow of the method for the personalized recommendation based on problem guiding according to the embodiment of the present invention.
As shown in Figure 4, the example flow of the method for the personalized recommendation based on problem guiding provided by the invention comprises:
S401: start;
S402: user's Website login application;
Perform step S403-S406, step S407-S410 respectively;
S403: user directly inputs problem;
S404: Chinese word segmentation;
S405: semantic analysis;
S406: obtain semantic topic.
S407: user directly inputs problem;
S408: recording user behavior;
S409: whether statistics hobby theme exceedes threshold values; If so, perform step S410, if not, perform step S408;
S410: obtain behavior theme.And,
S411: user's score data;
S412: user behavior data;
S413: article metadata;
S414: carry out data conversion according to theme;
That is, data filtering is carried out to user's score data, user behavior data and article metadata, get the ternary relation of multiple article theme.
Then, carry out according to different themes the problem guiding tree that genetic algorithm builds different themes, form problem guiding forest, that is: perform step S415-S419,
S415: population Selecting operation;
S416: crossing operation;
S417: mutation operator;
S418: whether optimal result exceeds threshold values; If so, perform step S419, if not, perform step S415;
S419: Generating Problems homing tree;
Tree-like being a problem of multiple article problem guiding guides forest.
S420: theme mates;
Namely, the hobby theme of user is determined according to the semantic topic obtained or behavior theme; Then the hobby theme of the user determined and problem guiding forest are carried out theme to mate, select and like the problem guiding that theme matches and set.
S421: export problem;
Namely, put question to user according to the article collection in the problem guiding tree node selected.
In the embodiment shown in fig. 4, can support to link up with user's direct dialogue text by semantic topic extraction assembly, can be supported in user by behavior subject distillation to browse web sites Active participation boot flow in process, these two kinds of modes can improve the dirigibility of guiding; Based on genetic algorithm Construct question homing tree, optimize the mode of conventional construction problem guiding tree, with multiple article as division item, the number that candidate divides item can increase along with number of articles exponentially, adopt traditional traversal contrast cannot choose optimum division item, that is: by Selecting operation, crossing operation and mutation operator, solve multiple problems that multiple article are chosen, guarantee that the article option information amount chosen is maximum.
Corresponding with said method, the present invention also provides a kind of personalized recommendation system based on problem guiding, and Fig. 5 shows the personalized recommendation system logical organization based on problem guiding according to the embodiment of the present invention.
As shown in Figure 5, the personalized recommendation system 500 based on problem guiding provided by the invention comprises: semantic topic acquiring unit 510, behavior theme acquiring unit 520, theme acquiring unit 530, problem guiding tree generation unit 540, hobby theme acquiring unit 550, theme matching unit 560 and problem output unit 570.
Wherein, semantic topic acquiring unit 510 obtains semantic topic for the problem inputted according to user.
Behavior theme acquiring unit 520 obtains behavior theme for the browsing content according to user.
Theme acquiring unit 530 for carrying out data filtering to user behavior data, user's score data and article metadata, and obtains multiple article themes of user, article and fancy grade ternary relation.
Problem guiding tree generation unit 540 for build based on genetic algorithm with multiple article themes of obtaining one to one problem guiding set.
Hobby theme acquiring unit 550, for according to semantic topic or behavior theme, obtains hobby theme.
Theme matching unit 560 mates for being set with problem guiding by hobby theme, determines that the problem guiding matched with hobby theme is set.
Problem output unit 570, for carrying out problem output according to the problem guiding tree determined to user, obtains user preference data thus carries out personalized recommendation.
Fig. 6 shows the structure of the semantic topic acquiring unit according to the embodiment of the present invention, and as shown in Figure 6, semantic topic acquiring unit 510 comprises: Chinese word segmentation module 511, semantic module 512 and semantic topic acquisition module 513.
Wherein, the paragraph sentence in natural language description text is split as word for utilizing Chinese Word Automatic Segmentation by Chinese word segmentation module 511; Chinese Word Automatic Segmentation comprises maximum matching algorithm, the longest equal word algorithm and minimum variation algorithm;
Each word in the paragraph sentence that semantic module 512 inputs for counting user and the similarity degree of article theme;
Semantic topic acquisition module 513 themes as the semantic topic of this statement for choosing maximum similarity.
Fig. 7 shows the structure of the behavior theme acquiring unit according to the embodiment of the present invention, and as shown in Figure 7, behavior theme acquiring unit 520 comprises: user behavior acquisition module 521, fancy grade acquisition module 522 and behavior theme acquisition module 523.
Wherein, user behavior acquisition module 521 obtains user behavior for the browsing content according to user;
Fancy grade acquisition module 522 is for obtaining user to the fancy grade of article theme according to user behavior;
Behavior theme acquisition module 523 is for choosing the maximum and article theme exceeding pre-set threshold value of user preferences degree as behavior theme.
In addition, theme acquiring unit 530, according to the selection of user to article each in the list containing multiple article, obtains the article that user likes most.
In addition, Fig. 8 shows the problem guiding tree generation module structure according to the embodiment of the present invention, as shown in Figure 8, problem guiding tree generation unit 540 comprises: population Selecting operation module 541, crossing operation module 542, mutation operator module 543 and groups of users distribute module 544.
Wherein, population Selecting operation module 541, for selecting winning individuality from colony, eliminates worst individual; Wherein, the N number of article collection of stochastic generation is adopted to carry out crossing operation in the initial stage of population Selecting operation, after completing crossing operation and mutation operator at every turn, from the N number of article collection of candidate item collection random selecting, and calculate the average fitness of the article collection population in population Selecting operation, wherein, N is greater than 1.
Crossing operation module 542 for any two article collection of random selecting, and carries out exchanging the new article collection of generation one to any two article, and meets newly-generated article and concentrate the article not have repetition.
Mutation operator module 543 carries out random selecting replacing for any article concentrated any article generated after crossing operation, and the article met after variation concentrate the article not having to repeat, and obtains the average fitness of the article collection population after variation.
Groups of users distributes the average fitness of module 544 for the article collection population in the average fitness of the article collection population that generates according to crossing operation and mutation operator, population Selecting operation and pre-set threshold value, children User group allocation is carried out, Generating Problems homing tree according to current item theme.
The more specifically reciprocal process of above-mentioned each module or unit, see the description in method flow, can repeat no more herein.
Can be found out by above-mentioned embodiment, the personalized recommendation method based on problem guiding that the present invention proposes and system, by obtaining semantic topic, can solve the problem that classic method cannot mate guiding flexibly, increases the dirigibility obtaining user preferences; By obtaining behavior theme, can solve traditional problem boot flow and fixedly can only be used in problem in register flow path, real-time assisted user buys commodity; By adopting genetic algorithm Construct question to guide forest, when can solve Sparse, causing problem of low quality of dividing into groups, improving the dirigibility of user grouping and the acquisition of Global Information amount.
The personalized recommendation method based on problem guiding and system that propose according to the present invention is described in an illustrative manner above with reference to accompanying drawing.But, it will be appreciated by those skilled in the art that the personalized recommendation method based on problem guiding and system that the invention described above is proposed, various improvement can also be made on the basis not departing from content of the present invention.Therefore, protection scope of the present invention should be determined by the content of appending claims.
Claims (10)
1., based on a personalized recommendation method for problem guiding, comprise two steps;
First step: obtain semantic topic or behavior theme and build multiple problem guiding tree; Wherein,
Problem according to user's input obtains semantic topic,
Browsing content according to user obtains behavior theme;
Data filtering is carried out to user behavior data, user's score data and article metadata, obtain multiple article themes of user, article and fancy grade ternary relation, and based on genetic algorithm build with the multiple article themes obtained one to one problem guiding set;
Second step: by theme coupling problem identificatioin homing tree; Wherein,
Hobby theme is obtained according to institute's semantic topic or described behavior theme;
Described hobby theme is set with described problem guiding and mates, determine that the problem guiding matched with described hobby theme is set, carry out problem output according to determined problem guiding tree to user, obtain user preference data thus carry out personalized recommendation.
2. as claimed in claim 1 based on the personalized recommendation method of problem guiding, wherein, obtain in the process of semantic topic in the problem inputted according to user,
The problem that user inputs is carried out Chinese word segmentation and semantic analysis, obtains semantic topic; Wherein,
In the process of described Chinese word segmentation, utilize Chinese Word Automatic Segmentation that the paragraph sentence in natural language description text is split as word; Described Chinese Word Automatic Segmentation comprises maximum matching algorithm, the longest equal word algorithm and minimum variation algorithm;
In the process of described semantic analysis, each word in the paragraph sentence of the problem of counting user input and the similarity degree of article theme;
In the process of described acquisition semantic topic, choose the semantic topic that maximum similarity themes as this statement.
3. as claimed in claim 1 based on the personalized recommendation method of problem guiding, wherein, obtaining in the process of behavior theme according to the browsing content of user,
Browsing content according to user obtains user behavior;
User is obtained to the fancy grade of article theme according to described user behavior, and using maximum for user preferences degree and the article theme exceeding pre-set threshold value as behavior theme.
4. as claimed in claim 1 based on the personalized recommendation method of problem guiding, wherein,
According to the selection of user to article each in the list containing multiple article, obtain the article that user likes most.
5. as claimed in claim 4 based on the personalized recommendation method of problem guiding, wherein, building with the multiple article themes process that problem guiding is set one to one obtained based on genetic algorithm,
Population Selecting operation, crossing operation, mutation operator Generating Problems homing tree are carried out to the user's evaluating data collection corresponding to each article theme obtained; Wherein,
In the process of described population Selecting operation, from colony, select winning individuality, eliminate worst individual; Wherein, stochastic generation N number of article collection is adopted to carry out described crossing operation in the initial stage of described population Selecting operation, after completing described crossing operation and described mutation operator at every turn, from the N number of article collection of candidate item collection random selecting, and calculate the average fitness of the article collection population in described population Selecting operation, wherein, N is greater than 1;
In the process of described crossing operation, any two article collection of each random selecting, and carry out exchanging the new article collection of generation one to any two article, and meet newly-generated article and concentrate the article not have repetition;
In the process of described mutation operator, random selecting replacing is carried out to any article that any article generated after described crossing operation are concentrated, and the article met after variation concentrate the article not having to repeat, and obtain the average fitness of the article collection population after variation.
6. as claimed in claim 5 based on the personalized recommendation method of problem guiding, wherein,
The average fitness of the article collection population in the average fitness of the article collection population generated according to described crossing operation and described mutation operator, described population Selecting operation and threshold values, children User group allocation is carried out, Generating Problems homing tree according to current item theme.
7., based on a personalized recommendation system for problem guiding, comprising:
Semantic topic acquiring unit, obtains semantic topic for the problem inputted according to user;
Behavior theme acquiring unit, obtains behavior theme for the browsing content according to user;
Theme acquiring unit, for carrying out data filtering to user behavior data, user's score data and article metadata, and obtains multiple article themes of user, article and fancy grade ternary relation,
Problem guiding tree generation unit, for build based on genetic algorithm with multiple article themes of obtaining one to one problem guiding set;
Hobby theme acquiring unit, for according to institute's semantic topic or described behavior theme, obtains hobby theme;
Theme matching unit, for being mated with described problem guiding forest by described hobby theme, determines that the problem guiding matched with described hobby theme is set;
Problem output unit, for carrying out problem output according to the problem guiding tree determined to user, obtaining user preference data thus carrying out personalized recommendation.
8. as claimed in claim 7 based on the personalized recommendation system of problem guiding, wherein,
Described theme acquiring unit, according to the selection of user to article each in the list containing multiple article, obtains the article that user likes most.
9. as claimed in claim 8 based on the personalized recommendation system of problem guiding, wherein, described problem guiding tree generation unit comprises:
Population Selecting operation module, for selecting winning individuality from colony, eliminates worst individual; Wherein, stochastic generation N number of article collection is adopted to carry out described crossing operation in the initial stage of described population Selecting operation, after completing described crossing operation and described mutation operator at every turn, from the N number of article collection of candidate item collection random selecting, and calculate the average fitness of the article collection population in described population Selecting operation, wherein, wherein, N is greater than 1;
Crossing operation module, for any two article collection of random selecting, and carries out exchanging the new article collection of generation one to any two article, and meets newly-generated article and concentrate the article not have repetition;
Mutation operator module, carries out random selecting replacing for any article concentrated any article generated after described crossing operation, and the article met after variation concentrate the article not having to repeat, and obtains the average fitness of the article collection population after variation.
10. as claimed in claim 9 based on the personalized recommendation system of problem guiding, wherein, described problem guiding tree generation unit also comprises:
Groups of users distributes module, for average fitness and the pre-set threshold value of the article collection population in the average fitness of the article collection population that generates according to described crossing operation and described mutation operator, described population Selecting operation, children User group allocation is carried out, Generating Problems homing tree according to current item theme.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410856318.XA CN104572982B (en) | 2014-12-31 | 2014-12-31 | Personalized recommendation method and system based on problem guiding |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410856318.XA CN104572982B (en) | 2014-12-31 | 2014-12-31 | Personalized recommendation method and system based on problem guiding |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104572982A true CN104572982A (en) | 2015-04-29 |
CN104572982B CN104572982B (en) | 2017-10-31 |
Family
ID=53089044
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410856318.XA Active CN104572982B (en) | 2014-12-31 | 2014-12-31 | Personalized recommendation method and system based on problem guiding |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104572982B (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106126541A (en) * | 2016-06-14 | 2016-11-16 | 华南师范大学 | Personalized solution based on big data knowledge storehouse recommends method and system |
CN106649842A (en) * | 2016-12-30 | 2017-05-10 | 上海博泰悦臻电子设备制造有限公司 | Cross recommendation method based on fusion data, system and vehicle machine |
CN106776701A (en) * | 2016-11-14 | 2017-05-31 | 东软集团股份有限公司 | The problem recommended for article determines method and apparatus |
CN107992459A (en) * | 2017-10-12 | 2018-05-04 | 如是科技(大连)有限公司 | Resume generation method and device |
CN107992610A (en) * | 2017-12-15 | 2018-05-04 | 泰康保险集团股份有限公司 | A kind of automatic question-answering method, device, equipment and storage medium |
CN108228910A (en) * | 2018-02-09 | 2018-06-29 | 艾凯克斯(嘉兴)信息科技有限公司 | It is a kind of that Recognition with Recurrent Neural Network is applied to the method on association select permeability |
CN110399471A (en) * | 2018-04-25 | 2019-11-01 | 北京快乐智慧科技有限责任公司 | A kind of guiding situational dialogues method and system |
CN110930969A (en) * | 2019-10-14 | 2020-03-27 | 科大讯飞股份有限公司 | Background music determination method and related equipment |
CN112288456A (en) * | 2020-01-22 | 2021-01-29 | 北京沃东天骏信息技术有限公司 | Article selection method and apparatus |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006004441A (en) * | 2005-06-24 | 2006-01-05 | Yokogawa Electric Corp | Information service system |
CN101296154A (en) * | 2007-04-28 | 2008-10-29 | 上海盛大网络发展有限公司 | Automatic answering system |
US20090012778A1 (en) * | 2007-07-05 | 2009-01-08 | Nec (China) Co., Ltd. | Apparatus and method for expanding natural language query requirement |
CN101354714A (en) * | 2008-09-09 | 2009-01-28 | 浙江大学 | Method for recommending problem based on probability latent semantic analysis |
CN101409690A (en) * | 2008-11-26 | 2009-04-15 | 北京学之途网络科技有限公司 | Method and system for obtaining internet user behaviors |
US20120310778A1 (en) * | 2011-06-03 | 2012-12-06 | Uc Group Limited | Systems and methods for clearing and settling transaction activity |
CN103455638A (en) * | 2013-09-26 | 2013-12-18 | 中国科学院自动化研究所 | Behavior knowledge extracting method and device combining reasoning and semi-automatic learning |
CN103995858A (en) * | 2014-05-15 | 2014-08-20 | 北京航空航天大学 | Individualized knowledge active pushing method based on task decomposition |
JP2014526747A (en) * | 2011-09-13 | 2014-10-06 | ▲騰▼▲訊▼科技(深▲セン▼)有限公司 | Data alignment method and apparatus |
-
2014
- 2014-12-31 CN CN201410856318.XA patent/CN104572982B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006004441A (en) * | 2005-06-24 | 2006-01-05 | Yokogawa Electric Corp | Information service system |
CN101296154A (en) * | 2007-04-28 | 2008-10-29 | 上海盛大网络发展有限公司 | Automatic answering system |
US20090012778A1 (en) * | 2007-07-05 | 2009-01-08 | Nec (China) Co., Ltd. | Apparatus and method for expanding natural language query requirement |
CN101354714A (en) * | 2008-09-09 | 2009-01-28 | 浙江大学 | Method for recommending problem based on probability latent semantic analysis |
CN101409690A (en) * | 2008-11-26 | 2009-04-15 | 北京学之途网络科技有限公司 | Method and system for obtaining internet user behaviors |
US20120310778A1 (en) * | 2011-06-03 | 2012-12-06 | Uc Group Limited | Systems and methods for clearing and settling transaction activity |
JP2014526747A (en) * | 2011-09-13 | 2014-10-06 | ▲騰▼▲訊▼科技(深▲セン▼)有限公司 | Data alignment method and apparatus |
CN103455638A (en) * | 2013-09-26 | 2013-12-18 | 中国科学院自动化研究所 | Behavior knowledge extracting method and device combining reasoning and semi-automatic learning |
CN103995858A (en) * | 2014-05-15 | 2014-08-20 | 北京航空航天大学 | Individualized knowledge active pushing method based on task decomposition |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106126541A (en) * | 2016-06-14 | 2016-11-16 | 华南师范大学 | Personalized solution based on big data knowledge storehouse recommends method and system |
CN106126541B (en) * | 2016-06-14 | 2019-07-19 | 华南师范大学 | Personalized solution recommended method and system based on big data knowledge base |
CN106776701A (en) * | 2016-11-14 | 2017-05-31 | 东软集团股份有限公司 | The problem recommended for article determines method and apparatus |
CN106776701B (en) * | 2016-11-14 | 2020-04-07 | 东软集团股份有限公司 | Problem determination method and device for item recommendation |
CN106649842A (en) * | 2016-12-30 | 2017-05-10 | 上海博泰悦臻电子设备制造有限公司 | Cross recommendation method based on fusion data, system and vehicle machine |
CN107992459A (en) * | 2017-10-12 | 2018-05-04 | 如是科技(大连)有限公司 | Resume generation method and device |
CN107992610A (en) * | 2017-12-15 | 2018-05-04 | 泰康保险集团股份有限公司 | A kind of automatic question-answering method, device, equipment and storage medium |
CN108228910A (en) * | 2018-02-09 | 2018-06-29 | 艾凯克斯(嘉兴)信息科技有限公司 | It is a kind of that Recognition with Recurrent Neural Network is applied to the method on association select permeability |
CN110399471A (en) * | 2018-04-25 | 2019-11-01 | 北京快乐智慧科技有限责任公司 | A kind of guiding situational dialogues method and system |
CN110930969A (en) * | 2019-10-14 | 2020-03-27 | 科大讯飞股份有限公司 | Background music determination method and related equipment |
CN110930969B (en) * | 2019-10-14 | 2024-02-13 | 科大讯飞股份有限公司 | Background music determining method and related equipment |
CN112288456A (en) * | 2020-01-22 | 2021-01-29 | 北京沃东天骏信息技术有限公司 | Article selection method and apparatus |
Also Published As
Publication number | Publication date |
---|---|
CN104572982B (en) | 2017-10-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104572982A (en) | Personalized recommendation method and system based on question guide | |
CN103209342B (en) | A kind of introduce video popularity and the collaborative filtered recommendation method of user's interests change | |
CN103338403B (en) | Individual character program commending method in radio data system and this system | |
US20070282684A1 (en) | System and Method for Determining Affinity Profiles for Research, Marketing, and Recommendation Systems | |
Khalid et al. | A literature review of implemented recommendation techniques used in massive open online courses | |
CN106951471A (en) | A kind of construction method of the label prediction of the development trend model based on SVM | |
CN107704996A (en) | A kind of Teacher Evaluation System based on sentiment analysis | |
CN105302880A (en) | Content correlation recommendation method and apparatus | |
Evripidou et al. | Quaestio-it. com: a social intelligent debating platform | |
CN109408726B (en) | Question answering person recommendation method in question and answer website | |
Vidal et al. | Application of social media for consumer research | |
CN105975609A (en) | Industrial design product intelligent recommendation method and system | |
CN110347893A (en) | A kind of individualized learning content recommendation system based on subspace clustering | |
CN103235777B (en) | A kind of internet public feelings acquisition methods of user demand guidance | |
Stewart et al. | Build it and they will come? Patron engagement via Twitter at historically black college and university libraries | |
Huang et al. | Information fusion oriented heterogeneous social network for friend recommendation via community detection | |
Al-Otaibi et al. | Finding influential users in social networking using sentiment analysis | |
Hajikhani et al. | Brand analysis in social network services: Results from content analysis in twitter regarding the us smartphone market | |
CN111583363A (en) | Visual automatic generation method and system for image-text news | |
KR102559449B1 (en) | Server and method for providing charge-free lerning curation service | |
CN115936939A (en) | Project recommendation method and system based on multiple preference degrees | |
Hodson et al. | Sharing sustainability stories: Case study of social media content curation for Canada research connections | |
Spyridou et al. | Sustainable online news projects: redefining production norms and practices | |
Rao et al. | Evaluating social responsible attitudes and opinions using sentiment analysis–an Indian sentiment | |
Sadeqi-Arani et al. | Content marketing scientific articles in the WOS: a bibliometric analysis |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |