CN106980646A - The method applied based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm - Google Patents
The method applied based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm Download PDFInfo
- Publication number
- CN106980646A CN106980646A CN201710106106.3A CN201710106106A CN106980646A CN 106980646 A CN106980646 A CN 106980646A CN 201710106106 A CN201710106106 A CN 201710106106A CN 106980646 A CN106980646 A CN 106980646A
- Authority
- CN
- China
- Prior art keywords
- popularity
- user
- project
- recommendation
- scoring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
The present invention relates to a kind of method applied based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm, it comprises the following steps：Step one, with the social psychology such as group psychology phenomenon and theory for foundation, influence of the research popularity to user, the relation between the change of analysis user interest and collective will, General Principle of the induction and conclusion popularity in recommendation；A kind of step 2, it is proposed that prediction scoring method of adjustment based on popularity.Based on the influence principle scored by popularity user, prediction scoring is adjusted according to project popularity and the variation relation of scoring in proposed algorithm, to improve the accuracy of proposed algorithm prediction；A kind of step 3, it is proposed that recommendation sort method based on popularity.This method fusion conventional recommendation standard sum term mesh popularity two indices carry out recommendation sequence, it is to avoid the defect such as interest predicated error or the impersonal theory of recommendation for only being produced respectively according to user's past historical record or project popularity.
Description
Technical field
Applied the present invention relates to a kind of based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm
Method.
Background technology
With the development and popularization of Internet technology, people become increasingly dependent on network and carry out various activities.But, it is current mutual
Working application and service are continuously increased, and abundant information resources bring certain difficulty to user's selection.In this context, it is individual
Propertyization recommendation method has become an important component of the Internet, applications.For proposed algorithm, it can predict exactly
User interest, recommends user, so that it is measurement one to improve user's clicking rate or purchase volume by most possible things interested
The basic standard of individual proposed algorithm effect.
Current Collaborative Filtering Recommendation Algorithm only according to user it is past browse historical behavior or score predict user
Environment or group behavior are not accounted in the interest in future, algorithm to individual consumer's scoring and the influence of housing choice behavior.And according to
The correlation theory of social psychology research, user interest is continually changing, while being also highly susceptible to external environment and group
The influence of body behavior and change.
The content of the invention
It is an object of the invention to for current Collaborative Filtering Recommendation Algorithm recommend it is not good in the case of, improve recommend it is accurate
Exactness and user satisfaction, a kind of analyzed based on popularity the influencing mechanism of user interest and its in recommendation calculation is provided for this
The method applied in method.This method is with the social psychology such as group psychology phenomenon and theory for foundation, and research popularity is to user
The influence of interest, the project popularity of drawing is increased over time, and first raise reduces again.For the higher project of popularity,
Diversity of values is smaller, and this is also mainly due to individual group psychology, and project is more popular, and scoring more reaches unanimity；It is more popular,
The viewpoint of user is more little affected by the influence of other users, is judged according to the hobby of oneself, and user's scoring will compare point
Dissipate.This method proposes a kind of recommendation optimized algorithm, it is contemplated that the factor of popularity, show that the prediction recommendation score of project is more accorded with
True scoring is closed, the accuracy of recommendation is improved.This method is improved to traditional recommendation sort algorithm, by setting weight
The factor, weighs the influence of conventional recommendation standard and popularity factor to user interest.
To reach above-mentioned purpose, idea of the invention is that：Relation first between analysis project popularity and user's scoring,
Based on the influence principle scored by popularity user, according to project popularity and the change of scoring in proposed algorithm
Relation pair prediction scoring is adjusted, to improve the accuracy of proposed algorithm prediction.Merge conventional recommendation standard sum term mesh popular
Spend two indices and carry out recommendation sequence, it is to avoid it is pre- only to go over the interest that historical record or project popularity produce respectively according to user
Survey the defects such as the impersonal theory of error or recommendation.Improve the clicking rate recommended and the degree of accuracy.
Conceived according to foregoing invention, the present invention is adopted the following technical scheme that：
A kind of method applied based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm, including
Following steps：
Step one, the General Principle of popularity and user interest is proposed：With social psychology phenomenon and theory for foundation, grind
Study carefully influence of the popularity to user interest, the relation between the change of analysis user interest and collective will, induction and conclusion popularity
General Principle in recommendation；
Step 2, proposes a kind of prediction scoring method of adjustment based on popularity：The influence scored with popularity user
Based on action principle, prediction scoring is adjusted according to project popularity and the variation relation of scoring in proposed algorithm,
To improve the accuracy of proposed algorithm prediction；
Step 3, proposes a kind of recommendation sort method based on popularity：This method merges conventional recommendation standard sum term mesh
Popularity two indices carry out recommendation sequence, it is to avoid only according to user go over that historical record or project popularity produce respectively it is emerging
The defect such as interesting predicated error or the impersonal theory of recommendation.
The scoring method of adjustment of the prediction based on popularity of the step 2, wherein, popularity refers to that a certain project is obtained
The user's degree of concern obtained, refers to the welcome degree of media file in a period of time of weighing, it reflects files in stream media at certain
The probability that one moment was asked by client, usual popularity is defined as in interval of time, and media file is by time of program request
Number, can fit the function H that popularity is changed over time.
The step 2 comprises the following steps：
Step 2 11：Interest similar users, the collaborative filtering based on user are excavated using the collaborative filtering of user
Recommendation is broadly divided into 3 stages：Data are represented；It was found that nearest-neighbors；Produce recommended project collection：Data represent to use user-item
Mesh rating matrix come represent obtain explicit data；It was found that nearest-neighbors are exactly the behavior of scoring and most like some of active user
User, finds the core that nearest-neighbors are Collaborative Filtering Recommendation Algorithms, and the accuracy of nearest-neighbors is directly connected to recommendation
The recommendation effect of algorithm is general in Collaborative Filtering Recommendation Algorithm to be judged whether by calculating the similitude between user closely
It is adjacent；The scoring that recommended project collection treats recommended project according to neighbor user is produced, targeted customer is tried to achieve by score calculation formula
The prediction marking of commending system is treated, then the higher project of selection scoring is recommended, general to use TOP-N proposed algorithms；
Recommended in collaborative filtering commonly using the mode of score in predicting, that is, the project for selecting N number of prediction scoring higher is pushed away
Recommend to targeted customer；Score in predicting formula is commonly defined as：
Wherein, Pi,tRepresent that predictions of the user i to project t is scored, n is the number of similar neighborhood, sim (i, j) is user i
With user j similarity, Rj,tIt is scorings of the user j to project t,For average scores of the user j to all items,For institute
There is average score of the user to project t that scored.
Step 2 12：Prediction scoring adjustment, the pre- test and appraisal for the collaborative filtering of adjustment that scored based on Popularity prediction
Point, prediction scoring formula is P '=P+T*f ((H-H ')/T), and wherein P represents neighbours' prediction scoring that project scores according to history,
P ' represents that the prediction of project is truly scored, and H represents project current popularity, and f represents the relation of popularity change and scoring change,
H ' represents the prediction popularity of project, and T represents the time interval apart from current time, the formula consider the popularity of project because
The influence of element, the variation tendency of stage popularity according to residing for project scores to adjust prediction during prediction；
Step 2 13：Compare the accuracy of two methods prediction scoring, relatively more traditional collaborative filtering is with being based on stream
The prediction scoring MAE of the collaborative filtering of row degree prediction scoring adjustment.
The recommendation sort method based on popularity of the step 3, including：
Hot Item Sets：Project popularity is more than the Item Sets of the average popularity of whole data set；
Cold Item Sets：Project popularity is less than the Item Sets of the average popularity of whole data set；
Recommend sequence：Recommend overall target function R=α * General Rank+ (1- α) * Popularity of sequence, its
Middle α is weight factor, and General Rank are conventional recommendation items selection standard, and Popularity is the prevalence of destination item
Degree；The formula is the fusion to two interest factors of influence, and for different proposed algorithms, the weight shared by each standard is perhaps
Different, α value needs correspondingly to be adjusted, and recommends efficiency to determine the α values of disparity items by comparing.
The step 3 comprises the following steps：
Step 3 11：Based on traditional collaborative filtering, a new recommendation order standard is designed, that is, merges item
The comprehensive standard of mesh prediction scoring and project popularity is ranked up to project to be measured, is set up and is recommended sequence overall target function；
Step 3 12：Experimental Comparison is carried out by adjusting weight factor α, to find to promote hot and cold Item Sets to recommend to tie
The optimal α of fruit.
The method applied based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm of the present invention,
Compared with prior art, with following breakthrough and remarkable advantage：
Universal phenomenon and rule that the inventive method changes according to user interest with popularity, are proposed based on the pre- of popularity
Test and appraisal point adjustment proposed algorithm and the recommendation sort method being combined based on conventional recommendation index with popularity factor, and based on association
The improvement of two aspects is realized with filter algorithm, the validity of innovatory algorithm by experimental verification.Present invention is primarily based on shadow
The popularity factor for ringing user interest is researched and analysed, and the variation tendency of user interest is predicted based on psychology.From
Under the influence of the factors such as many psychology, more popular project can more attract more user, although the project may score relatively low, because
This general proposed algorithm record (such as user scoring, browse project) only according to user's history and recommend be it is inaccurate,
For new user, the algorithm also can preferably solve the problems, such as the cold start-up of generally existing；In addition, user is to popular article
Scoring occurs correspondingly to change also with the variation tendency of popularity, and mainly project popularity initial stage is ground herein
Study carefully, prediction scoring is adjusted according to the change of popularity, predictablity rate is improved.
Brief description of the drawings
Fig. 1 is the flow chart of the inventive method.
Fig. 2 is that high scoring movie streams row degree and scoring change over time relation.
Fig. 3 is that middle scoring film popularity and scoring change over time relation.
Fig. 4 is that lower assessment point film popularity and scoring change over time relation.
Fig. 5 is different popularity films scoring situation.
Fig. 6 is that tradition CF is compared with based on popularity adjustment CF score in predicting accuracys.
Fig. 7 is the F value changes situation maps based on the different weight factors of hot-activity purpose.
Fig. 8 is the F value changes situation maps of the different weight factors based on cold project.
Embodiment
The preferred embodiments of the present invention are further described below in conjunction with accompanying drawing.
As shown in figure 1, in the present embodiment, it is of the invention based on popularity to the influencing mechanism analysis of user interest and its
The method applied in proposed algorithm, is tested mainly for Movie Lens data sets.Movie Lens data sets have collected
In 1000209 assessed values of 6040 users in 2000 to 3900 films, user's scoring is the integer from 1 to 5, and numerical value is got over
Height shows that user is bigger to the interest of the film.Score data mainly includes representing Customs Assigned Number User ID, represents film numbering
The time Timestamp tetra- that Movie ID, expression user score film scoring Rating, the user of film, i.e., it is as follows
Shown in table：
| User ID | Movie ID | Rating | Timestamp |
| 1~6040 | 1~3900 | 1~5 | 956703932~1046454590 |
Step one, the General Principle of popularity and user interest is proposed, with the social psychology such as group psychology phenomenon and reason
By for foundation, influence of the research popularity to user interest, the relation between the change of analysis user interest and collective will, and base
In influence of the experimental analysis popularity to user's scoring and behavior, General Principle of the induction and conclusion popularity in recommendation；
Step 2, proposes a kind of prediction scoring method of adjustment based on popularity.The influence scored with popularity user
Based on action principle, prediction scoring is adjusted according to project popularity and the variation relation of scoring in proposed algorithm,
To improve the accuracy of proposed algorithm prediction.Contrast experiment is carried out based on Movie Lens data sets, test result indicates that changing
The proposed algorithm entered is more accurate compared with conventional method；
Step 3, proposes a kind of recommendation sort method based on popularity.This method merges conventional recommendation standard sum term mesh
Popularity two indices carry out recommendation sequence, it is to avoid only according to user go over that historical record or project popularity produce respectively it is emerging
The defect such as interesting predicated error or the impersonal theory of recommendation.Test result indicates that improved proposed algorithm is effectively improved recommendation
Clicking rate and the degree of accuracy.
The step one includes：
Step 11, is tested mainly for Movie Lens data sets, and overall system is estimated using sample in mathematics
Analysis method is counted, the randomly drawing sample from data set calculates within each period the average popularity of sample and average respectively
Scoring, is divided into three groups by score data, i.e., high (Rating 4-5), in (Rating 3-4), lower assessment divide (Rating 0-3).
Step 12, to analysis of experimental results, reference picture 2,3,4,5, in three groups of scorings, film popularity is with the time
Variation tendency is basically identical, and film popularity is increased over time, and first raise reduces again.For the higher electricity of popularity
Shadow, diversity of values is smaller, and this is also mainly due to individual group psychology, and film is more popular, and scoring more reaches unanimity；Do not flow
OK, the viewpoint of user is more little affected by the influence of other users, is judged according to the hobby of oneself, and film scoring will compare
It is scattered.
The step 2 includes：
Step 2 11：Interest similar users, the collaborative filtering based on user are excavated using the collaborative filtering of user
Recommendation is broadly divided into 3 stages：Data are represented；It was found that nearest-neighbors；Produce recommended project collection.
Data represent the explicit data for representing to obtain using user-project rating matrix；It was found that nearest-neighbors are exactly to comment
Branch is some users most like with active user, finds the core that nearest-neighbors are Collaborative Filtering Recommendation Algorithms, most
The accuracy that neighbour occupies is directly connected to the recommendation effect of proposed algorithm.It is general in Collaborative Filtering Recommendation Algorithm to be used by calculating
Similitude between family judges whether neighbour.Calculating the method for similarity at present mainly has：Cosine similarity, amendment cosine phase
Like property and Pearson correlation coefficient isometry method；The scoring that recommended project collection treats recommended project according to neighbor user is produced,
The prediction marking that targeted customer treats commending system is tried to achieve by score calculation formula, then the higher project of selection scoring is carried out
Recommend, it is general to use TOP-N proposed algorithms.Recommended in collaborative filtering commonly using the mode of score in predicting, i.e.,
The higher project recommendation of the N number of prediction scoring of selection is to targeted customer.Score in predicting formula is commonly defined as：
Wherein, Pi,tRepresent that predictions of the user i to project t is scored, n is the number of similar neighborhood, sim (i, j) is user i
With user j similarity, Rj,tIt is scorings of the user j to project t,For average scores of the user j to all items,For institute
There is average score of the user to project t that scored.
Step 2 12：Prediction scoring adjustment, based on Popularity prediction score adjustment collaborative filtering (PCF) it is pre-
Test and appraisal point, prediction scoring formula P '=P+T*f ((H-H ')/T), P represents neighbours' prediction scoring that project scores according to history, P '
The prediction of expression project is truly scored, and f represents the relation of popularity change and scoring change, and H represents project current popularity, H '
The prediction popularity of expression project, T represents the time interval apart from current time.The formula considers the popularity factor of project
Influence, during prediction according to residing for project stage popularity variation tendency come adjust prediction scoring.Wherein, popularity：Refer to certain
User's degree of concern that one project is obtained, refers to the welcome degree of media file in a period of time of weighing, and it reflects stream matchmaker
The probability that body file is at a time asked by client, usual popularity is defined as in interval of time, media file
By the number of times of program request.The function that popularity is changed over time can be fitted by analyzing test data.
The popularity of high score film changes fitting formula:H=-13.05*ln (Δ T)+51.543；
In the popularity change fitting formula of point film be:H=-12.95*ln (Δ T)+51.762；
The popularity of low point of film changes fitting formula:H=-16.854*ln (Δ T)+80.534.
Step 2 13：Compare the accuracy of two methods prediction scoring, relatively more traditional collaborative filtering (GCF) and base
In the prediction scoring MAE of the collaborative filtering (PCF) of Popularity prediction scoring adjustment.With reference to Fig. 6, commented based on Popularity prediction
Divide the accuracy of the prediction scoring of the collaborative filtering (PCF) of adjustment higher than traditional collaborative filtering (GCF).
The step 3 includes：
Step 3 11：Based on traditional collaborative filtering, a new recommendation order standard is designed, that is, merges item
The comprehensive standard of mesh prediction scoring and project popularity is ranked up to project to be measured, is set up and is recommended sequence overall target function.
Step 3 12：Experimental Comparison is carried out by adjusting weight factor α, to find to promote hot and cold Item Sets to recommend to tie
The optimal α of fruit.
Wherein, hot Item Sets：Project popularity is more than the Item Sets of the average popularity of whole data set.
Cold Item Sets：Project popularity is less than the Item Sets of the average popularity of whole data set.
Recommend sequence overall target function：Recommend overall target function R=α * General Rank+ (1- α) * of sequence
Popularity, α are weight factor, and General Rank are conventional recommendation items selection standard, and Popularity is target item
Purpose popularity.The formula is the fusion to two interest factors of influence.For different proposed algorithms, shared by each standard
Weight may be different, and α value needs correspondingly to be adjusted, and recommends efficiency to determine the α values of disparity items by comparing.
Reference picture 7 and Fig. 8, the recommendation degree that the popularity factor is added when recommending are more than what exclusive use scoring was recommended
The degree of accuracy.As α=0, represent that fertilizer index is recommended using popularity value as standard completely, during α=1, expression recommendation refers to
Mark to predict that scoring is recommended as standard, is equal to traditional collaborative filtering completely.It is based entirely on pushing away for popularity
Recommend efficiency and be more than the recommendation efficiency for being based entirely on prediction scoring.Therefore, judge whether user is interested in this to score completely
And it is unilateral to produce further behavior.But, although the clicking rate for being based entirely on popularity is high, can not use the party
Method.Because this method is the hobby for pushing content to user, user not being considered of blindness, user processing information can be increased
Degree of difficulty.But, the method based on popularity can preferably handle the cold start-up problem that new user is brought.Therefore,
User recommend to be best strategy with reference to popularity and prediction two factors of scoring.In addition, for hot Item Sets,
With the increase of scoring weight α, downward trend is presented in F values (evaluation recommendation results), represents in stream of items row order section, project
Pouplarity is the main function that leading user selects film.In Fig. 7, for hot Item Sets, when α is 0.3~0.4, F values
Highest, preferably, i.e. the recommendation of pop project, project popularity accounts for 70%~60% or so weight, illustrated recommendation effect
Recommendation of the popular stage popularity to project has important effect.And for cold project, in Fig. 8, show the increasing with α
Plus, the fluctuation range of F values is than larger, when α is smaller, that is, represents that the weight that project popularity is accounted for is larger, recommendation effect is but very low,
Because for cold project, people would not be influenceed to select object of interest by group psychology again, but according to the interest of oneself
Go selection.When α is when between 0.4~0.6, recommendation effect is preferable.
Claims (5)
1. a kind of method applied based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm, its feature
It is, comprises the following steps：
Step one, the General Principle of popularity and user interest is proposed：With social psychology phenomenon and theory for foundation, research stream
Influence of the row degree to user interest, the relation between the change of analysis user interest and collective will, induction and conclusion popularity is being pushed away
General Principle in recommending；
Step 2, proposes a kind of prediction scoring method of adjustment based on popularity：The influence scored with popularity user
Based on principle, prediction scoring is adjusted according to project popularity and the variation relation of scoring in proposed algorithm, to carry
The accuracy of high proposed algorithm prediction；
Step 3, proposes a kind of recommendation sort method based on popularity：This method fusion conventional recommendation standard sum term mesh is popular
Spend two indices and carry out recommendation sequence, it is to avoid it is pre- only to go over the interest that historical record or project popularity produce respectively according to user
Survey the defects such as the impersonal theory of error or recommendation.
2. according to claim 1 answered based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm
Method, it is characterised in that the scoring method of adjustment of the prediction based on popularity of the step 2, wherein, popularity refers to
User's degree of concern that a certain project is obtained, refers to the welcome degree of media file in a period of time of weighing, it reflects stream
The probability that media file is at a time asked by client, usual popularity is defined as in interval of time, media text
Part can fit the function H that popularity is changed over time by the number of times of program request.
3. according to claim 1 or 2 analyzed and its in proposed algorithm the influencing mechanism of user interest based on popularity
The method of middle application, it is characterised in that the step 2 comprises the following steps：
Step 2 11：Interest similar users, the collaborative filtering recommending based on user are excavated using the collaborative filtering of user
It is broadly divided into 3 stages：Data are represented；It was found that nearest-neighbors；Produce recommended project collection：Data represent to comment using user-project
Sub-matrix come represent obtain explicit data；It was found that nearest-neighbors are exactly the behavior of scoring and the most like some use of active user
Family, finds the core that nearest-neighbors are Collaborative Filtering Recommendation Algorithms, and the accuracy of nearest-neighbors is directly connected to recommendation and calculated
The recommendation effect of method, typically neighbour is judged whether in Collaborative Filtering Recommendation Algorithm by calculating the similitude between user；
The scoring that recommended project collection treats recommended project according to neighbor user is produced, targeted customer is tried to achieve by score calculation formula and treated
The prediction marking of commending system, the project for then selecting scoring higher is recommended, general to use TOP-N proposed algorithms；In association
With being recommended in filter algorithm commonly using the mode of score in predicting, that is, select the higher project recommendation of N number of prediction scoring to
Targeted customer；Score in predicting formula is commonly defined as:
Wherein, Pi,tRepresent that predictions of the user i to project t is scored, n is the number of similar neighborhood, sim (i, j) is user i and use
Family j similarity, Rj,tIt is scorings of the user j to project t,For average scores of the user j to all items,For it is all
Score average score of the user to project t；
Step 2 12：Prediction scoring adjustment, the prediction scoring for the collaborative filtering of adjustment that scored based on Popularity prediction, in advance
Test and appraisal point formula is P '=P+T*f ((H-H ')/T), and wherein P represents neighbours' prediction scoring that project scores according to history, P ' tables
The true scoring of aspect purpose prediction, f represents the relation of popularity change and scoring change, and H represents project current popularity, H ' tables
Aspect purpose predicts popularity, and T represents the time interval apart from current time, and the formula considers the popularity factor of project
Influence, the variation tendency of stage popularity according to residing for project scores to adjust prediction during prediction；
Step 2 13：Compare the accuracy of two methods prediction scoring, relatively more traditional collaborative filtering is with being based on popularity
The prediction scoring MAE of the collaborative filtering of prediction scoring adjustment.
4. according to claim 1 answered based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm
Method, it is characterised in that the recommendation sort method based on popularity of the step 3, including：
Hot Item Sets：Project popularity is more than the Item Sets of the average popularity of whole data set；
Cold Item Sets：Project popularity is less than the Item Sets of the average popularity of whole data set；
Recommend sequence：Recommend sequence overall target function R=α * General Rank+ (1- α) * Popularity, wherein α be
Weight factor, General Rank are conventional recommendation items selection standard, and Popularity is the popularity of destination item；The public affairs
Formula is the fusion to two interest factors of influence, for different proposed algorithms, and the weight shared by each standard is perhaps different,
α value needs correspondingly to be adjusted, and recommends efficiency to determine the α values of disparity items by comparing.
5. being analyzed and its influencing mechanism of user interest in proposed algorithm based on popularity according to claim 1 or 4
The method of middle application, it is characterised in that the step 3 comprises the following steps：
Step 3 11：Based on traditional collaborative filtering, a new recommendation order standard is designed, i.e. convergence project is pre-
The comprehensive standard of test and appraisal point and project popularity is ranked up to project to be measured, is set up and is recommended sequence overall target function；
Step 3 12：Experimental Comparison is carried out by adjusting weight factor α, to find to promote hot and cold Item Sets recommendation results
Optimal α.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710106106.3A CN106980646A (en) | 2017-02-27 | 2017-02-27 | The method applied based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710106106.3A CN106980646A (en) | 2017-02-27 | 2017-02-27 | The method applied based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106980646A true CN106980646A (en) | 2017-07-25 |
Family
ID=59338178
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710106106.3A Pending CN106980646A (en) | 2017-02-27 | 2017-02-27 | The method applied based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106980646A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108182290A (en) * | 2018-01-30 | 2018-06-19 | 深圳市富途网络科技有限公司 | A kind of evaluation method for the sequence of community content temperature |
| CN108958247A (en) * | 2018-07-02 | 2018-12-07 | 深圳市益鑫智能科技有限公司 | A kind of guided robot |
| CN109408730A (en) * | 2018-12-06 | 2019-03-01 | 重庆理工大学 | A method of reducing Matthew effect in film recommender system |
| CN109409921A (en) * | 2018-08-31 | 2019-03-01 | 阿里巴巴集团控股有限公司 | The recognition methods of the popularity of electronic equipment, device, server |
| CN110674344A (en) * | 2019-08-28 | 2020-01-10 | 北京爱奇艺智慧娱乐科技有限公司 | Method, device, equipment and storage medium for generating model and recommending film |
| CN110825967A (en) * | 2019-10-31 | 2020-02-21 | 中山大学 | Recommendation list re-ranking method for improving diversity of recommendation system |
| CN110995828A (en) * | 2019-11-29 | 2020-04-10 | 北京邮电大学 | Network resource caching method, device and system |
| CN113592588A (en) * | 2021-07-25 | 2021-11-02 | 北京慧橙信息科技有限公司 | E-commerce platform commodity recommendation system and method based on big data collaborative filtering technology |
| CN114510645A (en) * | 2022-04-08 | 2022-05-17 | 浙大城市学院 | Method for solving long-tail recommendation problem based on extraction of effective multi-target groups |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101287082A (en) * | 2008-05-16 | 2008-10-15 | 华东师范大学 | Collaborative filtered recommendation method introducing hotness degree weight of program |
| CN102073709A (en) * | 2010-12-31 | 2011-05-25 | 中国科学技术大学 | Degree-reduced nearest neighbor (DRNN)-based similarity computing method |
| CN104809243A (en) * | 2015-05-15 | 2015-07-29 | 东南大学 | Mixed recommendation method based on excavation of user behavior compositing factor |
-
2017
- 2017-02-27 CN CN201710106106.3A patent/CN106980646A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101287082A (en) * | 2008-05-16 | 2008-10-15 | 华东师范大学 | Collaborative filtered recommendation method introducing hotness degree weight of program |
| CN102073709A (en) * | 2010-12-31 | 2011-05-25 | 中国科学技术大学 | Degree-reduced nearest neighbor (DRNN)-based similarity computing method |
| CN104809243A (en) * | 2015-05-15 | 2015-07-29 | 东南大学 | Mixed recommendation method based on excavation of user behavior compositing factor |
Non-Patent Citations (1)
| Title |
|---|
| 牛抗抗: "流行度对用户户兴趣的影响机制分析及其在推荐算法中的应用研究", 《中国优秀硕学位论全数据库信息科技辑》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108182290A (en) * | 2018-01-30 | 2018-06-19 | 深圳市富途网络科技有限公司 | A kind of evaluation method for the sequence of community content temperature |
| CN108958247A (en) * | 2018-07-02 | 2018-12-07 | 深圳市益鑫智能科技有限公司 | A kind of guided robot |
| CN109409921A (en) * | 2018-08-31 | 2019-03-01 | 阿里巴巴集团控股有限公司 | The recognition methods of the popularity of electronic equipment, device, server |
| CN109408730A (en) * | 2018-12-06 | 2019-03-01 | 重庆理工大学 | A method of reducing Matthew effect in film recommender system |
| CN109408730B (en) * | 2018-12-06 | 2021-08-24 | 重庆理工大学 | Method for reducing Martha effect in film recommendation system |
| CN110674344A (en) * | 2019-08-28 | 2020-01-10 | 北京爱奇艺智慧娱乐科技有限公司 | Method, device, equipment and storage medium for generating model and recommending film |
| CN110825967A (en) * | 2019-10-31 | 2020-02-21 | 中山大学 | Recommendation list re-ranking method for improving diversity of recommendation system |
| CN110825967B (en) * | 2019-10-31 | 2023-04-07 | 中山大学 | Recommendation list re-ranking method for improving diversity of recommendation system |
| CN110995828A (en) * | 2019-11-29 | 2020-04-10 | 北京邮电大学 | Network resource caching method, device and system |
| CN113592588A (en) * | 2021-07-25 | 2021-11-02 | 北京慧橙信息科技有限公司 | E-commerce platform commodity recommendation system and method based on big data collaborative filtering technology |
| CN113592588B (en) * | 2021-07-25 | 2023-10-03 | 深圳市瀚力科技有限公司 | E-commerce platform commodity recommendation system and method based on big data collaborative filtering technology |
| CN114510645A (en) * | 2022-04-08 | 2022-05-17 | 浙大城市学院 | Method for solving long-tail recommendation problem based on extraction of effective multi-target groups |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106980646A (en) | The method applied based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm | |
| CN104935963B (en) | A kind of video recommendation method based on timing driving | |
| CN103745100B (en) | A kind of method of the collaborative filtering recommending of the dominant explicit feedback of project-based mixing | |
| CN109783734A (en) | A kind of mixing Collaborative Filtering Recommendation Algorithm based on item attribute | |
| CN105488233A (en) | Reading information recommendation method and system | |
| CN109815402A (en) | Collaborative Filtering Recommendation Algorithm based on user characteristics | |
| CN103761237A (en) | Collaborative filtering recommending method based on characteristics and credibility of users | |
| CN103514304A (en) | Project recommendation method and device | |
| CN104008193B (en) | A kind of information recommendation method based on group of typical user discovery technique | |
| CN101482884A (en) | Cooperation recommending system based on user predilection grade distribution | |
| CN103559622A (en) | Characteristic-based collaborative filtering recommendation method | |
| CN104166732B (en) | Project collaboration filtering recommendation method based on global scoring information | |
| CN105023178B (en) | A kind of electronic commerce recommending method based on ontology | |
| CN107330727A (en) | A kind of personalized recommendation method based on hidden semantic model | |
| CN105338408B (en) | Video recommendation method based on time factor | |
| CN106846029B (en) | Collaborative filtering recommendation algorithm based on genetic algorithm and novel similarity calculation strategy | |
| KR20170079429A (en) | A clustering based collaborative filtering method with a consideration of users' features and movie recommendation system using thereof | |
| Dong et al. | Research of hybrid collaborative filtering algorithm based on news recommendation | |
| Zhao et al. | Academic social network-based recommendation approach for knowledge sharing | |
| CN110059257B (en) | Project recommendation method based on score correction | |
| CN106651520A (en) | E-business platform long-tail service recommending method and system | |
| CN106951459B (en) | Improvement collaborative filtering recommending method based on Information Entropy | |
| CN106204153A (en) | A kind of two-staged prediction Top N proposed algorithm based on attribute proportion similarity | |
| CN109684536A (en) | A kind of book recommendation method and system based on article k- nearest neighbor algorithm | |
| Shen et al. | A new user similarity measure for collaborative filtering algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170725 |
|
| RJ01 | Rejection of invention patent application after publication |