CN108198045A - The design method of mixing commending system based on e-commerce website data mining - Google Patents
The design method of mixing commending system based on e-commerce website data mining Download PDFInfo
- Publication number
- CN108198045A CN108198045A CN201810091512.1A CN201810091512A CN108198045A CN 108198045 A CN108198045 A CN 108198045A CN 201810091512 A CN201810091512 A CN 201810091512A CN 108198045 A CN108198045 A CN 108198045A
- Authority
- CN
- China
- Prior art keywords
- user
- item
- data
- algorithm
- articles
- 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
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000013461 design Methods 0.000 title claims abstract description 14
- 238000007418 data mining Methods 0.000 title claims abstract description 12
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 70
- 238000001914 filtration Methods 0.000 claims abstract description 24
- 238000013501 data transformation Methods 0.000 claims abstract description 6
- 239000011159 matrix material Substances 0.000 claims description 15
- 238000004458 analytical method Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000004141 dimensional analysis Methods 0.000 claims description 2
- 238000005065 mining Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 6
- 230000006399 behavior Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 3
- 238000004140 cleaning Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- 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)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention relates to a kind of design methods of the mixing commending system based on e-commerce website data mining, include the following steps：Related data required for obtaining algorithm is pre-processed to initial data, is mainly included：Data cleansing, data transformation and attribute reservation；The design of mixing proposed algorithm：The present invention is using the proposed algorithm mixed based on hidden factor model algorithm and the collaborative filtering based on article；The design of commending system method of evaluating performance.The present invention well solves the problems such as recommendation results personalization level is not high, and recommendation accuracy rate is not high, and coverage rate is low, and real-time is poor, new user's cold start-up and not high recommendation results explanation degree.
Description
Technical Field
The invention relates to the technical field of recommendation systems, in particular to a design method of a hybrid recommendation system based on e-commerce website data mining.
Background
With the rapid development of the computer industry and Internet services, electronic commerce is widely applied and popularized with the advantages of convenience, rapidness, comprehensiveness and the like, and gradually forms a global, distributed and dynamic information resource library, which becomes one of the main modes for vast consumers to browse or purchase goods. However, the problem of information overload caused by rapid expansion of internet scale and coverage causes excessive information to be presented simultaneously, and users cannot obtain useful parts from the information, thereby affecting the sales capability of the e-commerce website. For example, eBay china has about 200 million shops, Amazon has millions of goods. In contrast, the knowledge of the consumer is very limited. In a network environment, how to respond to the needs of customers accurately in real time to help customers quickly find needed goods needs to apply an e-commerce recommendation system.
The recommended technology in the field of e-commerce may play the following role:
(1) the method helps the user to find the interested articles, saves the user time and improves the user experience of the user;
(2) the loyalty of the user to the e-commerce website is improved, if the recommendation system can accurately find the interest points of the user and recommend proper resources to the user, the user can generate dependence on the e-commerce website, and therefore a stable enterprise loyalty customer group is established.
However, the conventional e-commerce recommendation algorithm generally has the problems of low personalization degree of recommendation results, low coverage rate, low popularity of recommendation results, poor real-time performance, low interpretation degree of recommendation results and the like, and is suitable for new users to start in a cold mode.
Disclosure of Invention
The invention aims to provide a design method of a hybrid recommendation system based on e-commerce website data mining, which can improve the coverage rate of recommendation results.
The technical scheme adopted by the invention for solving the technical problems is as follows: the design method of the hybrid recommendation system based on the data mining of the electronic commerce website comprises the following steps:
(1) preprocessing original data and acquiring related data required by an algorithm;
(2) recommending an algorithm by adopting a mode of mixing a hidden factor model algorithm and an article-based collaborative filtering algorithm;
(3) and evaluating the recommended algorithm by adopting three indexes of accuracy, recall rate and coverage rate.
The step (1) is specifically as follows: carrying out multi-dimensional analysis on the data to obtain the intrinsic rule of the original data, finding out the data which is irrelevant to an analysis target or needs to be processed by a model, and carrying out data duplication removal, data transformation and data classification on the data.
The step (2) specifically comprises the following substeps:
(21) establishing a user-article similarity matrix;
(22) establishing an article similarity matrix;
(23) recommending similar articles to a target user according to the historical preference of the user by using a collaborative filtering algorithm based on the articles;
(24) and mining the implicit interests of the user for recommendation based on a hidden factor model algorithm.
In the user-item similarity matrix of step (21), the abscissa represents the user, the ordinate represents the item, and the number in the matrix represents the user score, where 0 represents the user's dissatisfaction with the item, 1 represents the user's non-satisfaction with the item, and 2 represents the user's satisfaction with the item.
The step (23) is specifically as follows: analyzing a data set of a user and an article, finding out similar articles according to whether the user browses the articles, and recommending the similar articles to a target user according to the historical preference of the user; suppose a user is Ui(i-1, 2,3, …, n) articles Mj(j ═ 1,2,3, …, m), the article-based collaborative filtering algorithm is largely divided into two steps:
first, for the target user UiAnd its article M with scorejCalculating the item M according to the historical preference data of the user to the itemjSimilarity Sim (j, i) with other scored articles, finding the article MjItem set N with high similarityu(ii) a Wherein the similarity Indicating the user u's score for item i,represents the average score that user u has scored on the items he has browsed;
second, according to all the article sets NuTo select NuMiddle target user UiRecommending the possibly liked items which are not browsed by the target user to the target user and predicting the score; wherein, indicates the preference degree, r, of user u to website ju,iIndicating the preference degree of the user u for the website i, which is the website browsed by the user.
The step (24) is specifically as follows: obtaining an interest vector P (u) of a user u; obtaining a category vector q (m) of an item m; by usingCalculating the preference of the user to the item, wherein rumRepresenting a user's rating of the item; p is a radical ofukMeasures the relation between the interest of the user u and the kth hidden class, qkmMeasuring the relation between the item m and the k hidden class; and sorting in a descending order according to the preference degree, and outputting the top K as recommendations.
And the step (3) is specifically to determine the K value with the highest comprehensive index as the recommended number of the articles to the user by the recommendation system according to the recall rate, the accuracy and the coverage rate curve under different K values.
Advantageous effects
Due to the adoption of the technical scheme, compared with the prior art, the invention has the following advantages and positive effects:
the method improves the traditional project-based collaborative filtering algorithm, weights the influence of the user activity on the similarity of the articles when calculating the similarity, and uses a corrected similarity calculation formula, thereby improving the coverage rate of the recommendation result.
The LFM-based algorithm is an algorithm based on machine learning, is suitable for a system which lacks user interest information and website category information but has a large amount of user behaviors, and can well solve the problem of cold start of a new user.
The recommendation algorithm based on the mixing of the hidden factor model (LFM-based) algorithm and the item-based collaborative filtering algorithm (item-based CF) can well solve the problems of low personalization degree, low recommendation accuracy, low coverage rate, poor real-time performance, cold start of a new user, low interpretation degree of the recommendation result and the like of the recommendation result.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a data processing flow diagram of the data pre-processing stage of the present invention;
FIG. 3 is a generic user-item similarity matrix model
FIG. 4 is a general item similarity matrix model
FIG. 5 is a schematic diagram of an item-based collaborative filtering algorithm;
FIG. 6 is a flow chart of the LFM algorithm of the present invention;
FIG. 7 is a schematic diagram of the hybrid mixing algorithm of the present invention.
Detailed Description
The invention will be further illustrated with reference to the following specific examples. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Further, it should be understood that various changes or modifications of the present invention may be made by those skilled in the art after reading the teaching of the present invention, and such equivalents may fall within the scope of the present invention as defined in the appended claims.
The embodiment of the invention relates to a design method of a hybrid recommendation system based on e-commerce website data mining, which comprises the following steps as shown in figure 1: (1) preprocessing raw data to obtain relevant data required by an algorithm, and mainly comprises the following steps: data cleaning, data transformation and attribute reservation; (2) design of the hybrid recommendation algorithm: the invention adopts a recommendation algorithm based on the mixing of a hidden factor model (LFM-based) algorithm and an article-based collaborative filtering algorithm (item-based CF); (3) and recommending the design of the system performance evaluation method.
The above steps are further described in detail with reference to the following examples:
1. the step (1) of preprocessing the original data to obtain the relevant data required by the algorithm mainly comprises the following steps: data cleaning, data transformation and attribute reservation, specifically comprising:
as shown in fig. 2, firstly, distribution analysis is performed on each dimension of the web page type, the number of clicks, the web page ranking, and the like in the original data, and the intrinsic rules are obtained. And the possible reasons for the occurrence of the result are explained by verifying the data. The method mainly aims at analyzing websites browsed by the user for 1 time, and needs to perform personalized recommendation aiming at the user on the webpages to help the user to find out the webpages interested in or needed by the user.
On the basis of exploration analysis of raw data, data irrelevant to an analysis target or needing to be processed by a model is found and processed. The data processing method comprises the following steps: data cleansing, data integration and data transformation. Through the processing modes, the original data is processed into the human input data required by the model.
On this basis, according to the input data requirement of the recommendation system model, attribute specification needs to be performed on the processed data, and attributes required by the model are extracted.
2. Designing a hybrid recommendation algorithm in the step (2): the invention adopts a recommendation algorithm based on the mixing of a hidden factor model (LFM-based) algorithm and an article-based collaborative filtering algorithm (item-based CF), and specifically comprises the following steps:
2.1 establishing a user-item similarity matrix (U-V matrix)
As shown in fig. 3, fig. 3 is a general user-item similarity matrix model, in which the abscissa represents a user, the ordinate represents an item, and the number in the table represents the rating of the user to the website. Where 0 represents user dissatisfaction with the item, 1 represents no browsing of the item, and 2 represents user satisfaction with the item.
2.2 establishing an item similarity matrix (V-V matrix)
As shown in fig. 4, fig. 4 is a general article similarity matrix model, and the similarity calculation formula is calculated by using formula (2) in fig. 2.3.
2.3 item-based collaborative filtering algorithm (item-based CF) principle and steps:
as shown in fig. 5, fig. 5 is a schematic diagram of an article-based collaborative filtering algorithm, and a general processing procedure of the article-based collaborative filtering system includes analyzing data sets of a user and an article, finding similar articles according to whether the user browses the articles or not, and recommending similar articles to a target user according to the historical preference of the user. Fig. 5 is a diagram of a collaborative filtering recommendation system based on items, from which it can be seen that user a likes item a and item C, user B likes item a, item B, and item C, and user C likes item a. From these historical preferences of the monthly user, it can be analyzed that item a and item C are relatively similar, and that people who like item a all like item C, it can be inferred based on this data that user C is likely to like item C as well, so the system will recommend item C to user C.
Suppose a user is Ui(i ═ 1,2,.. n), website Mj(j＝1，2，...，m)，UiTo MjIs given a score of ri，j. According to the above processing procedure, the collaborative filtering algorithm based on articles is mainly divided into two steps:
first, for the target user UiAnd website M with scorejCalculating the website M according to the historical preference data of the user to the websitejSimilarity Sim (j, i) with other scored websites, finding the website MjHigh-similarity website set Nu。
Second, according to all web sites NuTo select NuMiddle target user UiWebsites that may be liked and that the target user has not browsed are recommended to the target user and the score is predicted.
The similarity between the network stations is generally calculated by using a modified cosine degree equation Sim (j, i).
In the formula, Mu，iRepresents the user u's score for web site i,representing the average score that user u has scored on the websites he has browsed.
The item-based collaborative filtering algorithm (item-based CF) calculates the user's preference for the website by the following formula.
In the formula, rujRepresenting the preference of user u to web site j, web site i being the web site browsed by the user, rujIndicating the degree of preference of user u for website i (for the implicit feedback dataset, if user u has past behavior for website i, then rui1). Then according to rujTo candidate websiteAnd (5) sequencing rows, and recommending websites with high scores for the user.
The item-based collaborative filtering algorithm (item-based CF) is used for recommending by calculating the similarity of the items instead of the similarity of the user, the stability is high, the similarity calculation with large workload can be completed offline in advance, the efficiency is improved, and reasonable recommendation explanation is provided for the recommendation result by using the historical behavior of the user; on the other hand, the corrected collaborative filtering algorithm based on the articles also obviously improves the coverage rate of the recommendation result.
2.4 algorithm principle and steps based on hidden factor model (LFM-based):
based on historical behavior data such as user browsing or scoring, the implicit interest of the user is mined out based on an LFM-based algorithm, namely implicit factors, then the user or a website is classified by the implicit factors, and finally recommendation is carried out through the implicit factors.
Based on the hidden factor model (LFM-based) algorithm, the key problems are how to determine the hidden factors from the scoring data and how to calculate the relationship between the user or website and the hidden factors. Singular value decomposition techniques may be used here. In the recommendation based on the hidden factor model, the user's preference degree for the video is calculated using the following formula.
In the formula, ruiRepresenting the grade of the user u to the website i; p is a radical ofukMeasures the relation between the interest of the user u and the kth hidden class, qkiThe relationship between website i and the kth hidden class is measured.
Fig. 6 is a flowchart of the LFM algorithm, which mainly includes the following steps:
firstly, obtaining an interest vector P (u) of a user u;
secondly, acquiring a category vector q (i) of a website i;
thirdly, calculating the preference degree of the user to the website;
fourthly, sorting according to the preference degree;
and fifthly, outputting a Top-K website list.
Compared with the traditional collaborative filtering algorithm based on articles and users, the hidden factor-based (LFM-based) algorithm has higher accuracy, recall rate and coverage rate, and the recommendation effect is obviously better than that of the traditional recommendation algorithm.
By adopting a recommendation algorithm based on the mixing of a hidden factor model (LFM-based) algorithm and an article-based collaborative filtering algorithm (item-based CF), the weakness of the algorithm caused by a simplified recommendation method can be overcome or alleviated. As the points of sight of the users to the website are different, and the recommendation results generated by the recommendation algorithm usually represent respective angles, a single recommendation result cannot meet the requirements of various crowds, and in order to ensure that the recommendation results have diversity and higher coverage rate, a hybrid mixing method can be used for combining the recommendation results according to a certain proportion and simultaneously presenting the recommendation results to the users.
Fig. 7 is a schematic diagram of a hybrid algorithm of the present invention, in a candidate item generating stage and a scoring generating stage, each recommendation algorithm generates a candidate item set according to unified input data, and ranks candidate items in the candidate item sets, and finally the system comprehensively presents the ranking results of each recommendation algorithm.
3. The design of the method for evaluating the performance of the recommendation system in the step (3) specifically comprises the following steps:
the method adopts three indexes of Precision (Precision), Recall (Recall) and Coverage (Coverage) to evaluate the experimental result. And determining the K value with the highest comprehensive index as the number of the recommended articles of the recommendation system to the user according to the curves of the three performance evaluation indexes under different K values.
Recommending N articles (marked as R (u)) to the user u, enabling the article set liked by the user u on the test set to be T (u), and evaluating the precision of a recommendation algorithm through the accuracy/recall rate:
recall describes how many proportions of user-item score records are included in the final recommendation list, and accuracy describes how many proportions of the final recommendation list are of the occurred user-item score records.
The coverage rate reflects the capability of the recommendation algorithm to find the long tail, and the higher the coverage rate is, the recommendation algorithm can recommend the articles in the long tail to the user. Here, the present invention takes the following coverage definitions:
the coverage rate indicates how large a proportion of the items are contained in the final recommendation list. If all items are recommended to at least one user, then the coverage is 100%.
The design of the performance evaluation method of the recommendation system in the step (3) further includes drawing a recall rate, an accuracy rate and a coverage rate curve under different K values, so that a proper K value is selected as the number of the articles recommended to the user by the recommendation system.
It is not easy to find that the traditional project-based collaborative filtering algorithm is improved, the influence of the user activity on the similarity of the articles is weighted when the similarity is calculated, and a corrected similarity calculation formula is used, so that the coverage rate of the recommendation result is improved. The LFM-based algorithm is an algorithm based on machine learning, is suitable for a system which lacks user interest information and website category information but has a large amount of user behaviors, and can well solve the problem of cold start of a new user. The recommendation algorithm based on the mixing of the hidden factor model (LFM-based) algorithm and the item-based collaborative filtering algorithm (item-based CF) can well solve the problems of low personalization degree, low recommendation accuracy, low coverage rate, poor real-time performance, cold start of a new user, low interpretation degree of the recommendation result and the like of the recommendation result.
Claims (7)
1. A design method of a hybrid recommendation system based on e-commerce website data mining is characterized by comprising the following steps:
(1) preprocessing original data and acquiring related data required by an algorithm;
(2) recommending an algorithm by adopting a mode of mixing a hidden factor model algorithm and an article-based collaborative filtering algorithm;
(3) and evaluating the recommended algorithm by adopting three indexes of accuracy, recall rate and coverage rate.
2. The method for designing a hybrid recommendation system based on e-commerce website data mining as claimed in claim 1, wherein the step (1) is specifically as follows: carrying out multi-dimensional analysis on the data to obtain the intrinsic rule of the original data, finding out the data which is irrelevant to an analysis target or needs to be processed by a model, and carrying out data duplication removal, data transformation and data classification on the data.
3. The method for designing a hybrid recommendation system based on e-commerce website data mining as claimed in claim 1, wherein the step (2) comprises the following sub-steps:
(21) establishing a user-article similarity matrix;
(22) establishing an article similarity matrix;
(23) recommending similar articles to a target user according to the historical preference of the user by using a collaborative filtering algorithm based on the articles;
(24) and mining the implicit interests of the user for recommendation based on a hidden factor model algorithm.
4. The method of claim 3, wherein in the step (21), the abscissa represents the user and the ordinate represents the item, and the number in the matrix represents the user score, wherein 0 represents the user's dissatisfaction with the item, 1 represents the user's unsuspecting item, and 2 represents the user's satisfaction with the item.
5. The method for designing a hybrid recommendation system based on e-commerce website data mining as claimed in claim 3, wherein the step (23) is specifically as follows: analyzing a data set of a user and an article, finding out similar articles according to whether the user browses the articles, and recommending the similar articles to a target user according to the historical preference of the user; suppose a user is Ui(i-1, 2,3, …, n) articles Mj(j ═ 1,2,3, …, m), based onThe collaborative filtering algorithm of the articles is mainly divided into two steps:
first, for the target user UiAnd its article M with scorejCalculating the item M according to the historical preference data of the user to the itemjSimilarity Sim (j, i) with other scored articles, finding the article MjItem set N with high similarityu(ii) a Wherein the similarityMu，iIndicating the user u's score for item i,represents the average score that user u has scored on the items he has browsed;
second, according to all the article sets NuTo select NuMiddle target user UiItems that may be liked and that the target user has not browsed are recommended to the target user and a score is predicted, wherein,rujindicates the preference degree, r, of user u to website ju,iIndicating the preference degree of the user u for the website i, which is the website browsed by the user.
6. The method for designing a hybrid recommendation system based on e-commerce website data mining as claimed in claim 3, wherein the step (24) is specifically as follows: obtaining an interest vector P (u) of a user u; obtaining a category vector q (m) of an item m; by usingCalculating the preference of the user to the item, wherein rumRepresenting a user's rating of the item; p is a radical ofukMeasures the relation between the interest of the user u and the kth hidden class, qkmMeasuring the relation between the item m and the k hidden class; sorting in descending order according to preference degree and outputtingThe first K are recommended.
7. The method as claimed in claim 1, wherein the step (3) is specifically performed by determining the K value with the highest comprehensive index as the number of recommended articles for the user by the recommendation system according to the recall rate, accuracy rate and coverage rate curves at different K values.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810091512.1A CN108198045A (en) | 2018-01-30 | 2018-01-30 | The design method of mixing commending system based on e-commerce website data mining |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810091512.1A CN108198045A (en) | 2018-01-30 | 2018-01-30 | The design method of mixing commending system based on e-commerce website data mining |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108198045A true CN108198045A (en) | 2018-06-22 |
Family
ID=62592039
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810091512.1A Pending CN108198045A (en) | 2018-01-30 | 2018-01-30 | The design method of mixing commending system based on e-commerce website data mining |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108198045A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109063120A (en) * | 2018-08-01 | 2018-12-21 | 北京邮电大学 | A kind of collaborative filtering recommending method and device based on cluster |
| CN109190043A (en) * | 2018-09-07 | 2019-01-11 | 北京三快在线科技有限公司 | Recommended method and device, storage medium, electronic equipment and recommender system |
| CN110059248A (en) * | 2019-03-21 | 2019-07-26 | 腾讯科技（深圳）有限公司 | A kind of recommended method, device and server |
| CN110110139A (en) * | 2019-04-19 | 2019-08-09 | 北京奇艺世纪科技有限公司 | The method, apparatus and electronic equipment that a kind of pair of recommendation results explain |
| CN110119465A (en) * | 2019-05-17 | 2019-08-13 | 哈尔滨工业大学 | Merge the mobile phone application user preferences search method of LFM latent factor and SVD |
| CN110807676A (en) * | 2020-01-07 | 2020-02-18 | 同盾控股有限公司 | Long-tail user mining method and device, electronic equipment and storage medium |
| CN111310042A (en) * | 2020-02-13 | 2020-06-19 | 研祥智能科技股份有限公司 | Collaborative filtering recommendation method and system based on cloud computing |
| CN111325598A (en) * | 2020-01-22 | 2020-06-23 | 深圳大学 | Article recommendation method and device and terminal equipment |
| CN111859157A (en) * | 2020-08-05 | 2020-10-30 | 王孝良 | Recommendation method for expressing relation between information |
| CN111915383A (en) * | 2019-05-09 | 2020-11-10 | 第四范式（北京）技术有限公司 | Window-based article cold start recommendation method and device |
| CN112541407A (en) * | 2020-08-20 | 2021-03-23 | 同济大学 | Visual service recommendation method based on user service operation flow |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102789499A (en) * | 2012-07-16 | 2012-11-21 | 浙江大学 | Collaborative filtering method on basis of scene implicit relation among articles |
| CN103309967A (en) * | 2013-06-05 | 2013-09-18 | 清华大学 | Collaborative filtering method and system based on similarity propagation |
| CN103617540A (en) * | 2013-10-17 | 2014-03-05 | 浙江大学 | E-commerce recommendation method of tracking user interest changes |
| CN106886559A (en) * | 2016-12-28 | 2017-06-23 | 浙江浙大网新集团有限公司 | The collaborative filtering method of good friend's feature and similar users feature is incorporated simultaneously |
| CN107194492A (en) * | 2017-04-13 | 2017-09-22 | 南京邮电大学 | The optimization method that a kind of businessman based on position social networks is recommended |
| CN107256494A (en) * | 2017-05-18 | 2017-10-17 | 深圳大学 | A kind of item recommendation method and device |
| CN107301583A (en) * | 2017-05-26 | 2017-10-27 | 重庆邮电大学 | It is a kind of that method is recommended based on user preference and the cold start-up trusted |
| CN107391713A (en) * | 2017-07-29 | 2017-11-24 | 内蒙古工业大学 | A kind of method and system for solving the problems, such as cold start-up in collaborative filtering recommending technology |
-
2018
- 2018-01-30 CN CN201810091512.1A patent/CN108198045A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102789499A (en) * | 2012-07-16 | 2012-11-21 | 浙江大学 | Collaborative filtering method on basis of scene implicit relation among articles |
| CN103309967A (en) * | 2013-06-05 | 2013-09-18 | 清华大学 | Collaborative filtering method and system based on similarity propagation |
| CN103617540A (en) * | 2013-10-17 | 2014-03-05 | 浙江大学 | E-commerce recommendation method of tracking user interest changes |
| CN106886559A (en) * | 2016-12-28 | 2017-06-23 | 浙江浙大网新集团有限公司 | The collaborative filtering method of good friend's feature and similar users feature is incorporated simultaneously |
| CN107194492A (en) * | 2017-04-13 | 2017-09-22 | 南京邮电大学 | The optimization method that a kind of businessman based on position social networks is recommended |
| CN107256494A (en) * | 2017-05-18 | 2017-10-17 | 深圳大学 | A kind of item recommendation method and device |
| CN107301583A (en) * | 2017-05-26 | 2017-10-27 | 重庆邮电大学 | It is a kind of that method is recommended based on user preference and the cold start-up trusted |
| CN107391713A (en) * | 2017-07-29 | 2017-11-24 | 内蒙古工业大学 | A kind of method and system for solving the problems, such as cold start-up in collaborative filtering recommending technology |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109063120A (en) * | 2018-08-01 | 2018-12-21 | 北京邮电大学 | A kind of collaborative filtering recommending method and device based on cluster |
| CN109063120B (en) * | 2018-08-01 | 2021-05-28 | 北京邮电大学 | Collaborative filtering recommendation method and device based on clustering |
| CN109190043A (en) * | 2018-09-07 | 2019-01-11 | 北京三快在线科技有限公司 | Recommended method and device, storage medium, electronic equipment and recommender system |
| CN109190043B (en) * | 2018-09-07 | 2021-04-23 | 北京三快在线科技有限公司 | Recommendation method and device, storage medium, electronic device and recommendation system |
| CN110059248A (en) * | 2019-03-21 | 2019-07-26 | 腾讯科技（深圳）有限公司 | A kind of recommended method, device and server |
| CN110059248B (en) * | 2019-03-21 | 2022-12-13 | 腾讯科技（深圳）有限公司 | Recommendation method and device and server |
| CN110110139A (en) * | 2019-04-19 | 2019-08-09 | 北京奇艺世纪科技有限公司 | The method, apparatus and electronic equipment that a kind of pair of recommendation results explain |
| CN111915383A (en) * | 2019-05-09 | 2020-11-10 | 第四范式（北京）技术有限公司 | Window-based article cold start recommendation method and device |
| CN110119465A (en) * | 2019-05-17 | 2019-08-13 | 哈尔滨工业大学 | Merge the mobile phone application user preferences search method of LFM latent factor and SVD |
| CN110119465B (en) * | 2019-05-17 | 2023-06-13 | 哈尔滨工业大学 | Mobile phone application user preference retrieval method integrating LFM potential factors and SVD |
| CN110807676A (en) * | 2020-01-07 | 2020-02-18 | 同盾控股有限公司 | Long-tail user mining method and device, electronic equipment and storage medium |
| CN111325598A (en) * | 2020-01-22 | 2020-06-23 | 深圳大学 | Article recommendation method and device and terminal equipment |
| CN111310042A (en) * | 2020-02-13 | 2020-06-19 | 研祥智能科技股份有限公司 | Collaborative filtering recommendation method and system based on cloud computing |
| CN111859157A (en) * | 2020-08-05 | 2020-10-30 | 王孝良 | Recommendation method for expressing relation between information |
| CN112541407A (en) * | 2020-08-20 | 2021-03-23 | 同济大学 | Visual service recommendation method based on user service operation flow |
| CN112541407B (en) * | 2020-08-20 | 2022-05-13 | 同济大学 | Visual service recommendation method based on user service operation flow |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108198045A (en) | The design method of mixing commending system based on e-commerce website data mining | |
| Mathew et al. | Book Recommendation System through content based and collaborative filtering method | |
| JP6356744B2 (en) | Method and system for displaying cross-website information | |
| CN110222272A (en) | A kind of potential customers excavate and recommended method | |
| US9311650B2 (en) | Determining search result rankings based on trust level values associated with sellers | |
| CN105426528A (en) | Retrieving and ordering method and system for commodity data | |
| Badriyah et al. | A hybrid recommendation system for E-commerce based on product description and user profile | |
| CN103559622A (en) | Characteristic-based collaborative filtering recommendation method | |
| CN107330727A (en) | A kind of personalized recommendation method based on hidden semantic model | |
| CN105138508A (en) | Preference diffusion based context recommendation system | |
| CN109446402B (en) | Searching method and device | |
| CN110619559A (en) | Method for accurately recommending commodities in electronic commerce based on big data information | |
| CN113344671A (en) | Trust factor fused personalized recommendation model and construction method | |
| TW201734909A (en) | Method and apparatus for identifying target user | |
| CN110619552A (en) | Member shopping data mining algorithm comprehensive engine | |
| CN108694234A (en) | A kind of service recommendation model based on improvement collaborative filtering | |
| Yao et al. | User-based and item-based collaborative filtering recommendation algorithms design | |
| Abdullah et al. | Infrequent purchased product recommendation making based on user behaviour and opinions in E-commerce sites | |
| Azizah et al. | the Influences of Flash Sale Promotion on Shopping Enjoyment and Impulse Buying of Shopee Consumers Indonesia | |
| Bajpai et al. | Survey paper on dynamic recommendation system for e-commerce | |
| Singh et al. | Implementation of knowledge based collaborative filtering and machine learning for E-Commerce recommendation system | |
| Nalchigar et al. | A large-scale study of online shopping behavior | |
| Yu | Merging attribute characteristics in collaborative filtering to alleviate data sparsity and cold start | |
| Mishra et al. | Solving cold start problem using MBA | |
| Sommer et al. | Recommender systems: the case of repeated interaction in matrix factorization |
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: 20180622 |
|
| RJ01 | Rejection of invention patent application after publication |