CN110188268A - A kind of personalized recommendation method based on label and temporal information - Google Patents
A kind of personalized recommendation method based on label and temporal information Download PDFInfo
- Publication number
- CN110188268A CN110188268A CN201910425039.0A CN201910425039A CN110188268A CN 110188268 A CN110188268 A CN 110188268A CN 201910425039 A CN201910425039 A CN 201910425039A CN 110188268 A CN110188268 A CN 110188268A
- Authority
- CN
- China
- Prior art keywords
- label
- user
- matrix
- preference
- project
- 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
-
- 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
- G06F16/9536—Search customisation based on social or collaborative filtering
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)
Abstract
The invention proposes a kind of personalized recommendation method based on label and temporal information, method includes the following steps: collecting user's history behavioral data, the historical behavior of the user is obtained to the label preference of user according to grading, label and temporal information, the frequency and time effect that are then based on label carry out the correlation between estimation label and project, prediction grading finally is carried out to preference pattern data using matrix factorisation algorithm, personalized recommendation is carried out to user.Association's factorization method that the present invention is combined by using temporal information with label data, can efficiently solve the overfitting problem as caused by sparse ratings data, can effectively improve the validity and user satisfaction of recommendation results.
Description
Technical field
The invention belongs to information recommendation field, in particular to a kind of personalized recommendation method, more specifically, being to propose
A kind of personalized recommendation method based on label and temporal information.
Background technique
With the development of social networks and e-commerce platform, the phenomenon that resulting in information overload, so that user is difficult to obtain
Take the information for meeting user demand, products & services.In addition, when user faces excessive unrelated selection, user experience and purchase
Buying conversion ratio can be remarkably decreased.In this background, the information filtering tool as a kind of personalization, recommender system are solving to believe
Breath overload, help user find the message context oneself really needed and play more and more important guiding function.Nowadays, in order to
It improves service quality, recommender system has been applied to the every field of internet, such as Amazon, TripAdvisor and Ali
Ba Ba etc..
Currently, common recommended technology can be divided into collaborative filtering (CF), content-based recommendation technology (CB) and based on knowing
The recommended technology (KB) of knowledge.However, existing recommended technology ratings data specific to the most of recommender systems of processing is sparse
Property problem when it is all weaker, lead to recommend accuracy decline, thus how in the case where data height is sparse training is effective
Recommender system model is still a great challenge.
Summary of the invention
It is a kind of based on label and temporal information the purpose of the invention is to propose in place of overcome the deficiencies in the prior art
Personalized recommendation method, it is ensured that the overfitting problem as caused by sparse ratings data can be efficiently solved, can effectively be mentioned
The method of the validity and user satisfaction of high recommendation results.
A kind of personalized recommendation method based on label and temporal information, comprising the following steps:
Step 1: for user when entering system platform, system platform can collect the historical behavior data of user automatically, go forward side by side
Row collection, storage and taxonomic revision；
Step 2: system is user tag according to the historical behavior data record being collected into, and is generated as analysis user preference
Data；
Step 3: system grades user tag according to preset rating scale, then passes through record label
Frequency is generated, generates label novelty degree further according to the temporal information of label, it is new by combination tag grade, label frequency and label
Clever degree generates user tag preference, to generate user-label preference matrix；
Step 4: and then before determining label and project according to the label frequency and label novelty degree that generate in step 3
Relationship, to obtain label-item association matrix, project is graded according to preset rating scale and is used by system
Family-project appraisal matrix；
Step 5: the relationship application matrix Factor minute between user tag preference, label grade and label and project is utilized
Resolving Algorithm constructs the collaboration singular value decomposition co-SVD model in matrix factorisation model:
Step 6: co-SVD model obtains user tag preference by integrated label grade, frequency and novelty, according to
The novelty of label frequency and a project determines the correlation between label and project, according to user tag preference and label
Correlation between project respectively defines co-SVD model to user-project appraisal matrix, user-label preference matrix
Factorization, the overfitting problem of collaboration processing label and temporal information source are carried out with label-item association matrix, and utilizes square
Array factor decomposition algorithm carries out prediction grading to preference pattern data, by prediction marking and queuing, obtain recommendation results, to user into
Row personalized recommendation.
Preferably, the personalized recommendation technology can be divided into collaborative filtering (CF), content-based recommendation technology (CB) and
Knowledge based engineering recommended technology (KB), wherein Collaborative Filtering Recommendation Algorithm is divided into: collaborative filtering memory-based and be based on model
Collaborative filtering, the matrix factorisation algorithm (MF) is a kind of collaborative filtering based on model.
The beneficial effects of the present invention are: the present invention proposes a kind of personalized recommendation side based on label and temporal information
Method, personalized recommendation solve the problems, such as information overload in many expert systems as a kind of information filtering method；In the present invention
The influence analyzed come measure time user preference using the novelty of label can effectively solve the problem that user preference dynamic change is asked
Topic；The present invention not only allows for the evaluations matrix between user and project, and passes through identification user and label (label and item
Mesh) between relationship, define user-label preference (label-item association) matrix, Harmonious Matrix point carried out to multiple matrixes
Solution；Over-fitting caused by this personalized recommendation method effective solution based on label and temporal information evaluation Sparse
Problem, while can significantly improve and recommend precision and user satisfaction sheet.
Detailed description of the invention
Fig. 1: for the present invention is based on the personalized recommendation method schematic diagrames of label and temporal information；
Fig. 2: for user of the present invention-label preference matrix structural schematic diagram；
Fig. 3: for label of the present invention-item association matrix schematic diagram；
Fig. 4: for user of the present invention-project appraisal matrix schematic diagram；
Fig. 5: for the SVD model structure of broad sense；
Fig. 6: for co-SVD model structure of the invention；
Fig. 7: for the personalized recommendation algorithm flow chart of broad sense；
Fig. 8: user's history behavioral data collection structure figure.
Specific embodiment
Below in conjunction with attached drawing, the technical characteristic and advantage above-mentioned and other to the present invention are clearly and completely described,
Obviously, described embodiment is only section Example of the invention, rather than whole embodiments.
Refering to fig. 1 shown in -8, a kind of personalized recommendation method based on label and temporal information, comprising the following steps:
Step 1: for user when entering system platform, system platform can collect the historical behavior data of user automatically, go forward side by side
Row is collected, storage and taxonomic revision, concrete mode are as shown in Figure 8；
Step 2: system is user tag according to the historical behavior data record being collected into, and is generated as analysis user preference
Data；
Step 3: system grades user tag according to preset rating scale, then passes through record label
Frequency is generated, generates label novelty degree further according to the temporal information of label, it is new by combination tag grade, label frequency and label
Clever degree generates user tag preference, to generate user-label preference matrix；
Step 4: and then before determining label and project according to the label frequency and label novelty degree that generate in step 3
Relationship, to obtain label-item association matrix, project is graded according to preset rating scale and is used by system
Family-project appraisal matrix；
Step 5: the relationship application matrix Factor minute between user tag preference, label grade and label and project is utilized
Resolving Algorithm constructs the collaboration singular value decomposition co-SVD model in matrix factorisation model:
Step 6: co-SVD model obtains user tag preference by integrated label grade, frequency and novelty, according to
The novelty of label frequency and a project determines the correlation between label and project, according to user tag preference and label
Correlation between project respectively defines co-SVD model to user-project appraisal matrix, user-label preference matrix
Factorization, the overfitting problem of collaboration processing label and temporal information source are carried out with label-item association matrix, and utilizes square
Array factor decomposition algorithm carries out prediction grading to preference pattern data, by prediction marking and queuing, obtain recommendation results, to user into
Row personalized recommendation.
Personalized recommendation technology can be divided into collaborative filtering (CF), content-based recommendation technology (CB) and Knowledge based engineering and push away
Technology (KB) is recommended, wherein Collaborative Filtering Recommendation Algorithm is divided into: collaborative filtering memory-based and the collaborative filtering based on model, institute
Stating matrix factorisation algorithm (MF) is a kind of collaborative filtering based on model, and MF algorithm is than the collaborative filtering based on memory
Algorithm has better precision of prediction.
The label that user generates implys that their preference, and it is dilute to alleviate data that these preferences can be used as the supplement of grading
Property problem is dredged, the definition of label grade is to observe that label reflects some of user's care in all properties of an article
Attribute, and conveyed the true intention that user uses label；The frequency of label refers to that ordinary user uses label for labelling project
Number；The novelty of label is hidden feature of the influence for measure time to label preference by excavation user and project.
System case study on implementation described above is only schematical, wherein the unit illustrated as separation module
It may or may not be physically separated, module shown as a unit may or may not be physics list
Member, it can it is in one place, or may be distributed on multiple subsystems.It can select according to the actual needs wherein
Some or all of the modules realize the purpose of disclosure scheme.Those of ordinary skill in the art are not making the creative labor
In the case where, it can it understands and implements.
Particular embodiments described above has carried out further the purpose of the present invention, technical scheme and beneficial effects
It is described in detail, it should be understood that the above is only a specific embodiment of the present invention, the protection being not intended to limit the present invention
Range.It particularly points out, to those skilled in the art, all within the spirits and principles of the present invention, that is done any repairs
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (2)
1. a kind of personalized recommendation method based on label and temporal information, which comprises the following steps:
Step 1: for user when entering system platform, system platform can collect the historical behavior data of user automatically, and be received
Collection, storage and taxonomic revision；
Step 2: system is user tag according to the historical behavior data record being collected into, and is generated as the number of analysis user preference
According to；
Step 3: system grades user tag according to preset rating scale, is then generated by record label
Frequency generates label novelty degree further according to the temporal information of label, passes through combination tag grade, label frequency and label novelty degree
User tag preference is generated, to generate user-label preference matrix；
Step 4: and then label and the pass before project are determined according to the label frequency and label novelty degree that generate in step 3
System, to obtain label-item association matrix, project is graded according to preset rating scale and is used by system
Family-project appraisal matrix；
Step 5: it is resolved using the relationship application matrix Factor minute between user tag preference, label grade and label and project
Method constructs the collaboration singular value decomposition co-SVD model in matrix factorisation model:
Step 6: co-SVD model obtains user tag preference, according to label by integrated label grade, frequency and novelty
The novelty of frequency and a project determines the correlation between label and project, according to user tag preference and label and item
Correlation between mesh respectively defines co-SVD model to user-project appraisal matrix, user-label preference matrix and mark
Label-item association matrix carry out Factorization, the overfitting problem of collaboration processing label and temporal information source, and utilization matrix because
Sub- decomposition algorithm carries out prediction grading to preference pattern data, by prediction marking and queuing, obtains recommendation results, carries out to user a
Propertyization is recommended.
2. a kind of personalized recommendation method based on label and temporal information according to claim 1, which is characterized in that institute
Collaborative filtering (CF), content-based recommendation technology (CB) and Knowledge based engineering recommended technology can be divided by stating personalized recommendation technology
(KB), wherein Collaborative Filtering Recommendation Algorithm is divided into: collaborative filtering memory-based and the collaborative filtering based on model, the matrix
Factoring algorithm (MF) is a kind of collaborative filtering based on model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910425039.0A CN110188268A (en) | 2019-05-21 | 2019-05-21 | A kind of personalized recommendation method based on label and temporal information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910425039.0A CN110188268A (en) | 2019-05-21 | 2019-05-21 | A kind of personalized recommendation method based on label and temporal information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110188268A true CN110188268A (en) | 2019-08-30 |
Family
ID=67717079
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910425039.0A Pending CN110188268A (en) | 2019-05-21 | 2019-05-21 | A kind of personalized recommendation method based on label and temporal information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110188268A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111125566A (en) * | 2019-12-11 | 2020-05-08 | 贝壳技术有限公司 | Information acquisition method and device, electronic equipment and storage medium |
| CN111260526A (en) * | 2020-01-20 | 2020-06-09 | 北京明略软件系统有限公司 | Figure track behavior analysis and estimation method and device |
| CN111931053A (en) * | 2020-08-10 | 2020-11-13 | 中国工商银行股份有限公司 | Item pushing method and device based on clustering and matrix decomposition |
| CN112417302A (en) * | 2020-12-08 | 2021-02-26 | 六晟信息科技(杭州)有限公司 | Big data-based information content intelligent analysis recommendation processing system |
| CN113643781A (en) * | 2021-06-25 | 2021-11-12 | 合肥工业大学 | Health intervention scheme personalized recommendation method and system based on time sequence early warning signal |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070214133A1 (en) * | 2004-06-23 | 2007-09-13 | Edo Liberty | Methods for filtering data and filling in missing data using nonlinear inference |
| CN105868372A (en) * | 2016-03-31 | 2016-08-17 | 广州酷狗计算机科技有限公司 | Label distribution method and device |
| CN109272390A (en) * | 2018-10-08 | 2019-01-25 | 中山大学 | The personalized recommendation method of fusion scoring and label information |
-
2019
- 2019-05-21 CN CN201910425039.0A patent/CN110188268A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070214133A1 (en) * | 2004-06-23 | 2007-09-13 | Edo Liberty | Methods for filtering data and filling in missing data using nonlinear inference |
| CN105868372A (en) * | 2016-03-31 | 2016-08-17 | 广州酷狗计算机科技有限公司 | Label distribution method and device |
| CN109272390A (en) * | 2018-10-08 | 2019-01-25 | 中山大学 | The personalized recommendation method of fusion scoring and label information |
Non-Patent Citations (2)
| Title |
|---|
| LING LUO等: "Personalized recommendation by matrix co-factorization with tags and time information", 《EXPERT SYSTEMS WITH APPLICATIONS》 * |
| 郭娣: "融合标签流行度和时间权重的矩阵分解推荐算法", 《小型微型计算机系统》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111125566A (en) * | 2019-12-11 | 2020-05-08 | 贝壳技术有限公司 | Information acquisition method and device, electronic equipment and storage medium |
| CN111125566B (en) * | 2019-12-11 | 2021-08-31 | 贝壳找房(北京)科技有限公司 | Information acquisition method and device, electronic equipment and storage medium |
| CN111260526A (en) * | 2020-01-20 | 2020-06-09 | 北京明略软件系统有限公司 | Figure track behavior analysis and estimation method and device |
| CN111931053A (en) * | 2020-08-10 | 2020-11-13 | 中国工商银行股份有限公司 | Item pushing method and device based on clustering and matrix decomposition |
| CN112417302A (en) * | 2020-12-08 | 2021-02-26 | 六晟信息科技(杭州)有限公司 | Big data-based information content intelligent analysis recommendation processing system |
| CN113643781A (en) * | 2021-06-25 | 2021-11-12 | 合肥工业大学 | Health intervention scheme personalized recommendation method and system based on time sequence early warning signal |
| CN113643781B (en) * | 2021-06-25 | 2023-07-04 | 合肥工业大学 | Personalized recommendation method and system for health intervention scheme based on time sequence early warning signal |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110188268A (en) | A kind of personalized recommendation method based on label and temporal information | |
| CN104391849B (en) | Incorporate the collaborative filtering recommending method of time contextual information | |
| CN102866992B (en) | A kind of method and device showing product information in webpage | |
| CN108021929A (en) | Mobile terminal electric business user based on big data, which draws a portrait, to establish and analysis method and system | |
| CN102609533A (en) | Kernel method-based collaborative filtering recommendation system and method | |
| CN103473354A (en) | Insurance recommendation system framework and insurance recommendation method based on e-commerce platform | |
| CN104899763A (en) | Personalized recommendation method based on bilateral diffusion of bipartite network | |
| CN110852852A (en) | Industrial Internet product recommendation system and method | |
| CN102254028A (en) | Personalized commodity recommending method and system which integrate attributes and structural similarity | |
| CN101454771A (en) | System and method of segmenting and tagging entities based on profile matching using a multi-media survey | |
| CN107464068A (en) | Enterprise development trend forecasting method and its system based on neutral net | |
| CN103345699A (en) | Personalized food recommendation method based on commodity forest system | |
| CN105335518A (en) | Method and device for generating user preference information | |
| CN111949887A (en) | Item recommendation method and device and computer-readable storage medium | |
| CN101753573A (en) | RSS information recommendation method based on collaborative filtering | |
| CN106157156A (en) | A kind of cooperation recommending system based on communities of users | |
| CN102664744B (en) | Group-sending recommendation method in network message communication | |
| CN115860880B (en) | Personalized commodity recommendation method and system based on multi-layer heterogeneous graph convolution model | |
| Muñoz-Leiva et al. | Detecting salient themes in financial marketing research from 1961 to 2010 | |
| CN105654361A (en) | Method and system for assessing credit based on community O2O | |
| CN105654267A (en) | Cold-chain logistic stowage intelligent recommendation method based on spectral cl9ustering | |
| CN103412865A (en) | Method and system for carrying out notification on website projects | |
| CN101986301B (en) | Inverse neighbor analysis-based collaborative filtering recommendation system and method | |
| CN103854206A (en) | Method and device for analyzing group characteristics | |
| CN107844585A (en) | A kind of recommendation method for considering overlapping community's effect |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190830 |