CN108810640B - Television program recommendation method - Google Patents
Television program recommendation method Download PDFInfo
- Publication number
- CN108810640B CN108810640B CN201810624032.7A CN201810624032A CN108810640B CN 108810640 B CN108810640 B CN 108810640B CN 201810624032 A CN201810624032 A CN 201810624032A CN 108810640 B CN108810640 B CN 108810640B
- Authority
- CN
- China
- Prior art keywords
- program
- user
- information
- programs
- target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/47—End-user applications
- H04N21/475—End-user interface for inputting end-user data, e.g. personal identification number [PIN], preference data
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
- H04N21/4508—Management of client data or end-user data
- H04N21/4532—Management of client data or end-user data involving end-user characteristics, e.g. viewer profile, preferences
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
- H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
- H04N21/4662—Learning process for intelligent management, e.g. learning user preferences for recommending movies characterized by learning algorithms
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
- H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
- H04N21/4667—Processing of monitored end-user data, e.g. trend analysis based on the log file of viewer selections
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
- H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
- H04N21/4668—Learning process for intelligent management, e.g. learning user preferences for recommending movies for recommending content, e.g. movies
Abstract
The invention requests to protect a recommendation method of a television program, which comprises the following steps: firstly, receiving language information input by a user; obtaining characteristic elements according to the language information, retrieving and evaluating the electronic program information in the electronic program list database by taking the characteristic elements as retrieval key words, and extracting related electronic program information according to the similarity; constructing a statistical model by using the feature set and a machine learning method; matching programs in the electronic program list database by using the statistical model; outputting the matching result to a user; thirdly, the user sends program request information, the rating information of the user on the program is obtained, a user-program rating matrix is obtained, and the program is preliminarily recommended; if the matching result is half the same as the matching result, recommending the program according to the matching result, if the matching result is not half the same as the matching result, selecting the program with the second highest score as the target program, repeating the steps for filling until more than half of the program is the same as the target program, and recommending the program list to the user. The mixed recommendation method can improve the accuracy of program recommendation.
Description
Technical Field
The invention belongs to the technical field of recommendation, and particularly belongs to a recommendation method of television programs.
Background
The traditional method for recommending programs aims to acquire the types, the albums and the singers of the programs heard by a user and recommend the programs of the corresponding program types, the albums or the singers of the programs to the user, and most of the current music recommendation methods are based on collaborative filtering.
Disclosure of Invention
The present invention is directed to solving the above problems of the prior art. A television program recommendation method for improving recommendation accuracy is provided.
The technical scheme of the invention is as follows:
a method for recommending television programs, comprising the steps of:
firstly, receiving language information input by a user; extracting related electronic program information from an electronic program list database storing the electronic program information according to the language information; performing feature selection on the extracted electronic program information to obtain feature elements, performing retrieval and evaluation processing on the electronic program information in the electronic program list database by taking the feature elements as retrieval keywords, extracting relevant electronic program information from the electronic program list database according to the processing result, and performing feature selection on the electronic program information extracted according to the language information and the electronic program information extracted according to the feature elements to obtain new feature elements, wherein the feature elements specifically comprise: when the received language information is a keyword set, carrying out logic calculation on the keyword set and then extracting related electronic program information from the electronic program list database; and acquiring the associated information of the new feature elements from a knowledge base with stored language knowledge, and constructing a feature set, wherein the method specifically comprises the following steps: when the received language information is a phrase or a sentence, word segmentation processing is firstly carried out, word segmentation results are calculated to obtain a space model preferred by a user, then the similarity between the space model and the electronic program information in the electronic program table database is calculated, and related electronic program information is extracted according to the similarity; constructing a statistical model by using the feature set and a machine learning method; matching programs in the electronic program list database by using the statistical model; outputting the matching result to a user;
thirdly, the user sends program request information, the rating information of the user on the programs is obtained, a user-program rating matrix is obtained, a historical rating record set is built for each user and each program, a user set is built at the same time, and the programs in the program set are sorted according to the rating programs of the user from large to small; selecting a program with the highest score as a target program according to the sequence of the programs in the user set; calculating the similarity between the rest programs and the target program according to the user-program scoring matrix; selecting programs with similarity greater than a first set value with the target program to construct a priority program set of the target program; calculating the common historical score difference mean value of the target program and each priority program, and if the two user histories do not have common programs, not calculating; selecting programs with the common score difference mean value smaller than a second set value to construct a final similar program set, and filling a user-program score matrix by using the similar program set of the target program; selecting the most similar program from the unfilled data in the user-program scoring matrix by adopting a similarity threshold method and a common scoring difference mean value, and refilling the user-program scoring matrix by using the similar program to perform preliminary recommendation on the program; if the matching result is half the same as the matching result, recommending the program according to the matching result, if the matching result is not half the same as the matching result, selecting the program with the second highest score as the target program, repeating the steps for filling until more than half of the program is the same as the target program, and recommending the program list to the user.
Further, the process of constructing the priority program set of the target program is as follows:
selecting a program with the highest score as a target program according to the sequence of the user in the user set; calculating the similarity between other programs and a target program by using a user-online program scoring matrix and through a Pearson correlation coefficient formula;
wherein, simu,vRepresenting the similarity of program u and program v, Iu,vFor a common scoring set of merchants for program u and program v, Rui、RviThe scores of the program u and the program v for the merchant i respectively,
respectively mean scores of the program u and the program v; and selecting programs with similarity greater than alpha to the target program to construct a preferred neighbor program set p _ N (u) of the target program.
Further, the selecting process of the most similar program is as follows:
calculating the average value of the historical common score differences of the target program and each priority program, wherein the calculation formula is as follows:
wherein avg (u, v) is the sum of programs u and vHistorical common score mean, I, of program vu′,vNews cast set, R, for historical joint scoring of target program u and program vui、RviRespectively scoring the program i for the program u and the program v; and selecting the priority programs with the common score difference mean value smaller than beta to construct the final most similar program N (u).
Further, the step of calculating the daily score of each program in the program playing log according to the weight of the program source and whether the program is completely played includes: by the formula: calculating the current day score of each program in the program playing log, wherein snow is the score of the current day of the program, weight is the weight of the program source, list represents whether the program is completely played, when the playing time of the program is not less than 85% of the real time length of the program, the program is considered to be completely played, a value of 1 is given, when the playing time of the program is less than 85% of the real time length of the program, the song is considered not to be completely played, and a value of 0 is given; the full play is 1, and the non-full play is 0.
The invention has the advantages of
On one hand, the invention extracts the relevant electronic program information from the electronic program list database according to the language information input by the user, performs characteristic selection to obtain characteristic elements, calls the information stored in the knowledge base to expand the characteristic elements to obtain the characteristic set of the user interest and hobby space, and constructs a statistical model by using the characteristic set and a machine learning method, so as to match the electronic program list database and output a matching result to the user, thereby realizing program recommendation, solving the problem of 'cold start' in the prior art, and improving the precision, performance and practicability of the program recommendation. The method is executed at the user side, and does not involve the acquisition of the personal information of the user at the network side server side or the user side, so that the privacy information of the user can be fully ensured not to be leaked, and the confidentiality is improved. In addition, the characteristic elements can be used as search keywords to search and evaluate the electronic program list database, and then program preselection is carried out again according to the processing result, so that the user interest space can be further expanded, and the program recommendation precision can be improved. On the other hand, the neighbor is screened by adding the historical common score difference mean value, and the neighbor with larger score difference with the target item is removed, so that the selection of the similar neighbor set is more accurate, the program which one user dislikes is effectively prevented from being recommended as the program which the other user likes, and the recommendation is more accurate. The invention firstly carries out the first filling from the perspective of the user and then carries out the second filling from the perspective of the program, so that the filling of the sparse user-program scoring matrix is more complete. And simultaneously, each step of filling adopts a dynamic filling mode, the target programs are sequentially selected according to the size of the grading number for filling, and the matrix filled each time is the matrix filled by the last target. The dynamic filling increases the common evaluation number of the programs, the similarity calculation is more accurate, the filling of a sparse user-program evaluation matrix is more accurate, the finally given recommendation list is more in line with the mind of the user, the stickiness of the user to the recommendation system is improved, and when the evaluation results of the two are different, the program with the second highest evaluation number is selected again as the target program, the steps are repeated for filling to obtain the recommended program, so that the limitation of the program with the highest evaluation number is avoided, the program with the second highest evaluation number is recommended, and the recommendation is more humanized.
Drawings
Fig. 1 is a flow chart of a program recommendation method based on a hybrid recommendation algorithm according to a preferred embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be described in detail and clearly with reference to the accompanying drawings. The described embodiments are only some of the embodiments of the present invention.
The technical scheme for solving the technical problems is as follows:
fig. 1 shows a method for recommending a television program, which includes the following steps:
firstly, receiving language information input by a user; extracting related electronic program information from an electronic program list database storing the electronic program information according to the language information; performing feature selection on the extracted electronic program information to obtain feature elements, performing retrieval and evaluation processing on the electronic program information in the electronic program list database by taking the feature elements as retrieval keywords, extracting relevant electronic program information from the electronic program list database according to the processing result, and performing feature selection on the electronic program information extracted according to the language information and the electronic program information extracted according to the feature elements to obtain new feature elements, wherein the feature elements specifically comprise: when the received language information is a keyword set, carrying out logic calculation on the keyword set and then extracting related electronic program information from the electronic program list database; and acquiring the associated information of the new feature elements from a knowledge base with stored language knowledge, and constructing a feature set, wherein the method specifically comprises the following steps: when the received language information is a phrase or a sentence, word segmentation processing is firstly carried out, word segmentation results are calculated to obtain a space model preferred by a user, then the similarity between the space model and the electronic program information in the electronic program table database is calculated, and related electronic program information is extracted according to the similarity; constructing a statistical model by using the feature set and a machine learning method; matching programs in the electronic program list database by using the statistical model; outputting the matching result to a user;
thirdly, the user sends program request information, the rating information of the user on the programs is obtained, a user-program rating matrix is obtained, a historical rating record set is built for each user and each program, a user set is built at the same time, and the programs in the program set are sorted according to the rating programs of the user from large to small; selecting a program with the highest score as a target program according to the sequence of the programs in the user set; calculating the similarity between the rest programs and the target program according to the user-program scoring matrix; selecting programs with similarity greater than a first set value with the target program to construct a priority program set of the target program; calculating the common historical score difference mean value of the target program and each priority program, and if the two user histories do not have common programs, not calculating; selecting programs with the common score difference mean value smaller than a second set value to construct a final similar program set, and filling a user-program score matrix by using the similar program set of the target program; selecting the most similar program from the unfilled data in the user-program scoring matrix by adopting a similarity threshold method and a common scoring difference mean value, and refilling the user-program scoring matrix by using the similar program to perform preliminary recommendation on the program; if the matching result is half the same as the matching result, recommending the program according to the matching result, if the matching result is not half the same as the matching result, selecting the program with the second highest score as the target program, repeating the steps for filling until more than half of the program is the same as the target program, and recommending the program list to the user.
Preferably, the process of constructing the priority program set of the target program is as follows:
selecting a program with the highest score as a target program according to the sequence of the user in the user set; calculating the similarity between other programs and a target program by using a user-online program scoring matrix and through a Pearson correlation coefficient formula;
wherein, simu,vRepresenting the similarity of program u and program v, Iu,vFor a common scoring set of merchants for program u and program v, Rui、RviThe scores of the program u and the program v for the merchant i respectively,
respectively mean scores of the program u and the program v; and selecting programs with similarity greater than alpha to the target program to construct a preferred neighbor program set p _ N (u) of the target program.
Preferably, the selection process of the most similar program is as follows:
calculating the average value of the historical common score differences of the target program and each priority program, wherein the calculation formula is as follows:
whereinAvg (u, v) is the mean of the historical common scores of program u and program v, Iu′,vNews cast set, R, for historical joint scoring of target program u and program vui、RviRespectively scoring the program i for the program u and the program v; and selecting the priority programs with the common score difference mean value smaller than beta to construct the final most similar program N (u).
Preferably, the step of calculating the daily score of each program in the program play log according to the weight of the program source and whether the program is completely played includes: by the formula: calculating the current day score of each program in the program playing log, wherein snow is the score of the current day of the program, weight is the weight of the program source, list represents whether the program is completely played, when the playing time of the program is not less than 85% of the real time length of the program, the program is considered to be completely played, a value of 1 is given, when the playing time of the program is less than 85% of the real time length of the program, the song is considered not to be completely played, and a value of 0 is given; the full play is 1, and the non-full play is 0.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.
Claims (2)
1. A method for recommending television programs, comprising the steps of:
firstly, receiving language information input by a user; according to the language information, extracting the related electronic program information from the electronic program list database storing the electronic program information, which specifically comprises the following steps: when the received language information is a keyword set, carrying out logic calculation on the keyword set and then extracting related electronic program information from the electronic program list database;
carrying out feature selection on the extracted electronic program information to obtain feature elements, carrying out retrieval and evaluation processing on the electronic program information in the electronic program list database by taking the feature elements as retrieval keywords, and extracting related electronic program information from the electronic program list database according to the processing result; then, carrying out feature selection on the electronic program information extracted according to the language information and the electronic program information extracted according to the feature elements to obtain new feature elements; acquiring the associated information of the new characteristic elements from a knowledge base with stored language knowledge, and constructing a characteristic set;
when the received language information is a phrase or a sentence, word segmentation processing is firstly carried out, word segmentation results are calculated to obtain a space model preferred by a user, then the similarity between the space model and the electronic program information in the electronic program table database is calculated, and related electronic program information is extracted according to the similarity; constructing a statistical model by using the feature set and a machine learning method; matching programs in the electronic program list database by using the statistical model; outputting the matching result to the user;
thirdly, the user sends program request information, the rating information of the user on the programs is obtained, a user-program rating matrix is obtained, a historical rating record set is built for each user and each program, a user set is built at the same time, and the programs in the program set are sorted according to the rating programs of the user from large to small; selecting a program with the highest score as a target program according to the sequence of the programs in the user set; calculating the similarity between the rest programs and the target program according to the user-program scoring matrix; selecting programs with similarity greater than a first set value with the target program to construct a priority program set of the target program; calculating the common historical score difference mean value of the target program and each priority program, and if the two user histories do not have common programs, not calculating; selecting programs with the common score difference mean value smaller than a second set value to construct a final similar program set, and filling a user-program score matrix by using the similar program set of the target program; selecting the most similar program from the unfilled data in the user-program scoring matrix by adopting a similarity threshold method and a common scoring difference mean value, and refilling the user-program scoring matrix by using the similar program to perform preliminary recommendation on the program; if the matching result is half the same as the matching result, recommending the program according to the matching result, if the matching result is not half the same as the matching result, selecting the program with the second highest score as the target program, repeating the steps for filling, and recommending the program list to the user until more than half of the programs are the same;
the process of constructing the priority program set of the target program is as follows:
selecting a program with the highest score as a target program according to the sequence of the user in the user set; calculating the similarity between other programs and a target program by using a user-online program scoring matrix and through a Pearson correlation coefficient formula;
wherein, simu,vRepresenting the similarity of program u and program v, Iu,vFor a common scoring set of merchants for program u and program v, Rui、RviThe scores of the program u and the program v for the merchant i respectively,
respectively mean scores of the program u and the program v; selecting programs with similarity larger than alpha with the target program to construct an optimal neighbor program set p _ N (u) of the target program;
the selection process of the most similar program is as follows:
calculating the average value of the historical common score differences of the target program and each priority program, wherein the calculation formula is as follows:
wherein avg (u, v) is the average value of the historical common scores of the program u and the program v, Iu′,vCo-scoring a set of television programs, R, for the history of a target program u and a program vui、RviRespectively scoring the program i for the program u and the program v; and selecting the priority programs with the common score difference mean value smaller than beta to construct the final most similar program N (u).
2. The method of claim 1, wherein the step of calculating the daily rating of each program in the program play log according to the weight of the program source and whether the program is completely played comprises: by the formula: calculating the current day score of each program in the program playing log, wherein, snore is the score of the current day of the program, weight is the weight of the program source, listen represents whether the program is completely played, when the playing time of the program is not less than 85% of the real time length of the program, the program is considered to be completely played, a value of 1 is given, when the playing time of the program is less than 85% of the real time length of the program, the program is considered not to be completely played, and a value of 0 is given; the full play is 1, and the non-full play is 0.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810624032.7A CN108810640B (en) | 2018-06-15 | 2018-06-15 | Television program recommendation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810624032.7A CN108810640B (en) | 2018-06-15 | 2018-06-15 | Television program recommendation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108810640A CN108810640A (en) | 2018-11-13 |
| CN108810640B true CN108810640B (en) | 2020-12-18 |
Family
ID=64086860
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810624032.7A Active CN108810640B (en) | 2018-06-15 | 2018-06-15 | Television program recommendation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108810640B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109640176B (en) * | 2018-12-18 | 2021-01-22 | 北京字节跳动网络技术有限公司 | Method and apparatus for generating information |
| CN111767470A (en) * | 2020-06-30 | 2020-10-13 | 山东汇贸电子口岸有限公司 | Article recommendation method, device, equipment and computer-readable storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102231166A (en) * | 2011-07-12 | 2011-11-02 | 浙江大学 | Collaborative recommendation method based on social context |
| CN102541920A (en) * | 2010-12-24 | 2012-07-04 | 华东师范大学 | Method and device for improving accuracy degree by collaborative filtering jointly based on user and item |
| CN103514255A (en) * | 2013-07-11 | 2014-01-15 | 江苏谐云智能科技有限公司 | Method for collaborative filtering recommendation based on item level types |
| CN104317900A (en) * | 2014-10-24 | 2015-01-28 | 重庆邮电大学 | Multiattribute collaborative filtering recommendation method oriented to social network |
| CN105023178A (en) * | 2015-08-12 | 2015-11-04 | 电子科技大学 | Main body-based electronic commercere commendation method |
| CN106610970A (en) * | 2015-10-21 | 2017-05-03 | 上海文广互动电视有限公司 | Collaborative filtering-based content recommendation system and method |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7630916B2 (en) * | 2003-06-25 | 2009-12-08 | Microsoft Corporation | Systems and methods for improving collaborative filtering |
| US7444313B2 (en) * | 2003-09-03 | 2008-10-28 | Microsoft Corporation | Systems and methods for optimizing decision graph collaborative filtering |
| WO2012079254A1 (en) * | 2010-12-17 | 2012-06-21 | 北京交通大学 | Program recommending device and program recommending method |
| CN107977373B (en) * | 2016-10-21 | 2020-09-08 | 北京酷我科技有限公司 | Song recommendation method |
-
2018
- 2018-06-15 CN CN201810624032.7A patent/CN108810640B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102541920A (en) * | 2010-12-24 | 2012-07-04 | 华东师范大学 | Method and device for improving accuracy degree by collaborative filtering jointly based on user and item |
| CN102231166A (en) * | 2011-07-12 | 2011-11-02 | 浙江大学 | Collaborative recommendation method based on social context |
| CN103514255A (en) * | 2013-07-11 | 2014-01-15 | 江苏谐云智能科技有限公司 | Method for collaborative filtering recommendation based on item level types |
| CN104317900A (en) * | 2014-10-24 | 2015-01-28 | 重庆邮电大学 | Multiattribute collaborative filtering recommendation method oriented to social network |
| CN105023178A (en) * | 2015-08-12 | 2015-11-04 | 电子科技大学 | Main body-based electronic commercere commendation method |
| CN106610970A (en) * | 2015-10-21 | 2017-05-03 | 上海文广互动电视有限公司 | Collaborative filtering-based content recommendation system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108810640A (en) | 2018-11-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109101479B (en) | Clustering method and device for Chinese sentences | |
| CN108304437B (en) | automatic question answering method, device and storage medium | |
| CN108153876B (en) | Intelligent question and answer method and system | |
| CN107993724B (en) | Medical intelligent question and answer data processing method and device | |
| CN107609101B (en) | Intelligent interaction method, equipment and storage medium | |
| CN107797984B (en) | Intelligent interaction method, equipment and storage medium | |
| CN109829104B (en) | Semantic similarity based pseudo-correlation feedback model information retrieval method and system | |
| US20210056571A1 (en) | Determining of summary of user-generated content and recommendation of user-generated content | |
| CN106709040B (en) | Application search method and server | |
| CN109189990B (en) | Search word generation method and device and electronic equipment | |
| WO2020119063A1 (en) | Expert knowledge recommendation method and apparatus, computer device, and storage medium | |
| CN112667794A (en) | Intelligent question-answer matching method and system based on twin network BERT model | |
| US20040249808A1 (en) | Query expansion using query logs | |
| CN105809473B (en) | Training method for matching model parameters, service recommendation method and corresponding device | |
| JP6355840B2 (en) | Stopword identification method and apparatus | |
| CN103885937A (en) | Method for judging repetition of enterprise Chinese names on basis of core word similarity | |
| CN110990533B (en) | Method and device for determining standard text corresponding to query text | |
| CN109829045A (en) | A kind of answering method and device | |
| CN110866102A (en) | Search processing method | |
| CN115905489B (en) | Method for providing bidding information search service | |
| CN112100396A (en) | Data processing method and device | |
| CN112632261A (en) | Intelligent question and answer method, device, equipment and storage medium | |
| CN112417119A (en) | Open domain question-answer prediction method based on deep learning | |
| CN108810640B (en) | Television program recommendation method | |
| CN115374362A (en) | Multi-way recall model training method, multi-way recall device and electronic equipment |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20201130 Address after: 315000 Kechuang building (14-2), (14-4), No.16, Buzheng lane, Haishu District, Ningbo City, Zhejiang Province Applicant after: Zhejiang Guangye Software Technology Co., Ltd Address before: No. 12, No. 12, Lake Yun street, Chongqing, Chongqing Applicant before: CHONGQING ZHIAO TECHNOLOGY Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |