CN111586089A - Client-side and server-side content recommendation system and method based on vector scoring - Google Patents
Client-side and server-side content recommendation system and method based on vector scoring Download PDFInfo
- Publication number
- CN111586089A CN111586089A CN202010201636.8A CN202010201636A CN111586089A CN 111586089 A CN111586089 A CN 111586089A CN 202010201636 A CN202010201636 A CN 202010201636A CN 111586089 A CN111586089 A CN 111586089A
- Authority
- CN
- China
- Prior art keywords
- content
- dimension
- score
- user
- scoring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/55—Push-based network services
-
- 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
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
- G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
- G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Bioethics (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to a system and a method for recommending contents of a client and a server based on vector scoring, wherein the system for recommending contents of the client comprises a system content scoring acquisition module, a user preference scoring module, a relation scoring module and a recommending module, and the content scoring acquisition module is used for acquiring one or more content dimension scores of a plurality of online contents; the user preference scoring module is used for acquiring user preference dimension scores; the relation scoring module determines the relation scoring of the online content by the user based on the content dimension scoring and the user dimension preference scoring of the online content; and the recommending module recommends the content to the user according to the ranking of the relation scores. The method is based on a vector scoring mechanism, and calculates the relation score matched with the user preference according to the user preference dimension score and the content dimension score, so that the related data types are few, the algorithm is simple, the calculation is rapid, the efficiency is high, and the user privacy is effectively protected.
Description
Technical Field
The invention relates to the field of internet application, in particular to a system and a method for recommending contents of a client and a server based on vector scoring.
Background
With the development of network technology, network users are increasing rapidly, and content providers providing various content services for users are increasing, for example, video websites providing video services, e-commerce platforms providing commodity transactions, reading websites providing novel reading, and various websites providing various information. Various content recommendation techniques are also developed in a large number for the purpose of attracting users and recommending contents/services. Generally, a content provider collects user data, such as personal information, network behavior data, browsed content data, etc., at a server, analyzes the user data to obtain personal preferences of the user, and calculates and recommends content matching the personal preferences of the user to the user based on the content provided by the content provider. However, the existing content recommendation method involves a large variety of data and a complex algorithm, so that the recommendation efficiency is not high. In addition, the existing content recommendation method needs to collect user information at a server, such as personal information of the user, browsing data and the like, so that the personal privacy of the user is leaked, the user is reluctant to reuse the service of the content provider, and the user is lost.
Disclosure of Invention
Aiming at the technical problems in the prior art, the invention provides a system and a method for recommending contents of a client and a server based on vector scoring, which are used for simplifying data, improving recommendation efficiency and protecting user privacy.
In order to solve the above technical problem, according to one aspect of the present invention, the present invention provides a client content recommendation system based on vector scoring, the system comprising a content score obtaining module, a user preference scoring module, a relationship scoring module and a recommendation module, wherein the content score obtaining module is configured to obtain one or more content dimension scores of a plurality of online contents; the user preference scoring module is configured to obtain a user preference dimension score; the relationship scoring module is configured to determine a relationship score for online content by the user based on a content dimension score and a user dimension preference score for the online content; the recommendation module is configured to recommend content to a user in an order of the relationship scores.
According to another aspect of the present invention, the present invention provides a method for recommending content on a client based on vector scoring, which includes the following steps:
obtaining one or more content dimension scores for a plurality of online content;
acquiring a user preference dimension score;
determining a plurality of relationship scores for a plurality of content and the user based on one or more content dimension scores and user preference dimension scores for a plurality of online content; and
and recommending the content to the user according to the ranking of the plurality of content relation scores.
According to another aspect of the present invention, the present invention provides a server content recommendation system based on vector scoring, the system comprising a content scoring providing module and a content providing module, wherein the content scoring providing module is configured to provide a content dimension scoring list in response to a request from an application client; the content providing module is configured to respond to a content request from an application client and send content requested by the content request to the application client.
According to another aspect of the present invention, the present invention provides a server content recommendation method based on vector scoring, including the following steps:
receiving a request from an application client, and providing a content dimension scoring list to the application client; and
the method comprises the steps of receiving a content request from an application client, and sending content requested by the content request to the application client.
The invention constructs different content dimensions for the online content, sets different content dimension scores for the online content according to the specific online content, and the content dimension scores of the online content can be persistently stored in a database to provide basic data for calculating the relationship scores between the online content and different users. By constructing the user preference dimension corresponding to the content dimension and the score thereof, the user preference data is simplified, and when the user preference and the content are matched and the recommended content is determined, the algorithm is simple and the calculation amount is small, so that the calculation is fast and the recommendation efficiency is high; the content is recommended to the user at the client based on the data generated in the process that the user uses the content client, so that the privacy of the user is protected, and the purpose of content recommendation is achieved.
Drawings
Preferred embodiments of the present invention will now be described in further detail with reference to the accompanying drawings, in which:
FIG. 1 is a schematic block diagram illustrating an application of a system for recommending content to a client based on vector scoring according to an embodiment of the present invention;
FIG. 2 is a functional block diagram of a client-side content recommendation system based on vector scoring according to one embodiment of the present invention;
FIG. 3 is a functional block diagram of a user preference scoring module according to one embodiment of the present invention;
FIG. 4 is a functional block diagram of a recommendation module according to one embodiment of the present invention;
FIG. 5 is a flowchart of a method for client-side content recommendation based on vector scoring, according to one embodiment of the invention;
FIG. 6 is a flow diagram of obtaining user preference dimensions and their scores according to one embodiment of the invention;
FIG. 7 is a functional block diagram of a server-side content recommendation system based on vector scoring according to an embodiment of the present invention;
FIG. 8 is a functional block diagram of a content score construction module according to one embodiment of the present invention;
FIG. 9 is a flowchart of a server-side content recommendation method based on vector scoring according to an embodiment of the present invention; and
FIG. 10 is a flow diagram for building a content dimension score for an application, according to one embodiment of the invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In the following detailed description, reference is made to the accompanying drawings that form a part hereof and in which is shown by way of illustration specific embodiments of the application. In the drawings, like numerals describe substantially similar components throughout the different views. Various specific embodiments of the present application are described in sufficient detail below to enable those skilled in the art to practice the teachings of the present application. It is to be understood that other embodiments may be utilized and structural, logical or electrical changes may be made to the embodiments of the present application.
Fig. 1 is a schematic block diagram of an application of a client content recommendation system based on vector scoring according to an embodiment of the present invention. The application in the invention can at least provide the user with browsing of online content, and the application can be a novel reading application, a video viewing application, a picture query application, a news browsing application, a commodity transaction application and the like according to the provided content. The application server 301 and the database 302 are located in the cloud, and the application client 10 and the client content recommendation system are located in the local user terminal. The two are connected with the cloud end through the network, and operations such as online content browsing and content recommendation are achieved. In one embodiment, the database 302 may store online content provided by the application, or the application server 301 connects with the content server 303, and the content server 303 provides specific content to the user. The database 302 also stores other data, such as user data, which includes personal information of the user, browsing records of the user, etc., and the database 302 also stores content description information of each content, content page link addresses, and a content dimension score list thereof. The list of content dimension scores includes each content dimension score for each online content. Wherein the content dimension corresponds to an online content category provided by the application. For example, in an application that provides novel reading material, the application classifies a plurality of categories for the novel provided, such as "history", "science fiction", "city", "suspicion", and so on. Correspondingly, the content dimension is also "history", "science fiction", "city", "suspicion", and the like. For a novel, its content may span multiple categories, such as being attributed to both the "history" category, the "science fiction" category, and the "suspicion" category. Thus, the corresponding content dimensions of the novel include three categories, "history", "science fiction" and "suspicion". The content of the novel is analyzed manually or in an AI manner, and the score corresponding to the three content dimensions can be determined according to the comparison scoring standard, such as 8 scores for "history", 5 scores for "science fiction", 5 scores for "suspense", and the like. The list of content dimensions stored in the database 302 is thus shown in table 1:
table 1:
where the numbers 0, 1, 2 … … are the numbers of the content in the database, xxxx, Yyyyyy, Sssss are the names of the novels, "history", "science fiction", "suspicion", etc. are the dimensions of the content, and the numbers 0-9 in the table are the specific scores.
The application server 301 may send the list of content dimensions to the client content recommendation system 20 in response to a request from the client content recommendation system, or may actively send the list of content dimensions to the application client 10 for storage in its local storage module 40.
FIG. 2 is a schematic block diagram of a client-side content recommendation system based on vector scoring according to one embodiment of the present invention.
The client content recommendation system 20 includes a content score obtaining module 21, a user preference score module 22, a relationship score module 23, and a recommendation module 24, where the content score obtaining module 21 is configured to obtain each content dimension score of each online content provided by an application. In one embodiment, the content score obtaining module 21 sends a request to the application server 301, and the application server 301 reads the content dimension score list from the database 302 and sends the content dimension score list to the content score obtaining module 21. In another embodiment, since the application server 301 actively pushes the content dimension score list to the application client 10 and stores the content dimension score list in the local storage module 40 of the user terminal 1, the content score obtaining module 21 queries the local storage module 40 to obtain the content in the content dimension score list.
The preference scoring module 22 is configured to obtain a user preference dimension score, wherein the preference dimension is one or more of the content dimensions. The user preference dimension score may be obtained in advance and stored in the local storage module 40 of the user terminal 1, and may be updated at any time according to a change in the user preference. The user preference scoring module 22 queries the local storage module 40 to determine whether a list of user preference dimension scores is stored locally. And if not, acquiring the user preference dimension score according to data generated in the process of browsing the content by the user. When the user browses the content, the application client 10 or/and the application server 301 records corresponding data, for example, browsing path, browsing content name, content amount thereof, and specific browsing amount. For example, according to a browsing record, the user clicks two categories of "history" and "science fiction", browses a novel mmm under the category of "science fiction", which has a total of 300 chapters, and the user reads the first 8 chapters. In another browsing record, the user browses a novel Nnnnn under the "City" category, which has 600 chapters in total, and the user reads the first chapter 1. From these browsing history data, the user preference scoring module 22 may obtain a user preference dimension score. Specifically, as shown in fig. 3, a functional block diagram of the user preference scoring module in one embodiment is shown. The user preference scoring module 22 includes the information extracting unit 221, the preference dimension building unit 222, and the preference scoring unit 223, wherein the information extracting unit 221 extracts user browsing content information from data generated in the process of browsing content by the user, and the user browsing content information includes content and browsing volume, such as the aforementioned novel name, all chapters, and reading volume. The browsing amount is such as the number of words of a novel, chapters, the playing time of a video and the like. The preference dimension building unit 22 builds a user preference dimension according to the content dimension of the browsed content, that is, counts a plurality of contents browsed by the user, and determines the content dimension of the plurality of contents as the user preference dimension. For example, the user browses three novels together, where the content dimension of the first novels and their scores are "history 3 score", "science fiction 8 score", the content dimension of the second novels is "science fiction 8 score" and "suspense 6 score", and the content dimension of the third novels is "suspense 8 score" and "city 9 score". Then "history", "science fiction", "suspicion" and "city" are determined as the user's preference dimension.
The preference scoring unit 23 determines the preference dimension score according to the user browsing content information and the scoring criteria. In one embodiment, the scoring criteria defines a calculation rule, determines a browsing depth according to the browsing amount, and determines a preference dimension score by using the browsing depth and a corresponding content dimension score. As in the above example, the user browses three novel in total, wherein the content dimension and the score of the first novel are history 3 and science fiction 8, and the reading depth is chapter 80/chapter 600; the content dimensions of the second novel are science fiction 8 and suspense 6, and the reading depth is 400 chapters/500 chapters; the third novel has a content dimension of 8 suspense and 9 City, and a reading depth of 65 chapters/1000 chapters. Firstly, the reading depths of the contents are respectively as follows:
the first novel: 80/600 ═ 0.13;
the second novel: 400/500 ═ 0.80;
the third novel: 65/1000 ═ 0.065.
And then obtaining the favorite dimension score y according to the content dimension score and the reading depth of each novel:
"history" y1 ═ 3 × 0.13 ═ 0.39;
"science fiction" y2 ═ (8 × 0.13+8 × 0.80)/2 ═ 3.72;
"suspense" y3 ═ (6 × 0.80+8 × 0.065)/2 ═ 2.66;
"urban" y4 ═ 9 × 0.065 ═ 0.59.
Thus, the user's favorite dimension scores are summarized in Table 2 below:
TABLE 2
| Preference dimension | History of | Science fiction | Suspense questions of the present invention | City of a city |
| Scoring | 0.39 | 3.72 | 2.66 | 0.59 |
As can be seen, the preference scoring unit 23 calculates the user preference dimension score according to the following equation 1-1:
wherein the content of the first and second substances,
scoring a content dimension for the ith content;
is the reading depth, x, of the ith contentiIs the reading amount of the ith content, XiIs the total content of the ith content, and n is the total content with the same content dimension.
The present invention is not limited to the above formula 1-1, and other factors (such as reading amount) may be added on the basis of the above formula to form other formulas according to different considered factors, and those skilled in the art can determine the corresponding formula according to the specific considered factors, and details are not repeated herein.
In one embodiment, the user preference dimension scores as shown in Table 2 are stored in local storage module 40 and updated at any time or on a regular basis.
The relationship scoring module 23 determines a relationship score of the user for the content based on a content dimension score and a user dimension preference score of the content. The relationship scoring module 3 determines the proximity of the content dimension score and the user dimension preference score by calculating the difference between the content dimension score and the user dimension preference score, wherein the greater the proximity, the more the content conforms to the user preference. And quantifying the proximity to obtain a relationship score. In one embodiment, the user's relationship score to a content is calculated using equation 1-2:
where RMSE is a relationship score, the lower its score, the greater the proximity of the content to the user's preferences. y istScoring the user's tth preference dimension,
the t-th content dimension of the content is scored.
The recommendation module 24 recommends content to the user in a low-to-high ranking of the relationship scores. In one embodiment, as shown in fig. 4, the recommending module 24 further includes a sorting unit 241, a push information generating unit 242, and a pushing unit 243, where the sorting unit 241 sorts a plurality of online contents provided by the application according to the content relationship score of each online content. After the calculation of the relationship scoring module 23, each online content provided by the application has a respective relationship score RMSE, and the sorting unit 241 sorts the relationship scores RMSE from low to high, that is, the lowest relationship score calculated according to the formula 1-2 is ranked first and is closest to the user preference. The push information generating unit 242 selects content description information of one or more top-ranked contents according to the set recommendation conditions, such as the number of contents recommended each time, and generates push information according to format requirements, such as that each recommended content should include a text introduction, a picture, a link, and the like. The content description information may be obtained by requesting the application server 301, or may be actively pushed to the application client by the application server 301, and stored in the local storage module 40 of the terminal. The content description information may include content text introduction, cover pictures, typical video clips, and the like. In some embodiments, an update interval and an update condition of the recommended content are further set, and when the update interval is reached or the update condition is met, the push information is regenerated. The pushing unit 243 pushes the push information to the application client 10. The application client 10 is provided with a recommendation display position, and after receiving the push information, the application client displays the push information in the recommendation display position.
Fig. 5 is a flowchart of a method for recommending content on a client side based on vector scoring according to an embodiment of the present invention. Wherein the method comprises the steps of:
and step S10, acquiring the content dimension scores of all online contents provided by the application. If the application server side actively pushes the content dimension scores to the application client side, the content dimension scores of all online contents can be obtained by inquiring the local storage module. If the application server side does not actively push the content dimension scores to the application client side, a request is sent to the application server side through the application client side, the application server side reads the content dimension score list from the database and sends the content dimension score list to the application client side, and therefore corresponding content dimension scores of all the contents can be obtained.
And 11, acquiring the favorite dimension score of the user. Wherein the preference dimension is one or more of content dimensions. The user preference dimension score may be calculated in advance, stored in the local storage module 40 of the user terminal 1, and updated at any time according to a change in the user preference. When the user favorite dimension score is obtained, the local storage module 40 may be queried to determine whether a user favorite dimension score list is stored locally. And if not, acquiring the user preference dimension score according to data generated in the process of browsing the content by the user. When a user browses contents, the application server side and/or the application client side can record browsing data of the user, a record is generated during each access, and each record at least comprises a browsed content name, the reading amount of the browsing and the like. By browsing the recorded data, the contents browsed by the user together and the total reading amount of the contents can be obtained. The preference dimension and the score of the user can be obtained through the data. As shown in fig. 6, the method includes the following steps:
and step S111, inquiring all browsing records and classifying according to browsing contents. For example, browsing records related to different contents are grouped into one group, so that all browsing record data can be grouped into a plurality of groups according to the browsing contents.
In step S112, a group of data is analyzed to extract the total reading amount. For example, the group data includes a plurality of records, each record describes the reading content, such as the chapter of a novel, and the total reading chapter, i.e., the total reading amount, is obtained through the plurality of records.
And step S113, calculating the reading depth of the user for the content according to the total reading amount and all chapter contents of the novel.
Step S114, judging whether the reading depth of all the contents is calculated, if not, returning to step S112 to continue the extraction and calculation of the next group of data, and if the reading depth of all the contents is analyzed, returning to step S115.
Step S115, building a user preference dimension. The content dimension and the score of each online content are obtained, repeated content dimensions are removed, and the content dimensions related to all the contents are used as user preference dimensions.
Step S116, a user preference dimension is taken.
In step S117, a corresponding content dimension score is acquired. For example, for the embodiment in the foregoing system, for the "science fiction" user preference dimension, the content dimension scores of the two online contents corresponding thereto are both 8 points; for the 'suspense' user preference dimension, the content dimension scores of the two online contents corresponding to the 'suspense' user preference dimension are respectively 6 and 8; for the 'history' user preference dimension, one content dimension corresponding to the 'history' user preference dimension is scored as 3; for the "urban" user preference dimension, one content dimension corresponding thereto is scored 9.
Step S118, calculating to obtain the user preference dimension score according to the content dimension of each content and the score and the reading depth thereof by using the formula 1-1.
Wherein the content of the first and second substances,
scoring a content dimension for the ith content;
is the reading depth, x, of the ith contentiIs the reading amount of the ith content, XiIs the total content of the ith content, and n is the total content with the same content dimension.
The user preference dimension score shown in table 2 is obtained by the calculation of equation 1-1.
Step S119, determining whether all the user preference dimension scores have been calculated. If all the user preference dimension scores are calculated, the process is ended, and if not, the process returns to the step S116, and then one user preference dimension is taken to calculate the score until all the user preference dimension scores are obtained.
And step S12, determining the relation score of the user to the content based on the content dimension score and the user dimension preference score of the content. In one embodiment, the user's relationship score for each content is calculated according to equations 1-2:
wherein, RMSE is the relation score, the lower the score is, the closer the content is to the user preference; y istFor the t-th user dimension preference score,
for the t-th content dimension of the contentAnd (5) grading the degree.
For example, for 3 contents in table 1 and the user preference dimension score in table 2, the relationship scores between the 3 contents and the user can be calculated by formula 1-2 as follows:
as can be seen from the calculated scores, the content yyyy best meets the user's preference. Moreover, as can be seen from the user preference score and the content dimension score of the content, the content yyyy is indeed most suitable for the user preference.
Equations 1-2 are merely one type of relational expression for calculating the relational score, and other equations similar to those described above may be used for calculation by one skilled in the art.
And step S13, recommending contents to the user according to the sequence of the relationship scores from low to high. After obtaining the relationship score between each online content and the user, the online contents are sorted, and since the smaller the relationship score value in this embodiment is, the closer the relationship score value is to the preference of the user, in this embodiment, the online contents are sorted in the order of the relationship scores from low to high, and then the top n online contents are selected as the push content according to the push condition, such as the push number n. And the recommended content can be displayed in the recommendation display position of the application client, wherein the content on the recommended content comes from the content recommendation system. For example, according to the push number n, the first n contents are selected, according to the format of the display requirement, if the format of the display requirement includes the requirements of text content and format thereof, picture and format thereof and link, the content descriptions of the first n contents are selected, necessary text introduction is extracted from the content descriptions, the text content is generated according to the format, then a picture is selected from the content descriptions, the picture is modified according to the size requirement, the contents are combined into push information according to the typesetting requirement and sent to the application client, and therefore the current n contents are displayed in the recommended display position of the application client.
In addition, the content in the recommended display position can be updated according to different updating conditions. For example, when the display period is reached, the next group of n contents is reselected to generate the push information. Or when the newly added user browses the content, which causes the user favorite dimension score to change, at the moment, the relationship score between the user and each content needs to be recalculated, which causes the relationship score ordering of the online content to change, and then new content needs to be recommended again.
Fig. 7 is a schematic block diagram of a server-side content recommendation system based on vector scoring according to an embodiment of the present invention, the server-side content recommendation system 50 includes a content scoring providing module 501 and a content providing module 502, wherein the content scoring providing module 501 is configured to provide a content dimension scoring list to the application client 10. In one embodiment, the content scoring module 501 actively pushes the content dimension scoring list to the application client 10 and periodically updates the content dimension scoring list. In another embodiment, the content score providing module 501 sends a list of content dimension scores to the application client 10 in response to a request from the application client.
The content providing module 502 is configured to respond to a content request of an application client, and provide corresponding content to the application client that sent the content request. When a user clicks a certain recommended content in the recommended display position in the application client 10, the application client 10 sends the content link and the content request together to the content providing module 502 through the application server 301, the content providing module 502 sends the corresponding content located in the database 302 or the content server 303 to the application client 10 through the application server 301 according to the content link, and the application client 10 receives and displays the corresponding content page to the user.
In order to obtain each content dimension score of the online content, in an embodiment, the server content recommendation system 50 further includes a content dimension score construction module 503, configured to construct one or more content dimension scores for the online content provided by the application. As shown in fig. 8, which is a schematic block diagram of a content dimension scoring construction module according to an embodiment of the present invention, the content dimension scoring construction module 503 includes a dimension construction unit 5031, a content analysis unit 5032, and a content scoring unit 5033, where the dimension construction unit 5031 constructs a plurality of content dimensions for scoring for the online content provided by the present application. For example, when a plurality of categories such as "science fiction", "suspicion", "history", "city", and the like are set in the novel reading application, all the categories are acquired, and each category is set as one content dimension, wherein the content category information of the application is stored in the database, and all the content categories can be acquired by reading corresponding data from the data. The content analysis unit 5032 performs content analysis on the online content based on the constructed content dimension. For example, content may be read from the content server 303 manually or in an AI mode, and content analysis may be performed on the content based on different content dimensions. Wherein different content dimensions have different criteria, for example, for content dimension "history", the degree of correlation with "history" is determined by obtaining keywords about time, person description, event description, etc. during analysis. The content scoring unit 5033 determines a score for each content dimension of the content according to the content analysis result and the scoring criteria for the corresponding content dimension. For example, the content dimension score is 10 levels from 0 to 9, a standard is determined for each level, and the content analysis result is compared with the level standards to obtain a corresponding level, that is, the content dimension score is obtained. The content dimension score construction module 503 obtains the content dimension score of each content, and stores the content dimension score in the database 302. As shown in table 1 above.
As shown in fig. 9, a method for recommending content by a server based on vector scoring according to an embodiment of the present invention includes:
step S50 provides the application client with a list of content dimension scores. In one embodiment, the content dimension score list is generated and stored in a database by the application server, and at this time, the content dimension score list can be read from the database. And the application server side actively pushes the content dimension scoring list to the application client side and updates the content dimension scoring list at regular time or in real time. Or, in this embodiment, the content dimension score list is sent to the application client in response to a request of the application client.
Step S51, based on the content request, provides the corresponding content page to the user application client that sent the content request. Referring to fig. 1, when a user clicks a certain recommended content in the recommended presentation position in the application client 10, the application client 10 sends the content link and a content request together to the application server, according to the content link, if the content is located in the database 302 or the content server 303, the application server reads a content page corresponding to the content link from the database 302 or the content server 303 and sends the content page to the application client 10, and the application client 10 receives and presents the content page to the user.
The server-side content recommendation method provided by the invention also provides a method for constructing the content dimension score of the online content. Specifically, as shown in fig. 10, it is a flowchart for constructing a content dimension score for an application, and specifically includes the following steps:
and step S61, constructing a content dimension for scoring according to the online content category provided by the application. Since the applications that generally provide online content store and manage the online content in a classified manner, the applications are not only used for management of content providers, but also facilitate users to quickly find favorite content. In this embodiment, online content categories are used as content dimensions. For example, for a novel reading application, its categories "science fiction", "suspicion", "history", "city" are set as content dimensions, respectively.
In step S62, one of the contents is analyzed. In one embodiment, each content dimension is provided with 10 levels, each level of each content dimension is provided with different criteria, for example, for the "historical" content dimension, the levels are 0-9 by keywords in terms of "time", "people's clothing description", "place wording", "appliances", etc., and the amount of application thereof. When analyzing the "historical" content dimension of a content, it is analyzed from several aspects.
Step S63, determining a score of each content dimension of the content according to the content analysis result and the score standard of the corresponding content dimension. In step S62, a content is analyzed in "time", "person clothing description", "place word", and "tool" to obtain keywords and their repetition, and the analysis result is compared with the level criteria in 0-9 levels to obtain the most suitable level, i.e., the dimension score of the content.
And step S64, judging whether all the online contents have been scored by the dimension of the contents, if the online contents have not been scored by the dimension of the contents, returning to step S62, and if the online contents have been scored by the dimension of the contents, ending the process.
After the content dimension scores of all online contents are obtained, the content dimension scores are stored in the database 303 in a table form, and the table can be updated as the contents increase.
The method is based on a vector scoring mechanism, the user favorite dimension score obtained by the user browsing data is matched with the content dimension scores of different contents at the user terminal, the algorithm is simple, the calculation amount is small, the recommended content can be quickly and efficiently obtained, and the browsing data of the user does not need to be sent to the server side, so that the user privacy is protected.
The above embodiments are provided only for illustrating the present invention and not for limiting the present invention, and those skilled in the art can make various changes and modifications without departing from the scope of the present invention, and therefore, all equivalent technical solutions should fall within the scope of the present invention.
Claims (28)
1. A client-side content recommendation system based on vector scoring, comprising:
a content score acquisition module configured to acquire one or more content dimension scores for a plurality of online content;
a user preference scoring module configured to obtain a user preference dimension score;
a relationship scoring module configured to determine a relationship score for online content by the user based on a content dimension score and a user dimension preference score for the online content; and
a recommendation module configured to recommend content to the user in an order of the relationship scores.
2. The system of claim 1, the content score acquisition module further configured to acquire at least a content dimension score list from an application server; or inquiring a locally stored content dimension score list pushed by the application server.
3. The system of claim 1, the user preference scoring module further configured to query a locally stored list of user preference dimension scores.
4. The system of claim 1, the user preference scoring module further configured to generate or update a user preference dimension score based on data generated during user browsing of content.
5. The system of claim 4, the user preference scoring module further configured to comprise:
an information extraction unit configured to extract user browsing content information from data generated during a user browsing content;
a preference dimension construction unit configured to construct a user preference dimension according to a content dimension of the browsing content; and
a preference scoring unit configured to determine the preference dimension score according to user browsing content information and a scoring criterion.
6. The system of claim 1, the relationship scoring module further configured to calculate a closeness of a content dimension score of content to a dimension preference score of the user, the closeness quantified as a relationship score of the content to the user.
7. The system of claim 6, the relationship scoring module further configured to quantify how close a content dimension score of online content is to a dimension preference score of the user using scoring relationships 1-2:
where RMSE is the user's relationship score to a content, ytScoring a user's tth preference dimension,
scoring a tth content dimension of the content.
8. The system of claim 1, the recommendation module further comprising:
a ranking unit configured to rank a plurality of contents according to a relationship score of contents and the user;
a push information generating unit configured to generate push information according to the one or more contents ordered in the top; and
a push unit configured to push the push information to an application client.
9. A method for recommending client content based on vector scoring comprises the following steps:
obtaining one or more content dimension scores for a plurality of online content;
acquiring a user preference dimension score;
determining a plurality of relationship scores for a plurality of content and the user based on one or more content dimension scores and user preference dimension scores for a plurality of online content; and
and recommending the content to the user according to the ranking of the plurality of content relation scores.
10. The method of claim 9, further comprising:
acquiring a content dimension score list of online content from an application server; or
And querying a locally stored content dimension score list pushed by the application server.
11. The method of claim 9, further comprising locally retrieving the stored latest user preference dimension scores; alternatively, the user preference dimension score is generated or updated based on data generated during the user's browsing of content.
12. The method of claim 11, wherein the user preference dimension score is obtained by:
extracting user browsing content information from data generated in the process of browsing content by a user, wherein the browsing content information at least comprises content dimension scores and browsing amount of the browsing content;
constructing a user preference dimension according to the content dimension of the browsed content; and
and calculating the corresponding user favorite dimension score according to the browsing amount of the user browsing content and the content dimension score.
13. The method of claim 12, further comprising:
classifying the information of the user browsing content according to the browsing content;
calculating the browsing depth of each browsing content according to the browsing amount of each browsing content;
acquiring a content dimension score of each browsing content; and
calculating a user preference dimension score according to equation 1-1:
wherein y is a user preference dimension score;
scoring a content dimension for the ith content;
is the browsing depth, x, of the ith contentiIs the browsing volume, X, of the ith contentiN is the total content of the ith content, and the total number of browsed contents with the same content dimension.
14. The method of claim 9, further comprising: and calculating the degree of closeness of the content dimension score of the content and the dimension preference score of the user, and taking the numerical closeness as the relation score of the content and the user.
15. The method of claim 14, quantifying the closeness of the content dimension score of online content to the user's dimension preference score using scoring relationships 1-2:
where RMSE is the user's relationship score to a content, ytScoring a user's tth preference dimension,
scoring a tth content dimension of the content.
16. The method of claim 9, further comprising:
sequencing the plurality of online contents according to the content relation score of each content;
generating push information according to the one or more contents which are sequenced at the front; and
and pushing the pushing information to an application client.
17. The method of claim 16, the push information comprising at least content recommendation information and a content link.
18. A server-side content recommendation system based on vector scoring, comprising:
a content score providing module configured to provide a list of content dimension scores in response to a request from an application client; and
a content providing module configured to send content requested by a content request to an application client in response to the content request from the application client.
19. The system of claim 18, further comprising a content score construction module configured to construct one or more content dimension scores for a plurality of online content.
20. The system of claim 19, the content score construction module further configured to comprise:
a dimension construction unit configured to construct a plurality of content dimensions for scoring;
a content analysis unit configured to perform content analysis on the online content based on the constructed content dimension; and
a content scoring unit configured to determine each content dimension score of the online content according to the content analysis result and a scoring standard of a corresponding content dimension.
21. The system of claim 19 or 20, the content score construction module connected to a database configured to store constructed content dimension scores into the database.
22. The system of claim 21, the content score providing module is connected to a database and configured to obtain online content and its content dimension scores from the database.
23. The system of claim 22, the content score provision module, upon receiving a request from an application client, provides a list of content dimension scores to the application client; or the content scoring providing module pushes the content dimension scoring list to the application client.
24. A server content recommendation method based on vector scoring comprises the following steps:
receiving a request from an application client, and providing a content dimension scoring list to the application client; and
the method comprises the steps of receiving a content request from an application client, and sending content requested by the content request to the application client.
25. The method of claim 24, further comprising: one or more content dimension scores are constructed for the online content.
26. The method of claim 25, further comprising:
constructing one or more content dimensions from the online content categories;
performing content analysis on the online content based on the constructed content dimension; and
and determining the score of each content dimension of the online content according to the content analysis result and the score standard of the corresponding content dimension.
27. The method of claim 25 or 26, further comprising: and storing the constructed content dimension scores into a database.
28. The method of claim 27, further comprising:
obtaining the content dimension scoring list from a database; and
sending the content dimension score list to an application client in response to a request of the application client; or, the content dimension score list is actively pushed to the application client.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010201636.8A CN111586089A (en) | 2020-03-20 | 2020-03-20 | Client-side and server-side content recommendation system and method based on vector scoring |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010201636.8A CN111586089A (en) | 2020-03-20 | 2020-03-20 | Client-side and server-side content recommendation system and method based on vector scoring |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111586089A true CN111586089A (en) | 2020-08-25 |
Family
ID=72119047
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010201636.8A Pending CN111586089A (en) | 2020-03-20 | 2020-03-20 | Client-side and server-side content recommendation system and method based on vector scoring |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111586089A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113434779A (en) * | 2021-07-22 | 2021-09-24 | 咪咕数字传媒有限公司 | Interactive reading method and device capable of intelligent recommendation, computing equipment and storage medium |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080140644A1 (en) * | 2006-11-08 | 2008-06-12 | Seeqpod, Inc. | Matching and recommending relevant videos and media to individual search engine results |
| US20090210246A1 (en) * | 2002-08-19 | 2009-08-20 | Choicestream, Inc. | Statistical personalized recommendation system |
| CN102004785A (en) * | 2010-11-30 | 2011-04-06 | 北京瑞信在线***技术有限公司 | Method and device for recommending music |
| CN103186527A (en) * | 2011-12-27 | 2013-07-03 | 北京百度网讯科技有限公司 | System for building music classification model, system for recommending music and corresponding method |
| CN103886090A (en) * | 2014-03-31 | 2014-06-25 | 北京搜狗科技发展有限公司 | Content recommendation method and device based on user favorites |
| US9270778B1 (en) * | 2007-07-05 | 2016-02-23 | Invent.Ly, Llc | Matching users based on behavior in a social network |
| CN106469199A (en) * | 2016-08-31 | 2017-03-01 | 福建天晴数码有限公司 | Method and system recommended by a kind of interactive mode picture |
| US20170124590A1 (en) * | 2015-10-29 | 2017-05-04 | Flipboard, Inc. | Digital magazine recommendations by topic |
| CN107492008A (en) * | 2017-08-09 | 2017-12-19 | 阿里巴巴集团控股有限公司 | Information recommendation method, device, server and computer-readable storage medium |
| CN107943871A (en) * | 2017-11-10 | 2018-04-20 | 深圳市华阅文化传媒有限公司 | Recommended user reads the method and device of the books of hobby |
| US20180268015A1 (en) * | 2015-09-02 | 2018-09-20 | Sasha Sugaberry | Method and apparatus for locating errors in documents via database queries, similarity-based information retrieval and modeling the errors for error resolution |
| CN109325818A (en) * | 2018-08-14 | 2019-02-12 | 中国平安人寿保险股份有限公司 | A kind of Products Show method, computer readable storage medium and terminal device |
| CN110096618A (en) * | 2019-05-10 | 2019-08-06 | 北京友普信息技术有限公司 | A kind of film recommended method based on fractional dimension sentiment analysis |
| CN110781377A (en) * | 2019-09-03 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Article recommendation method and device |
-
2020
- 2020-03-20 CN CN202010201636.8A patent/CN111586089A/en active Pending
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090210246A1 (en) * | 2002-08-19 | 2009-08-20 | Choicestream, Inc. | Statistical personalized recommendation system |
| US20080140644A1 (en) * | 2006-11-08 | 2008-06-12 | Seeqpod, Inc. | Matching and recommending relevant videos and media to individual search engine results |
| US9270778B1 (en) * | 2007-07-05 | 2016-02-23 | Invent.Ly, Llc | Matching users based on behavior in a social network |
| CN102004785A (en) * | 2010-11-30 | 2011-04-06 | 北京瑞信在线***技术有限公司 | Method and device for recommending music |
| CN103186527A (en) * | 2011-12-27 | 2013-07-03 | 北京百度网讯科技有限公司 | System for building music classification model, system for recommending music and corresponding method |
| CN103886090A (en) * | 2014-03-31 | 2014-06-25 | 北京搜狗科技发展有限公司 | Content recommendation method and device based on user favorites |
| US20180268015A1 (en) * | 2015-09-02 | 2018-09-20 | Sasha Sugaberry | Method and apparatus for locating errors in documents via database queries, similarity-based information retrieval and modeling the errors for error resolution |
| US20170124590A1 (en) * | 2015-10-29 | 2017-05-04 | Flipboard, Inc. | Digital magazine recommendations by topic |
| CN106469199A (en) * | 2016-08-31 | 2017-03-01 | 福建天晴数码有限公司 | Method and system recommended by a kind of interactive mode picture |
| CN107492008A (en) * | 2017-08-09 | 2017-12-19 | 阿里巴巴集团控股有限公司 | Information recommendation method, device, server and computer-readable storage medium |
| CN107943871A (en) * | 2017-11-10 | 2018-04-20 | 深圳市华阅文化传媒有限公司 | Recommended user reads the method and device of the books of hobby |
| CN109325818A (en) * | 2018-08-14 | 2019-02-12 | 中国平安人寿保险股份有限公司 | A kind of Products Show method, computer readable storage medium and terminal device |
| CN110096618A (en) * | 2019-05-10 | 2019-08-06 | 北京友普信息技术有限公司 | A kind of film recommended method based on fractional dimension sentiment analysis |
| CN110781377A (en) * | 2019-09-03 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Article recommendation method and device |
Non-Patent Citations (3)
| Title |
|---|
| 张晓琳等: "杰卡德相似系数在推荐***中的应用", 《计算机技术与发展》 * |
| 李改等: "一种新的社会化相似度计算方法", 《计算机应用研究》 * |
| 王中友等: "基于用户喜好的个性推荐***优化", 《杭州电子科技大学学报(自然科学版)》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113434779A (en) * | 2021-07-22 | 2021-09-24 | 咪咕数字传媒有限公司 | Interactive reading method and device capable of intelligent recommendation, computing equipment and storage medium |
| CN113434779B (en) * | 2021-07-22 | 2023-01-03 | 咪咕数字传媒有限公司 | Interactive reading method and device capable of intelligent recommendation, computing equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP5386663B1 (en) | Information processing apparatus, information processing method, information processing program, and recording medium | |
| USRE41899E1 (en) | System for ranking the relevance of information objects accessed by computer users | |
| JP5431727B2 (en) | Relevance determination method, information collection method, object organization method, and search system | |
| US20130185294A1 (en) | Recommender system, recommendation method, and program | |
| US20100306249A1 (en) | Social network systems and methods | |
| US20090077065A1 (en) | Method and system for information searching based on user interest awareness | |
| EP2353103A2 (en) | Method and system for determining topical relatedness of domain names | |
| JP2011227922A (en) | Personalization for sequencing of braced content within search result | |
| JP2000348041A (en) | Document retrieval method, device therefor and mechanically readable recording medium | |
| EP2778959A1 (en) | Information processing device, information processing method, and information processing program | |
| CN102037464A (en) | Search results with most clicked next objects | |
| KR20090033989A (en) | Method for advertising local information based on location information and system for executing the method | |
| CN112307366B (en) | Information display method and device and computer storage medium | |
| JP2013206407A (en) | Information recommendation method, device and program | |
| JP2014006757A (en) | Content distribution device | |
| CN112579854A (en) | Information processing method, device, equipment and storage medium | |
| US9600560B2 (en) | Search keyword and category association apparatus, search keyword and category association method, search keyword and category association program and recording medium | |
| CN114820123A (en) | Group purchase commodity recommendation method, device, equipment and storage medium | |
| JP6434954B2 (en) | Information processing apparatus, information processing method, and program | |
| CN110008396B (en) | Object information pushing method, device, equipment and computer readable storage medium | |
| CN114201680A (en) | Method for recommending marketing product content to user | |
| CN111586089A (en) | Client-side and server-side content recommendation system and method based on vector scoring | |
| CN109299368B (en) | Method and system for intelligent and personalized recommendation of environmental information resources AI | |
| JP6433270B2 (en) | Content search result providing system and content search result providing method | |
| JP2017079080A (en) | Information processing apparatus, information processing method, and information processing program |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200825 |
|
| WD01 | Invention patent application deemed withdrawn after publication |