User news recommendation method combined with story line, terminal equipment and medium

Abstract

The invention relates to a user news recommendation method in combination with a story line, terminal equipment and a medium, and the method comprises the steps: receiving a real-time news set and a user historical click news set, and carrying out the category labeling and feature extraction; after news in the historical click news set is divided according to the time intervals, clustering processing is conducted on division results according to categories, and a second news subset corresponding to each category in each time interval is obtained; calculating a feature vector of each second news subset; for each category, calculating a first similarity between corresponding different time intervals, and further constructing a directed graph representing a user story line; according to the feature vector of each node in the directed graph, calculating the category feature of each category; and selecting the news for recommendation according to the magnitude of a second similarity between the category feature and the news feature of the news of the corresponding category in the real-time news set. According to the method and the device, news meeting the previous browsing story line can be recommended to the user, and the satisfaction of the user can be improved.

Description

Technical Field

[0001] The present invention relates to the field of news recommendation, and in particular to a user news recommendation method combined with a story line, a terminal device and a medium. Background Art

[0002] Nowadays, various news events are constantly generated and spread on news websites, and it is difficult for the public to effectively read and understand such a large amount of information. Therefore, how to deeply mine and analyze fragmented data to form a user's story line path has become an important issue facing network users.

[0003] The goal of user storyline generation is to identify events that users are interested in from historical reading, construct nodes according to weights, form storylines and branches, and finally assemble them into an easy-to-understand time-based event evolution structure. Storyline generation faces two main challenges: (1) How to construct story branches that are highly relevant to user interests. (2) How to dynamically adapt to changes in user interests so as to timely recommend news stories that may be of interest to users in the future. However, in real life, users' interests are often more diverse and dynamic. Users' focus on events may change at different times and in different situations. This complexity makes it difficult for existing recommendation methods to effectively reflect users' real interests and needs if they only rely on timelines to connect events. This limitation makes it difficult for users to obtain highly relevant and coherent information when receiving news recommendations, which affects their understanding and participation in events. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes a user news recommendation method, terminal device and medium combined with a story line.

[0005] The specific plan is as follows:

[0006] A user news recommendation method combined with a story line comprises the following steps:

[0007] S1: receiving the real-time news set corresponding to the current time period and the historical click news set corresponding to the user to be recommended, and performing category labeling and feature extraction on each news in the real-time news set and the historical click news set to obtain news features of each news;

[0008] S2: setting a time range based on the release time of each news in the historical click news set, and dividing the time range into multiple time intervals;

[0009] S3: Correspond each news in the historical click news set to each time interval to obtain the first news subset corresponding to each time interval;

[0010] S4: clustering the first news subset corresponding to each time interval according to the category to obtain the second news subset corresponding to each category in each time interval;

[0011] S5: calculating a feature vector of each second news subset according to news features of each news in each second news subset;

[0012] S6: for each category, calculating a first similarity between feature vectors of any two second news subsets corresponding to different time intervals;

[0013] S7: constructing a directed graph representing the user story line corresponding to each category according to the magnitude relationship between the first similarity and the first similarity threshold; in the directed graph, the second news subsets included in the category are nodes, different nodes correspond to different time intervals, and an edge between two nodes indicates that the first similarity between the second news subsets corresponding to the two nodes is greater than the first similarity threshold, and the direction of the edge is from the earlier time interval to the later time interval;

[0014] S8: Calculate the category feature corresponding to each category according to the feature vector of the second news subset corresponding to each node in the directed graph corresponding to each category;

[0015] S9: Selecting news from the real-time news set for recommendation to the user to be recommended according to the second similarity between the category feature and the news feature of the news of the corresponding category in the real-time news set.

[0016] Furthermore, the current time period is within one day from the current moment.

[0017] Furthermore, the time frame is one year.

[0018] Furthermore, when dividing into multiple time intervals, the time intervals are divided into equal time intervals.

[0019] Furthermore, the method for extracting news features is: first extract the title of each news, then perform word segmentation and remove redundant data in the news title to obtain keywords, and finally calculate the news features of the news based on all the keywords.

[0020] Furthermore, the feature vector of the second news subset is an average value of news features of all the news contained therein.

[0021] Furthermore, a method of weighted summation of feature vectors is adopted when calculating category features, and the closer the time interval corresponding to the second news subset is to the current time, the greater the corresponding weight is.

[0022] A user news recommendation terminal device combined with a story line includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned method of the embodiment of the present invention are implemented.

[0023] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method described above in an embodiment of the present invention are implemented.

[0024] The present invention adopts the above technical solution, and through the association relationship between news and users, it can recommend news that matches the previously browsed story line to users, which helps to improve user satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 Shown is a flow chart of a method according to a first embodiment of the present invention.

[0026] Figure 2 Shown is a schematic diagram of a directed graph representing a user story line in this embodiment. DETAILED DESCRIPTION

[0027] To further illustrate various embodiments, the present invention provides drawings. These drawings are part of the disclosure of the present invention, which are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in this field should be able to understand other possible implementations and advantages of the present invention.

[0028] The present invention will now be further described with reference to the accompanying drawings and specific implementation methods.

[0029] Embodiment 1:

[0030] The embodiment of the present invention provides a user news recommendation method combined with a story line, such as Figure 1 As shown, the method comprises the following steps:

[0031] S1: receiving the real-time news set corresponding to the current time period and the historical click news set corresponding to the user to be recommended, and performing category labeling and feature extraction on each news in the real-time news set and the historical click news set to obtain news features of each news.

[0032] The current time period may be a time period close to the current time to ensure the real-time nature of the recommended news, such as within one day from the current time. The real-time news set corresponding to the current time period may be automatically collected by the crawler, which is all the news that can be collected.

[0033] The historical clicked news set corresponding to the user to be recommended is all the news clicked (browsed) by the user within a historical time range (such as one year).

[0034] The division of news categories can be set by technicians in this field according to their needs. It can be divided into major categories, such as entertainment, sports, society, etc., or divided into minor categories, such as dividing sports by event types, dividing by sports types, etc., and there is no limitation here.

[0035] Each piece of news needs to have its corresponding news features extracted. In the extraction of news features, those skilled in the art can implement it according to existing models and algorithms. In this embodiment, the title of each piece of news is first extracted, then word segmentation is performed and redundant data in the news title is removed to obtain keywords, and finally the news features of the news are calculated based on all the keywords.

[0036] S2: setting a time range based on the release time of each news in the historical click news set, and dividing the time range into multiple time intervals.

[0037] The historical clicked news set collected in this embodiment is the news clicked by the user within one year before the current time, so the time range is one year.

[0038] In the segmentation of time intervals, it is usually segmented at fixed time intervals, such as each time interval is one month, that is, the time range of one year is divided into 12 equal parts.

[0039] S3: Correspond each news in the historical click news set to each time interval to obtain a first news subset corresponding to each time interval.

[0040] Through step S3, the news in the same time interval in the historical click news set can be classified into a first news subset. For example, if 12 time intervals are included, 12 first news subsets are obtained.

[0041] S4: clustering the first news subset corresponding to each time interval according to categories to obtain second news subsets corresponding to each category in each time interval.

[0042] Through clustering, news belonging to the same category in each first news subset can be classified into a second news subset. If the total number of categories is 6, each first news subset is split into 6 second news subsets. It should be noted that if there is no news in a category, the second news subset corresponding to the category is empty.

[0043] Clustering processing can be performed using existing algorithms, such as the KMeans clustering algorithm, and is not limited here.

[0044] S5: Calculate a feature vector of each second news subset according to the news features of each news in each second news subset.

[0045] In this embodiment, the feature vector of the second news subset is set to be the average value of the news features of all the news included in it. In other embodiments, those skilled in the art can set other calculation methods of the feature vector, which is not limited here.

[0046] S6: For each category, calculate the first similarity between the feature vectors of any two second news subsets corresponding to different time intervals.

[0047] The first similarity calculation algorithm may adopt an existing algorithm, which is not limited here.

[0048] It should be noted that for each category, there may be some time intervals without news of this category, so only the time intervals with this category are considered.

[0049] S7: Based on the relationship between the first similarity and the first similarity threshold, construct a directed graph representing the user story line corresponding to each category; in the directed graph, the second news subsets included in the category are nodes, different nodes correspond to different time intervals, and the edge between two nodes indicates that the first similarity between the second news subsets corresponding to the two nodes is greater than the first similarity threshold, and the direction of the edge is from the earlier time interval to the later time interval.

[0050] For example, for a certain category, there are only 10 non-empty second news subsets corresponding to this category in 12 time intervals. Therefore, the first similarity between the two of these 10 second news subsets is calculated, and there are a total of C 1 2 0 = 45 first similarities. If 15 of the 45 first similarities calculated are greater than the first similarity threshold, then the directed graph corresponding to the category only contains the nodes corresponding to the 15 first similarities. If the first similarity between the second news subsets corresponding to two nodes is greater than the first similarity threshold, then an edge is drawn between the two nodes, and the direction of the edge is from the earlier time interval to the later time interval, such as from the node corresponding to January to the node corresponding to March. The directed graph drawn in this embodiment is as follows: Figure 2 shown.

[0051] S8: Calculate the category feature corresponding to each category according to the feature vector of the second news subset corresponding to each node in the directed graph corresponding to each category.

[0052] This embodiment adopts a method of weighted summation of feature vectors when calculating category features, and sets a corresponding weight that is greater when the time interval corresponding to the second news subset is closer to the current time.

[0053] S9: Selecting news from the real-time news set for recommendation to the user to be recommended according to the second similarity between the category feature and the news feature of the news of the corresponding category in the real-time news set.

[0054] The algorithm for calculating the second similarity may be the same as that for calculating the first similarity.

[0055] In this embodiment, according to the magnitude of the second similarity, the news with the largest second similarity is selected and recommended to the user. In addition, since there are multiple categories, each category has its corresponding maximum second similarity. When recommending, the corresponding news can be recommended to the user in descending order of the maximum second similarities corresponding to different categories.

[0056] The embodiment of the present invention can recommend news that matches a previously browsed story line to the user through the association relationship between the news and the user, which helps to improve the user's satisfaction.

[0057] Embodiment 2:

[0058] The present invention also provides a user news recommendation terminal device combined with a story line, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in the above-mentioned method embodiment of the first embodiment of the present invention when executing the computer program.

[0059] Further, as an executable solution, the user news recommendation terminal device combined with the story line can be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The user news recommendation terminal device combined with the story line may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the composition structure of the user news recommendation terminal device combined with the story line is only an example of the user news recommendation terminal device combined with the story line, and does not constitute a limitation on the user news recommendation terminal device combined with the story line. It may include more or less components than the above, or a combination of certain components, or different components. For example, the user news recommendation terminal device combined with the story line may also include input and output devices, network access devices, buses, etc., and the embodiments of the present invention do not limit this.

[0060] Further, as an executable solution, the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the user news recommendation terminal device combined with the story line, and uses various interfaces and lines to connect the various parts of the user news recommendation terminal device combined with the story line.

[0061] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the user news recommendation terminal device combined with the story line by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system and at least one application required for a function; the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0062] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method in the embodiment of the present invention are implemented.

[0063] If the module / unit integrated in the user news recommendation terminal device combined with the story line is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory) and software distribution medium, etc.

[0064] Although the present invention has been specifically shown and described in conjunction with the preferred embodiments, it should be understood by those skilled in the art that various changes may be made to the present invention in form and details without departing from the spirit and scope of the present invention as defined by the appended claims, all of which are within the scope of protection of the present invention.

Claims

1. A user news recommendation method combined with a story line, characterized in that: The following steps are involved: S1: receiving the real-time news set corresponding to the current time period and the historical click news set corresponding to the user to be recommended, and performing category labeling and feature extraction on each news in the real-time news set and the historical click news set to obtain news features of each news; S2: setting a time range based on the release time of each news in the historical click news set, and dividing the time range into multiple time intervals; S3: Correspond each news in the historical click news set to each time interval to obtain the first news subset corresponding to each time interval; S4: clustering the first news subset corresponding to each time interval according to the category to obtain the second news subset corresponding to each category in each time interval; S5: calculating a feature vector of each second news subset according to news features of each news in each second news subset; S6: for each category, calculating a first similarity between feature vectors of any two second news subsets corresponding to different time intervals; S7: constructing a directed graph representing the user story line corresponding to each category according to the magnitude relationship between the first similarity and the first similarity threshold; in the directed graph, the second news subsets included in the category are nodes, different nodes correspond to different time intervals, and an edge between two nodes indicates that the first similarity between the second news subsets corresponding to the two nodes is greater than the first similarity threshold, and the direction of the edge is from the earlier time interval to the later time interval; S8: Calculate the category feature corresponding to each category according to the feature vector of the second news subset corresponding to each node in the directed graph corresponding to each category; S9: Selecting news from the real-time news set for recommendation to the user to be recommended according to the second similarity between the category feature and the news feature of the news of the corresponding category in the real-time news set.

2. The user news recommendation method combined with story lines according to claim 1 is characterized by: The current time period is within one day from the current time.

3. The user news recommendation method combined with story lines according to claim 1 is characterized by: The time frame is one year.

4. The user news recommendation method combined with story lines according to claim 1 is characterized by: When splitting into multiple time intervals, split them at equal time intervals.

5. The user news recommendation method combined with story lines according to claim 1, characterized in that: The method of news feature extraction is: first extract the title of each news, then perform word segmentation and remove redundant data in the news title to obtain keywords, and finally calculate the news features of the news based on all keywords.

6. The user news recommendation method combined with story lines according to claim 1 is characterized by: The feature vector of the second news subset is the average value of the news features of all the news it contains.

7. The user news recommendation method combined with story lines according to claim 1 is characterized by: When calculating the category features, a method of weighted summation of feature vectors is adopted, and the closer the time interval corresponding to the second news subset is to the current time, the greater the corresponding weight is.

8. A user news recommendation terminal device combined with a story line, characterized by: The method comprises a processor, a memory and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 7 when executing the computer program.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images