A method for pushing information information, related devices, equipment and storage medium

Abstract

The application discloses a kind of information information push method, related device, equipment and storage medium.The method of the application includes: in response to information viewing request, the comprehensive ordering fraction corresponding to each information content in first information content set is acquired, comprehensive ordering fraction and time decay fraction and matching decay fraction are respectively positively correlated, time decay fraction and matching decay fraction are respectively negatively correlated with the published time amount of information content, and matching decay fraction is positively correlated with the number of keyword matched to information content;According to the comprehensive ordering fraction corresponding to each information content, first information content set is sorted according to the order of comprehensive ordering fraction from high to low, and information content sequence is obtained;The information corresponding to each information content in information content sequence is acquired;Information information sequence is pushed operation.The application can consider the real-time and importance of information content, so as to improve the accuracy and efficiency of user obtaining key information.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, related apparatus, device, and storage medium for pushing information. Background Technology

[0002] Some users have a need to access information; therefore, these users need to collect information from multiple media sources, including news websites, social media, industry reports, and officially released data. Analyzing this information allows for timely understanding of industry trends, providing data support for corporate strategy formulation, risk warning, and product development, enabling rapid response to complex environmental changes.

[0003] Since manually reviewing every piece of information is too costly, information can be obtained through personalized recommendations or popularity-based recommendations. Personalized recommendation schemes typically use algorithms such as collaborative filtering to provide a unique information experience for each individual. Popularity-based recommendations prioritize information that is more frequently accessed, displaying it in a ranking system.

[0004] However, the inventors have discovered at least the following problems with current solutions: personalized recommendations are prone to becoming siloed, causing users to miss relevant information. Furthermore, prioritizing trending topics may fail to provide users with the information they need. Currently, no effective solution has been proposed to address these issues. Summary of the Invention

[0005] This application provides a method, related apparatus, device, and storage medium for pushing information, which can take into account both the real-time nature and importance of the information content, thereby improving the accuracy and efficiency of users obtaining key information.

[0006] In view of this, this application provides a method for pushing information, including:

[0007] In response to a news viewing request, the system obtains the comprehensive ranking score for each news item in the first news content set. The comprehensive ranking score is positively correlated with the time decay score and the matching decay score. The time decay score is negatively correlated with the amount of time the news item has been published. The matching decay score is negatively correlated with the amount of time the news item has been published. The matching decay score is positively correlated with the number of keywords matched by the news item.

[0008] Based on the comprehensive ranking score corresponding to each piece of information, the first set of information content is sorted in descending order of comprehensive ranking score to obtain the information content sequence;

[0009] Obtain the information information corresponding to each piece of information in the information content sequence. The information information includes metadata corresponding to the information content, which is used to describe the attribute features related to the information content.

[0010] The system pushes information sequences consisting of information information corresponding to each piece of information, where each piece of information in the information sequence has a one-to-one correspondence with the information content in the same position in the information content sequence.

[0011] This application also provides an information push device, including:

[0012] The acquisition module is used to respond to information viewing requests and acquire the comprehensive ranking score corresponding to each piece of information in the first set of information content. The comprehensive ranking score is positively correlated with the time decay score and the matching decay score. The time decay score is negatively correlated with the amount of time the information content has been published. The matching decay score is negatively correlated with the amount of time the information content has been published. The matching decay score is positively correlated with the number of keywords matched by the information content.

[0013] The sorting module is used to sort the first set of information content according to the comprehensive sorting score corresponding to each information content, in descending order of the comprehensive sorting score, to obtain the information content sequence;

[0014] The acquisition module is also used to acquire information information corresponding to each piece of information in the information content sequence. The information information includes metadata corresponding to the information content, and the metadata is used to describe the attribute features related to the information content.

[0015] The push module is used to push information information sequences consisting of information information corresponding to each information content. Each piece of information in the information information sequence has a one-to-one correspondence with the information content in the same position in the information content sequence.

[0016] In one possible design, in another implementation of another aspect of the embodiments of this application, the information push device further includes a calculation module and a determination module;

[0017] The acquisition module is also used to acquire the publication time of each piece of information in the first set of information content;

[0018] The calculation module is used to calculate the time decay score of each piece of information in the first set of information content, based on the amount of time the information content has been published and a preset time score.

[0019] The calculation module is also used to sum up the preset matching scores corresponding to each of the at least one keyword in the keyword set for each piece of information in the first set of information content, so as to obtain the matching score of the information content if the information content matches at least one keyword in the keyword set.

[0020] The calculation module is also used to calculate the matching decay score of each piece of information in the first set of information content based on the amount of time the information content has been published and the matching score.

[0021] The determination module is used to determine the comprehensive ranking score of each piece of information in the first set of information content based on the time decay score and the matching decay score of the information content.

[0022] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0023] The acquisition module is also used to acquire the preset header part of the information content if the subject object of the information content matches any object in the header object set for each information content in the first information content set. The header object set includes the top K objects with the largest header index, where K is an integer greater than or equal to 1.

[0024] The calculation module is also used to calculate the header decay score of each piece of information in the first set of information content, based on the amount of time the information content has been published and the preset header part.

[0025] The determination module is specifically used to determine the comprehensive ranking score of each piece of information in the first set of information content based on the time decay score, matching decay score, and header decay score of the information content.

[0026] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0027] The acquisition module is specifically used to respond to information viewing requests. For each piece of information in the first set of information content, it acquires the publication time of the information content, wherein the information viewing request carries the target account identifier.

[0028] For each piece of information in the first set of information content, the time decay score of the information content is calculated based on the published time and the preset time score.

[0029] For each piece of information in the first set of information content, if the information content matches at least one keyword in the keyword set, the preset matching scores corresponding to each keyword of the at least one keyword are summed to obtain the matching score of the information content.

[0030] For each piece of information in the first set of information content, the matching decay score of the information content is calculated based on the amount of time the information content has been published and the matching score.

[0031] For each piece of information in the first set of information content, if the subject of the information content matches the target account's associated follower, the preset follower score for the information content is obtained.

[0032] For each piece of information in the first set of information content, the attention decay score of the information content is calculated based on the amount of time the information content has been published and the preset attention score;

[0033] For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, and attention decay score.

[0034] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0035] The acquisition module is also used to respond to information viewing requests. For each piece of information in the first set of information content, if the subject object of the information content matches any object in the header object set, the preset header part of the information content is acquired. The header object set includes the top K objects with the largest header index, where K is an integer greater than or equal to 1.

[0036] The calculation module is also used to calculate the header decay score of each piece of information in the first set of information content, based on the amount of time the information content has been published and the preset header part.

[0037] The acquisition module is specifically used to determine the comprehensive ranking score of each piece of information in the first set of information based on the information's time decay score, matching decay score, attention decay score, and header decay score.

[0038] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0039] The acquisition module is also used to acquire a set of candidate objects, wherein the set of candidate objects includes T objects, where T is an integer greater than or equal to K;

[0040] The determination module is also used to determine the head index of each object in the candidate object set based on at least one of the following: user activity, revenue, retention rate, and user rating score.

[0041] The determination module is also used to select the top K objects with the largest head index in the candidate object set as the head object set based on the head index of each object in the candidate object set.

[0042] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0043] The acquisition module is also used to respond to information viewing requests. For each piece of information in the first set of information content, if the industry type of the information content matches the industry type associated with the target account identifier, the preset industry score of the information content is acquired.

[0044] The determination module is also used to determine the industry decay score of each piece of information in the first set of information content based on the amount of time the information content has been published and the preset industry score.

[0045] The acquisition module is specifically used to determine the comprehensive ranking score of each piece of information in the first set of information based on the information's time decay score, matching decay score, attention decay score, top decay score, and industry decay score.

[0046] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0047] The acquisition module is specifically used to receive information viewing requests sent by the terminal;

[0048] Based on the information viewing request, obtain the comprehensive ranking score corresponding to each piece of information in the first set of information content;

[0049] The push module is specifically used to send a sequence of information information, consisting of information information corresponding to each piece of information, to the terminal so that the terminal can display the sequence of information information.

[0050] In one possible design, in another implementation of another aspect of the embodiments of this application, the information push device further includes a receiving module and a sending module;

[0051] The receiving module is used to receive a request from the terminal to view details of the target information after pushing the information information sequence consisting of information information corresponding to each information content. The target information is any one of the information information in the information information sequence.

[0052] The sending module is used to send the information content corresponding to the target information to the terminal based on the details viewing request, so that the terminal can display the information content corresponding to the target information.

[0053] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0054] The receiving module is also used to receive an information filtering request sent by the terminal after pushing the information information sequence consisting of information information corresponding to each information content, wherein the information filtering request carries at least one piece of information for constructing information filtering conditions.

[0055] The determination module is also used to determine all information that meets the information filtering conditions from the information information sequence based on the information filtering request;

[0056] The sorting module is also used to sort all the information that meets the information filtering conditions in descending order of the comprehensive sorting score, so as to obtain the information subsequence;

[0057] The sending module is also used to send information information subsequences to the terminal so that the terminal can display the information information subsequences.

[0058] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0059] The receiving module is also configured to receive a summary viewing request sent by the terminal, wherein the summary viewing request carries at least one piece of information for constructing summary generation conditions;

[0060] The acquisition module is also used to acquire a second set of information content that meets the summary generation conditions based on the summary viewing request;

[0061] The acquisition module is also used to generate summary information for each of the M key points based on the second information content set, where M and N are both integers greater than or equal to 1;

[0062] The acquisition module is also used to generate N original sentences corresponding to each key point based on the second set of information content and the summary information corresponding to each key point;

[0063] The sending module is also used to send the summary information corresponding to each key point and N original sentences to the terminal, so that the terminal can display the summary information corresponding to each key point and N original sentences.

[0064] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0065] The acquisition module is specifically used to segment each piece of information in the second set of information content into sentences to obtain the set of original sentences.

[0066] For each of the M key points, obtain the similarity between the summary information corresponding to the key point and each original sentence in the set of original sentences;

[0067] For each of the M key points, extract the top N original sentences from the set of original sentences that have the highest similarity to the summary information corresponding to the key point.

[0068] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0069] The receiving module is also used to receive key configuration requests sent by the terminal, wherein the key configuration requests carry M key identifiers;

[0070] The determination module is also used to determine M key points based on the key point configuration request, wherein the key points among the M key points have a one-to-one correspondence with the key point identifiers among the M key point identifiers.

[0071] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0072] The acquisition module is also used to acquire the image to be processed;

[0073] The acquisition module is also used to acquire information content based on the image to be processed through a graph-to-text model;

[0074] The acquisition module is also used to acquire the video to be processed;

[0075] The acquisition module is also used to obtain information content based on the video to be processed by generating a video description model;

[0076] The acquisition module is also used to acquire the voice to be processed;

[0077] The acquisition module is also used to acquire information content based on the speech to be processed through a speech recognition model.

[0078] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0079] The acquisition module is also used to acquire information content publishing sites;

[0080] The acquisition module is also used to add the information content to the first information content set if the publishing site of the published information content successfully matches the preset site.

[0081] In one possible design, in another implementation of another aspect of the embodiments of this application,

[0082] The acquisition module is also used to acquire a set of candidate information content;

[0083] The acquisition module is also used to add each piece of information in the candidate information content set to the first information content set if the title or summary of the information content meets the information matching conditions.

[0084] Among them, at least one of the following conditions must be met for information matching:

[0085] The title of the news content includes target keywords;

[0086] The summary of the information content includes target keywords;

[0087] The title of the news content includes seed words;

[0088] The summary of the information content includes seed words.

[0089] In another aspect, this application provides a computer device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the methods described above.

[0090] Another aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the methods described above.

[0091] Another aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the methods described above.

[0092] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0093] This application provides a method for pushing information. First, a comprehensive ranking score is obtained for each piece of information in a first set of information, where the comprehensive ranking score is positively correlated with both the time decay score and the matching decay score. Then, the first set of information is sorted from highest to lowest based on the comprehensive ranking score for each piece of information, resulting in an information content sequence. Next, information information corresponding to each piece of information in the information content sequence is obtained. Finally, an information information sequence consisting of the information information corresponding to each piece of information is pushed. Through this method, the time decay score reflects the impact of the real-time nature of the information on the ranking priority, while the matching decay score reflects the impact of the importance of the information on the ranking priority. Therefore, the comprehensive ranking score generated by the time decay score and the matching decay score can balance the real-time nature and importance of the information, thereby improving the accuracy and efficiency of users obtaining key information. Attached Figure Description

[0094] Figure 1 This is a schematic diagram illustrating the push of application-related information in an embodiment of this application;

[0095] Figure 2This is a schematic diagram illustrating the pushing of hardware product information in an embodiment of this application;

[0096] Figure 3 This is a schematic diagram illustrating the pushing of e-commerce information in an embodiment of this application;

[0097] Figure 4 This is a schematic diagram of an implementation environment for the information push method in this application embodiment;

[0098] Figure 5 This is a schematic diagram of another implementation environment for the information push method in this application;

[0099] Figure 6 This is a flowchart illustrating the information push method in an embodiment of this application;

[0100] Figure 7 This is a schematic diagram of an interface showing a sequence of information in an embodiment of this application;

[0101] Figure 8 This is a schematic diagram of an interface displaying information content in an embodiment of this application;

[0102] Figure 9 This is a schematic diagram of an interface for filtering information content in an embodiment of this application;

[0103] Figure 10 This is a schematic diagram of an interface illustrating smart summarization in an embodiment of this application;

[0104] Figure 11 This is a schematic diagram of an interface for selecting key points in an embodiment of this application;

[0105] Figure 12 This is a schematic diagram of an interaction flow of the information push method in an embodiment of this application;

[0106] Figure 13 This is a schematic diagram of an information push device in an embodiment of this application;

[0107] Figure 14 This is a schematic diagram of the structure of a computer device in an embodiment of this application. Detailed Implementation

[0108] This application provides a method, related apparatus, device, and storage medium for pushing information, which can take into account both the real-time nature and importance of the information content, thereby improving the accuracy and efficiency of users obtaining key information.

[0109] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0110] Some users have a need to access information, requiring them to collect articles from multiple media outlets daily. With thousands of articles published daily, manually reviewing each one is inefficient. Therefore, a system is needed to assist in filtering and inform users which information is more relevant. Currently, personalized or trending recommendations are commonly used to improve information access efficiency. However, these methods don't always match real-world business scenarios. For example, a user might need information about a game's download surge, but this information might not be trending or meet the criteria for personalized recommendations, leading to the user missing valuable news.

[0111] Based on this, this application provides a method for pushing information. Specifically, a comprehensive ranking score is generated based on the publication time of the information content and the matched keywords. This comprehensive ranking score decays over time, causing older information content to gradually lose popularity. Therefore, the information content is presented as an information stream in descending order of the comprehensive ranking score, thus helping users quickly obtain information.

[0112] Before introducing the specific methods of this application, the application scenarios of this application will be illustrated by example. It should be understood that the following application scenarios are merely illustrative and are not limited to these examples.

[0113] Application Scenario 1: Push notifications for application-related news and information;

[0114] By studying the app industry, users can accurately grasp market trends (e.g., the shift from social entertainment to utility-based applications) and user needs, identify competitive barriers and technological trends (e.g., the integration of virtual reality technology with mobile devices), and anticipate compliance risks. These factors are crucial for developing effective strategies.

[0115] For example, please refer to Figure 1 , Figure 1 This is a schematic diagram illustrating the push of application-related information in an embodiment of this application. As shown in the figure, the key information presented is designed to assist users in their analysis. This information originates from official platform data (e.g., app downloads, ratings, and reviews from major app stores), industry news media, data providers, and social media platforms. After clicking on the title of the information (e.g., "Application A Launches 'Blind Box Packages' - Traffic Anxiety Behind Consumer Controversy"), users can be redirected to the relevant page where the corresponding information content is displayed, thus facilitating the viewing of the complete information.

[0116] Application Scenario 2: Hardware product information;

[0117] Deep insights into the consumer electronics industry (e.g., hardware products like smartphones and tablets) are crucial for companies to maintain a leading edge in technological iteration and ecosystem competition. Industry-related information allows users to understand the global market landscape and technological evolution, helping them anticipate potential risks.

[0118] For example, please refer to Figure 2 , Figure 2 This is a schematic diagram illustrating the push of hardware product information in this embodiment of the application. The diagram shows key information to assist user analysis. This information originates from professional technology media (e.g., media outlets that publish extensive reviews, news, and industry analyses of hardware products), market research reports (e.g., regularly publishing research reports on the hardware market, providing market data and trend analysis), social media and forums, manufacturer websites, product launches, and industry exhibitions. After clicking on a title in the information (e.g., "Brand B's recycling rate for used mobile phones is less than 1%"), users can be redirected to the relevant page where the corresponding information content is displayed, thus facilitating viewing the complete information.

[0119] Application Scenario 3: Pushing e-commerce news and information;

[0120] By collecting and analyzing information from e-commerce platforms, users gain invaluable insights for businesses in developing effective growth strategies. The e-commerce industry is rapidly evolving, with new business models, technological applications, and market trends constantly emerging. Therefore, comprehensive and in-depth collection and analysis of information is crucial for accurately grasping market dynamics.

[0121] For example, please refer to Figure 3 , Figure 3This is a schematic diagram illustrating the push of e-commerce related information in this application embodiment. As shown in the figure, the key information presented is designed to assist users in their analysis. This information originates from the e-commerce platform itself (i.e., industry reports and statistical data released by the e-commerce platform, which are authoritative and targeted), professional data institutions, news media, industry websites, social media, and user review platforms, etc. After clicking on the title of the information (e.g., "The Underlying Logic of Live Streaming E-commerce's Contribution to Total Retail Sales Exceeding 12%), users can be redirected to the relevant page where the corresponding information content is displayed, thus facilitating the viewing of the complete information.

[0122] It should be noted that the above application scenarios are merely examples, and the information push method provided in this embodiment can also be applied to other scenarios, which are not limited here.

[0123] The following will combine Figure 4 and Figure 5 The implementation environment for each information push method is described below. Figure 4 This illustrates a method for pushing information in a single-machine scenario. Figure 5 The example shown illustrates a method for pushing information in a networked scenario.

[0124] I. Standalone scenario;

[0125] The method provided in this application can be applied to... Figure 4 The implementation environment shown includes a terminal 401. The terminal 401 involved in this application includes, but is not limited to, mobile phones, tablets, laptops, desktop computers, smart voice interaction devices, virtual reality devices, smart home appliances, vehicle terminals, and aircraft. The client is deployed on the terminal 401 and can run on the terminal 401 via a browser, a standalone app, or a mini-program.

[0126] In the above implementation environment, in step A1, the user triggers an information viewing request. In step A2, terminal 401 obtains the comprehensive ranking score corresponding to each piece of information in the first information content set. In step A3, terminal 401 sorts the first information content set according to the comprehensive ranking score from high to low, obtaining an information content sequence. In step A4, terminal 401 obtains the information information corresponding to each piece of information in the information content sequence. In step A5, terminal 401 pushes the information information sequence to the user. In step A6, the user can view the information information sequence.

[0127] II. Networked Scenarios;

[0128] The method provided in this application can be applied to... Figure 5The illustrated implementation environment includes a terminal 501 and a server 502, and the terminal 501 and server 502 can communicate via a network 503. The network 503 uses standard communication technologies and / or protocols, typically the Internet, but can also be any network, including but not limited to Bluetooth, a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), mobile, private networks, or any combination of virtual private networks. In some embodiments, customized or dedicated data communication technologies may be used to replace or supplement the aforementioned data communication technologies.

[0129] The server 502 involved in this application can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence (AI) platforms.

[0130] In the above implementation environment, in step B1, terminal 501 sends an information viewing request to server 502 via network 503. In step B2, server 502 obtains the comprehensive ranking score corresponding to each piece of information in the first information content set. In step B3, server 502 sorts the first information content set according to the comprehensive ranking score from high to low, obtaining an information content sequence. In step B4, server 502 obtains the information information corresponding to each piece of information in the information content sequence. In step B5, server 502 pushes the information information sequence to terminal 501 via network 503. In step B6, terminal 501 displays the information information sequence.

[0131] Based on the above introduction, the method for pushing information in this application will be described below. Please refer to [link / reference]. Figure 6 The information push method in this application embodiment can be completed independently by the server, independently by the terminal, or jointly by the terminal and the server. The method of this application includes:

[0132] S601. In response to the information viewing request, obtain the comprehensive ranking score corresponding to each piece of information in the first information content set, wherein the comprehensive ranking score is positively correlated with the time decay score, and the comprehensive ranking score is positively correlated with the matching decay score, the time decay score is negatively correlated with the amount of time the information content has been published, the matching decay score is negatively correlated with the amount of time the information content has been published, and the matching decay score is positively correlated with the number of keywords matched by the information content;

[0133] In one or more embodiments, a comprehensive ranking score needs to be calculated for each piece of information in the first information content set. One implementation is to start calculating the comprehensive ranking score for each piece of information after receiving an information viewing request. Another implementation is to pre-calculate the comprehensive ranking score for each piece of information, and then directly call the pre-calculated comprehensive ranking score after receiving an information viewing request.

[0134] Specifically, the overall ranking score includes at least a time decay score and a match decay score. A higher time decay score results in a higher overall ranking score, and similarly, a higher match decay score also results in a higher overall ranking score. Both time decay and match decay scores are negatively correlated with the publication time of the information content; that is, the time decay score decreases as the publication time increases, and correspondingly, the match decay score also decreases as the publication time increases. Therefore, information content that has been published for a longer period tends to have a lower overall ranking score and will be ranked lower. Furthermore, the match decay score is positively correlated with the number of keywords matched by the information content; that is, the more keywords included in the information content, the higher the match decay score. Considering the differences in writing styles among information content, some content may include a lot of invalid and repetitive descriptions. Therefore, when an article contains multiple identical keywords, it is usually counted as a single keyword match; that is, the number of times the same keyword appears in the article is not considered.

[0135] It is understood that the published time in this application can be counted in "hours," for example, if 30 hours have passed since the publication time, then the published time is "30 hours." The published time in this application can also be counted in "days," for example, the published time for information published today is "0 days," and the published time for information published yesterday is "1 day." In addition, other time units can be used, such as "minutes" or "weeks." For ease of explanation, this application uses "days" as an example of published time, but this should not be construed as a limitation of this application.

[0136] It should be noted that the term "in response to" in this application refers to the conditions or states upon which the execution of an operation depends, and one or more operations that can be executed when certain conditions or states are met. These operations can be real-time or have a certain delay.

[0137] S602. Based on the comprehensive ranking score corresponding to each piece of information, sort the first set of information content in descending order of comprehensive ranking score to obtain the information content sequence.

[0138] In one or more embodiments, the first set of information content is sorted in descending order of its comprehensive ranking score, thereby obtaining a sequence of information content arranged in descending order of comprehensive ranking score. A higher comprehensive ranking score indicates that the information content is more worthy of attention.

[0139] It should be noted that if at least two pieces of information have the same overall ranking score, they can be sorted according to the order of publication time (i.e., the newest information is displayed first), or according to the site affiliation (i.e., information from a specified site is displayed first), or according to the number of keywords hit (i.e., information with more keywords is displayed first), or other sorting mechanisms can be used, which are not limited here.

[0140] S603. Obtain the information information corresponding to each piece of information in the information content sequence, wherein the information information includes metadata corresponding to the information content, and the metadata is used to describe the attribute features related to the information content;

[0141] In one or more embodiments, for each piece of information, its corresponding information information can be extracted separately. The information information includes at least metadata of the information content, which describes attribute characteristics related to the information content. Furthermore, the information information may also include illustrations of the information content, etc.

[0142] Specifically, one implementation involves extracting the information information corresponding to each piece of information in the sequence after obtaining the information content sequence. Another implementation involves pre-extracting the information information corresponding to each piece of information and then directly determining the corresponding information information after obtaining the information content sequence.

[0143] It should be noted that the metadata involved in this application includes, but is not limited to, the title, author, publication time, keywords, etc. of the information content.

[0144] S604. Push the information information sequence consisting of information information corresponding to each information content, wherein each information information in the information information sequence has a one-to-one correspondence with the information content in the same position in the information content sequence.

[0145] In one or more embodiments, the information information corresponding to each piece of information is sorted according to its order in the information content sequence to obtain an information information sequence. Therefore, the information information in the information information sequence is also arranged in descending order according to the comprehensive ranking score.

[0146] Specifically, when the information sequence is generated at the terminal, it can be directly displayed to the user as an information stream. When the information sequence is generated at the server, it needs to be pushed to the terminal, and then the terminal displays the information sequence as an information stream.

[0147] This application provides a method for pushing information. Through the above method, time decay scores reflect the impact of the real-time nature of the information content on the ranking priority, while matching decay scores reflect the impact of the importance of the information content on the ranking priority. Therefore, the combined ranking score generated by time decay scores and matching decay scores can take into account both the real-time nature and importance of the information content, thereby improving the accuracy and efficiency of users obtaining key information.

[0148] Optionally, in the above Figure 6 In addition to one or more corresponding embodiments, another optional embodiment provided in this application may further include:

[0149] For each piece of information in the first set of information content, obtain the publication time of the information content;

[0150] For each piece of information in the first set of information content, the time decay score of the information content is calculated based on the published time and the preset time score.

[0151] For each piece of information in the first set of information content, if the information content matches at least one keyword in the keyword set, the preset matching scores corresponding to each keyword of the at least one keyword are summed to obtain the matching score of the information content.

[0152] For each piece of information in the first set of information content, the matching decay score of the information content is calculated based on the amount of time the information content has been published and the matching score.

[0153] For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the time decay score and the matching decay score of the information content.

[0154] In one or more embodiments, a method for calculating the comprehensive ranking score is introduced. As can be seen from the foregoing embodiments, it is necessary to calculate the comprehensive ranking score for each piece of information in the first information content set separately. Based on this, the following will use "number of days" as an example to illustrate the method.

[0155] Specifically, the configuration table includes preset time scores, preset matching scores for each keyword, and decay coefficients for each indicator. Therefore, the overall ranking score for the information content can be calculated as follows:

[0156]

[0157] Among them, S 综合 (doc,time) represents the overall ranking score of the information content. This represents the time decay of information content. ∑ 匹配词i S 匹配词i (doc) indicates the matching score of the information content. This represents the decay score for matching information content.

[0158] The time period represents the amount of time that has elapsed since the publication. For example, the current date minus the publication date means that the time period for news published today is 0, the time period for news published yesterday is 1, and so on.

[0159] doc indicates information content.

[0160] S 日期 S represents the preset time in minutes. 日期 =a, and, a>0.

[0161] k 日期 k represents the time decay coefficient. 日期 ∈(0,1).

[0162] k 匹配 k represents the matching attenuation coefficient. 匹配 ∈(0,1).

[0163] S 匹配词i b is the preset matching score for the matching word i. Different matching words i can have different preset matching scores. i ,Right now:

[0164]

[0165] The preset matching score is related to the analyst's focus; events with greater emphasis require higher preset matching scores. For example, the preset matching score for the matching word "online" is 5, and the preset matching score for the matching word "version" is 1. If the matching word "i" does not exist in the news content, the preset matching score is 0.

[0166] Secondly, this application provides a method for calculating a comprehensive ranking score. Using this method, the time decay score reflects the decay of information content over time, while the matching decay score reflects the degree of matching between the information content and the business requirements. Based on both the time decay score and the matching decay score, the comprehensive ranking score can be evaluated more comprehensively and accurately, thereby facilitating the obtaining of ranking results that better meet business needs.

[0167] Optionally, in the above Figure 6 In addition to one or more corresponding embodiments, another optional embodiment provided in this application may further include:

[0168] For each piece of information in the first set of information content, if the subject object of the information content matches any object in the set of header objects, then the preset header part of the information content is obtained. The set of header objects includes the top K objects with the largest header index, where K is an integer greater than or equal to 1.

[0169] For each piece of information in the first set of information content, the header decay score of the information content is calculated based on the amount of time the information content has been published and the preset header part.

[0170] For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score and matching decay score, specifically including:

[0171] For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, and header decay score.

[0172] In one or more embodiments, another method for calculating the comprehensive ranking score is introduced. As can be seen from the foregoing embodiments, it is necessary to calculate the comprehensive ranking score for each piece of information in the first information content set separately. Based on this, the following will use "number of days" as an example to illustrate the method.

[0173] Specifically, the configuration table includes preset time scores, preset matching scores for each keyword, preset headers, and attenuation coefficients for each metric. Therefore, the overall ranking score for the information content can be calculated as follows:

[0174]

[0175] Among them, S 综合 (doc,time) represents the overall ranking score of the information content. This represents the time decay of information content. ∑ 匹配词i S 匹配词i (doc) indicates the matching score of the information content. This represents the decay score for matching information content. This indicates the header decay score of the information content.

[0176] k 头部 k represents the head attenuation coefficient. 头部 ∈(0,1).

[0177] S 头部 (doc) is the default header. If the subject of the information content matches any object in the header object set, a default header greater than 0 will be obtained, i.e.:

[0178]

[0179] Taking the collection of information from the gaming industry as an example, the subject object can be the game name. If the subject object is "Game A" and the header object set includes "Game A", then a preset header score of 2 can be obtained. If the subject object is "Game B" and the header object set does not include "Game B", then a preset header score of 0 can be obtained.

[0180] Furthermore, this application provides another method for calculating the comprehensive ranking score. Using the above method, the time decay score reflects the decay of information content over time, the matching decay score reflects the degree of matching between information content and business needs, and the head decay score reflects the association between information content and leading entities (i.e., entities that hold a leading position in a specific field or market). Based on the time decay score, matching decay score, and head decay score, the comprehensive ranking score can be evaluated more comprehensively and accurately, thus facilitating the obtaining of ranking results that better meet business requirements.

[0181] Optionally, in the above Figure 6 Based on one or more corresponding embodiments, in another optional embodiment provided by this application, in response to an information viewing request, obtaining the comprehensive ranking score corresponding to each piece of information in the first information content set specifically includes:

[0182] In response to a news viewing request, for each news item in the first news content set, the publication time of the news item is obtained, wherein the news viewing request carries the target account identifier;

[0183] For each piece of information in the first set of information content, the time decay score of the information content is calculated based on the published time and the preset time score.

[0184] For each piece of information in the first set of information content, if the information content matches at least one keyword in the keyword set, the preset matching scores corresponding to each keyword of the at least one keyword are summed to obtain the matching score of the information content.

[0185] For each piece of information in the first set of information content, the matching decay score of the information content is calculated based on the amount of time the information content has been published and the matching score.

[0186] For each piece of information in the first set of information content, if the subject of the information content matches the target account's associated follower, the preset follower score for the information content is obtained.

[0187] For each piece of information in the first set of information content, the attention decay score of the information content is calculated based on the amount of time the information content has been published and the preset attention score;

[0188] For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, and attention decay score.

[0189] In one or more embodiments, another method for calculating the comprehensive ranking score is introduced. As can be seen from the foregoing embodiments, it is necessary to calculate the comprehensive ranking score for each piece of information in the first information content set separately. Based on this, the following will use "number of days" as an example to illustrate the method.

[0190] Specifically, the configuration table includes preset time scores, preset matching scores for each keyword, preset attention scores, and decay coefficients for each metric. Therefore, the overall ranking score for information content can be calculated as follows:

[0191]

[0192] Among them, S 综合 (doc,time) represents the overall ranking score of the information content. This represents the time decay of information content. ∑ 匹配词i S 匹配词i (doc) indicates the matching score of the information content. This represents the decay score for matching information content. This indicates the decrease in attention given to the information content.

[0193] k 关注 This indicates that the attenuation coefficient, k, is of interest. 关注 ∈(0,1).

[0194] S 关注(doc) represents the preset follower score. If the subject of the information content matches the follower of the target account, a preset follower score greater than 0 will be awarded.

[0195]

[0196] Taking the collection of information from the gaming industry as an example, the subject can be a game name. If the subject is "Game A" and the target account's associated followers include "Game A", then a preset follower score of 2 can be obtained. If the subject is "Game B" and the target account's associated followers do not include "Game B", then a preset follower score of 0 can be obtained.

[0197] Since each user needs to follow different objects, it is necessary to determine specific objects of interest based on the target account identifier (e.g., username, registered email, mobile phone number). For ease of understanding, please refer to Table 1, which is an example of collecting information content from the gaming industry. Table 1 provides an illustration of the object of interest list for different accounts.

[0198] Table 1

[0199] Account ID Target of attention 11223344 Game A 22334455 Game A, Game B, Game D 33445566 Game B, Game C 44556677 Game A, Game B, Game C, Game D

[0200] Therefore, after receiving an information viewing request, the target account identifier carried in the request can be used to determine the corresponding target account. For example, if the target account identifier is "11223344", then the target account is game A.

[0201] Furthermore, this application provides another method for calculating the comprehensive ranking score. Using the above method, the time decay score reflects the decay of information content over time, the matching decay score reflects the degree of matching between information content and business needs, and the attention decay score reflects the association between information content and the user's followed objects. Based on the time decay score, matching decay score, and attention decay score, the comprehensive ranking score can be evaluated more comprehensively and accurately, thereby facilitating the obtaining of ranking results that better meet business requirements.

[0202] Optionally, in the above Figure 6 In addition to one or more corresponding embodiments, another optional embodiment provided in this application may further include:

[0203] In response to a news viewing request, for each news content in the first news content set, if the subject object of the news content matches any object in the header object set, then the preset header part of the news content is obtained. The header object set includes the top K objects with the largest header index, where K is an integer greater than or equal to 1.

[0204] For each piece of information in the first set of information content, the header decay score of the information content is calculated based on the amount of time the information content has been published and the preset header part.

[0205] For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, and attention decay score, specifically including:

[0206] For each piece of information in the first set of information, a comprehensive ranking score is determined based on the information's time decay score, matching decay score, attention decay score, and head decay score.

[0207] In one or more embodiments, another method for calculating the comprehensive ranking score is introduced. As can be seen from the foregoing embodiments, it is necessary to calculate the comprehensive ranking score for each piece of information in the first information content set separately. Based on this, the following will use "number of days" as an example to illustrate the method.

[0208] Specifically, the configuration table includes preset time scores, preset matching scores for each keyword, preset headers, preset attention scores, and decay coefficients for each metric. Therefore, the overall ranking score for the information content can be calculated as follows:

[0209]

[0210] Among them, S 综合 (doc,time) represents the overall ranking score of the information content. This represents the time decay of information content. ∑ 匹配词i S 匹配词i (doc) indicates the matching score of the information content. This represents the decay score for matching information content. This indicates the decrease in attention given to the information content. This represents the header decay score of the information content. Max(·) indicates taking the maximum value.

[0211] Formula (7) can be used to obtain the comprehensive ranking score of an article's content within a date interval of time. Since S 日期 The value is a positive constant. Therefore, even if the news content does not contain any matching words, does not belong to the header object set, and is not an object followed by users, the news content will still have a positive overall ranking score that gradually decreases with date. The matching score is the sum of the preset matching scores of multiple matching words. If a piece of news content is associated with multiple matching words, its matching score will be larger. The preset attention score and preset header part are actually scores for the news content's affiliation. To avoid double-counting and causing this factor to have too large a weight, the maximum value is used for calculation.

[0212] Furthermore, this application provides another method for calculating the comprehensive ranking score. Using the above method, the time decay score reflects the decay of information content over time, the matching decay score reflects the degree of matching between information content and business needs, the attention decay score reflects the association between information content and the user's followed objects, and the header decay score reflects the association between information content and header objects. Based on the time decay score, matching decay score, attention decay score, and header decay score, the comprehensive ranking score can be evaluated more comprehensively and accurately, thus facilitating the obtaining of ranking results that better meet business needs.

[0213] Optionally, in the above Figure 6 In addition to one or more corresponding embodiments, another optional embodiment provided in this application may further include:

[0214] Obtain a set of candidate objects, where the set of candidate objects includes T objects, and T is an integer greater than or equal to K;

[0215] For each object in the candidate object set, determine the object's head index based on at least one of the following: user activity, revenue, retention rate, and user rating score.

[0216] Based on the head index of each object in the candidate object set, the top K objects with the largest head index in the candidate object set are selected as the head object set.

[0217] In one or more embodiments, a method for calculating the head index is described. As can be seen from the foregoing embodiments, the head object set includes K objects that belong to the head objects. In practical applications, the head index of an object can be used to determine whether the object belongs to the head objects.

[0218] Specifically, taking the collection of information content from the gaming industry as an example, let's assume the object is "games". Based on this, for each object (e.g., games) in the candidate object set, the object's head index can be calculated as follows:

[0219] S head_i =A_i×w1+B_i×w2+C_i×w3+D_i×w4; Formula (8)

[0220] Among them, S head_i Let A_i represent the head index of the i-th object. Let A_i represent the user activity level of the i-th object. Let B_i represent the revenue of the i-th object. Let C_i represent the retention rate of the i-th object. Let D_i represent the user rating score of the i-th object. Let w1 represent the first weight value, for example, 0.4. Let w2 represent the second weight value, for example, 0.3. Let w3 represent the third weight value, for example, 0.2. Let w4 represent the fourth weight value, for example, 0.1.

[0221] For example, taking the i-th object as game A, user activity can be the ratio of game A's average monthly user activity to the average of the top 10 games in the industry in terms of average monthly user activity. Revenue can be the ratio of game A's average monthly revenue to the average of the top 10 games in the industry in terms of average monthly revenue. Retention rate can be the ratio of game A's retention rate to the average of the top 10 games in the industry in terms of retention rate. User rating score can be the ratio of game A's user rating score to the maximum rating score.

[0222] For easier understanding, please refer to Table 2, which is a schematic diagram of the head index corresponding to the candidate object set.

[0223] Table 2

[0224] Candidate object set Head Index Game A 0.98 Game B 0.88 Game C 0.95 Game D 0.68 Game E 0.44

[0225] Based on this, taking K as 3 as an example, "Game A", "Game C" and "Game B" are used as the head object set.

[0226] Furthermore, this application provides a method for calculating the head index. Using this method, the number of active users reflects the user base, revenue reflects commercial value, retention rate reflects user stickiness, and user rating score represents user approval. The weighted head index can comprehensively and objectively assess the importance of an object.

[0227] Optionally, in the above Figure 6 In addition to one or more corresponding embodiments, another optional embodiment provided in this application may further include:

[0228] In response to a news viewing request, for each news item in the first news content set, if the industry type of the news item matches the industry type associated with the target account identifier, the preset industry score of the news item is obtained.

[0229] For each piece of information in the first set of information content, determine the industry decay score of the information content based on the amount of time the information content has been published and the preset industry score;

[0230] For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, and attention decay score, specifically including:

[0231] For each piece of information in the first set of information, a comprehensive ranking score is determined based on the information's time decay score, matching decay score, attention decay score, top-ranking decay score, and industry decay score.

[0232] In one or more embodiments, another method for calculating the comprehensive ranking score is introduced. As can be seen from the foregoing embodiments, it is necessary to calculate the comprehensive ranking score for each piece of information in the first information content set separately. Based on this, the following will use "number of days" as an example to illustrate the method.

[0233] Specifically, the configuration table includes preset time scores, preset matching scores for each keyword, preset header scores, preset attention scores, preset industry scores, and decay coefficients for each indicator. Therefore, the comprehensive ranking score for information content can be calculated as follows:

[0234]

[0235] Among them, S 综合 (doc,time) represents the overall ranking score of the information content. This represents the time decay of information content. This represents the industry attenuation score of information content. ∑ 匹配词i S 匹配词i (doc) indicates the matching score of the information content. This represents the decay score for matching information content. This indicates the decrease in attention given to the information content. This represents the header decay score of the information content. Max(·) indicates taking the maximum value.

[0236] k 行业 k represents the industry depreciation coefficient. 行业 ∈(0,1).

[0237] S 行业 (doc) represents the preset industry score. If the industry type of the information content matches the industry type associated with the target account identifier, a preset industry score greater than 0 will be obtained.

[0238]

[0239] Specifically, if the industry type of the information content is "gaming industry" and the industry type associated with the target account identifier is also "gaming industry", then a preset industry score of 2 can be obtained. If the industry type of the information content is "gaming industry" and the industry type associated with the target account identifier is "e-commerce industry", then a preset industry score of 0 can be obtained.

[0240] Furthermore, this application embodiment provides another method for calculating the comprehensive ranking score. Using the above method, the time decay score reflects the decay of information content over time; the matching decay score reflects the degree of matching between information content and business; the attention decay score reflects the association between information content and users' followed objects; the head decay score reflects the association between information content and head objects; and the industry decay score reflects the relevance of information content to a specific industry. Based on the time decay score, matching decay score, attention decay score, head decay score, and industry decay score, the comprehensive ranking score can be evaluated more comprehensively and accurately, thereby facilitating the obtaining of ranking results that better meet business needs.

[0241] Optionally, in the above Figure 6 Based on one or more corresponding embodiments, in another optional embodiment provided by this application, in response to an information viewing request, obtaining the comprehensive ranking score corresponding to each piece of information in the first information content set specifically includes:

[0242] Receive information viewing requests sent by the terminal;

[0243] Based on the information viewing request, obtain the comprehensive ranking score corresponding to each piece of information in the first set of information content;

[0244] The push operation involves pushing a sequence of information messages, each consisting of information messages corresponding to a specific message content.

[0245] Send a sequence of information information, consisting of information information corresponding to each piece of information, to the terminal so that the terminal can display the sequence of information information.

[0246] In one or more embodiments, a method for displaying information sequences in the form of an information stream is described. As can be seen from the foregoing embodiments, after generating the information sequence, it can be displayed through an enterprise's business information management platform (BIP). Users can access the information sequence with authorization, thereby assisting them in quickly obtaining key intelligence.

[0247] Specifically, for ease of understanding, please refer to Figure 7 , Figure 7 This is a schematic diagram of an interface showing an information information sequence in an embodiment of this application. The diagram shows an information information sequence displayed in the form of an information flow (showing four pieces of information information in the information information sequence). Taking one piece of information content as an example, the information information corresponding to the information content includes the title indicated by 701, the publishing site indicated by 702, the author indicated by 703, the keywords indicated by 704, and the publishing time indicated by 705.

[0248] It should be noted that the page usually displays news information formed by the daily increase in news content by default, and the news information that appears at the top is the news content with the higher comprehensive ranking score.

[0249] Secondly, this application provides a method for displaying information sequences in the form of an information stream. This method allows users to quickly locate important information without having to sift through a large amount of irrelevant or secondary information, thereby improving information acquisition efficiency.

[0250] Optionally, in the above Figure 6 Based on one or more corresponding embodiments, in another optional embodiment provided by this application, after pushing the information information sequence composed of information information corresponding to each information content, it may further include:

[0251] The receiving terminal sends a request to view details of target information, where the target information is any one of the information in the information sequence;

[0252] Based on the details viewing request, the system sends the information content corresponding to the target information to the terminal, so that the terminal can display the information content corresponding to the target information.

[0253] In one or more embodiments, a method for displaying information content is described. As can be seen from the foregoing embodiments, users can further view the corresponding information content (e.g., articles, videos, etc.) for information that they are interested in.

[0254] Specifically, for ease of understanding, please refer to Figure 8 , Figure 8 This is a schematic diagram of an interface displaying information content in an embodiment of this application. Figure 8 Image (A) shows a sequence of information displayed in the form of an information stream. The information sequence includes the information indicated by 801, the information indicated by 802, the information indicated by 803, and the information indicated by 804. Taking the user clicking on the title indicated by 8031 ​​as an example, this triggers a request to view details of the target information (i.e., the information indicated by 803).

[0255] It should be noted that the news content uses hyperlinks; clicking the title in the news article will redirect you to the publishing site to view the original text. Therefore, after clicking the title indicated by 8031, the following will be displayed: Figure 8 The interface shown in Figure (B) is where the information content is displayed. If you need to return to... Figure 8 In the interface shown in Figure (A), you can click the return control indicated by 805.

[0256] Secondly, this application provides a method for displaying information content. Using this method, users can directly access the corresponding information content by clicking on the information (e.g., the title), saving time and effort spent on manual searching and achieving the goal of quickly locating information content.

[0257] Optionally, in the above Figure 6 Based on one or more corresponding embodiments, in another optional embodiment provided by this application, after pushing the information information sequence composed of information information corresponding to each information content, it may further include:

[0258] The receiving terminal sends an information filtering request, wherein the information filtering request carries at least one piece of information for constructing information filtering conditions;

[0259] Based on the information filtering request, determine all information that meets the information filtering criteria from the information information sequence;

[0260] All information that meets the information filtering criteria are sorted in descending order of comprehensive ranking score to obtain information subsequences;

[0261] Send a subsequence of information information to the terminal so that the terminal can display the subsequence of information information.

[0262] In one or more embodiments, a method for filtering information content is described. As can be seen from the foregoing embodiments, since information sequences typically include a large amount of information, users can also set information filtering conditions by inputting relevant information. Based on this, information that meets the information filtering conditions is arranged in descending order of comprehensive ranking score to obtain information subsequences.

[0263] Specifically, for ease of understanding, please refer to Figure 9 , Figure 9 This is a schematic diagram of an interface for filtering information content in an embodiment of this application. As shown in the figure, 901 indicates a search filter box, and 902 indicates a query control. Exemplarily, when a user enters a search scope (e.g., title search or full-text search) and search terms in the search filter box indicated by 901, and clicks the query control indicated by 902, an information filtering request is triggered. Thus, information filtering conditions are constructed based on the user-input search scope and search terms. That is, if the title of the information content includes the user-specified search terms (e.g., "update"), the information filtering conditions are met.

[0264] 903 indicates the expansion control. After the user clicks the expansion control indicated by 903, the advanced search filter box indicated by 904 is displayed. Users can enter game type, article author, business category (e.g., games, technology, entertainment), site, game status, event type (e.g., reservation, skin, launch), site category (e.g., official account, game website), or global games through the advanced search filter box indicated by 904. For example, if a user enters the game status (e.g., "not online") and site category (e.g., "official account") in the search filter box indicated by 904, and clicks the query control indicated by 902, an information filtering request is triggered. Thus, information filtering conditions are constructed based on the user-input game status and site category. That is, if the information content originates from a site category and is in an "not online" state, the information filtering conditions are met.

[0265] Secondly, this application provides a method for filtering information content. Through this method, users can filter information by selecting fields, which on the one hand allows for quick location of the target data needed by the user and accurate matching of specific needs, thereby enhancing the flexibility of data processing. On the other hand, it can reduce information interference and simplify data complexity.

[0266] Optionally, in the above Figure 6 In addition to one or more corresponding embodiments, another optional embodiment provided in this application may further include:

[0267] The receiver terminal sends a summary viewing request, wherein the summary viewing request carries at least one piece of information for constructing summary generation conditions;

[0268] Based on the summary viewing request, obtain the second set of information content that meets the summary generation conditions;

[0269] Based on the second set of information content, generate summary information for each of the M key points, where M and N are both integers greater than or equal to 1;

[0270] Based on the second set of information content and the summary information corresponding to each key point, generate N original sentences corresponding to each key point;

[0271] Send a summary of each key point and N original sentences to the terminal so that the terminal can display the summary of each key point and N original sentences.

[0272] In one or more embodiments, a method for generating intelligent summaries based on information content is described. As can be seen from the foregoing embodiments, when a user notices an object (e.g., a game), they can click the title link and select to view a summary of recent information, thereby redirecting to the intelligent summary page. Users can view summaries generated from the recent information content of that object through the intelligent summary page.

[0273] It should be noted that intelligent summarization refers to a summary generated based on a large language model (LLM) to create a second set of information content. The summary lists the specified key points and provides the most relevant original text for each key point for verification. Therefore, the summary in this application includes key point information and several original sentences.

[0274] Specifically, for ease of understanding, please refer to Figure 10 , Figure 10 This is a schematic diagram of an interface illustrating a smart summary in an embodiment of this application. As shown in the figure, 1001 indicates a summary filtering box, and 1002 indicates a filtering control. For example, when a user enters a global game (e.g., "Game A"), a date range, and a site (e.g., site X and site Y) in the summary filtering box indicated by 1001, and clicks the filtering control indicated by 1002, a summary viewing request is triggered. The summary generation conditions are constructed based on the user's input of the global game, date range, and site.

[0275] Suppose that there are 10 news articles about game A published by site X or site Y within the date range specified by the user. These 10 news articles meet the summary generation conditions, that is, they constitute the second news content set.

[0276] Based on this, for the second set of information content, on the one hand, corresponding summary information is generated according to the specified M key points. On the other hand, since LLM is a generative model, it needs to generate answers based on questions, but it may output results that seem reasonable but are actually inconsistent with reality, logic, or input information. This phenomenon usually manifests as the model fabricating facts, inventing details, or giving seemingly confident but incorrect answers to uncertain questions; that is, it exhibits hallucinations. To avoid the hallucinations and misleading effects of LLM, the N most relevant original sentences for each key point can be extracted from the second set of information content.

[0277] For example, taking "game selling points" as the key point, 1003 is used to indicate the summary information corresponding to "game selling points". 1004 is used to indicate the 5 original sentences corresponding to "game selling points" (in this case, N equals 5). Furthermore, the website and title of the information content to which the original sentences belong can also be displayed, and clicking the title can jump to the original text to view details.

[0278] Secondly, this application provides a method for generating intelligent summaries based on information content. Through this method, users can filter information corresponding to fields to select information content, thereby assisting users in quickly querying summaries of recent information. At the same time, users can also view the original text details corresponding to each key point, ensuring they are not misled by the illusion of the model.

[0279] Optionally, in the above Figure 6 Based on one or more corresponding embodiments, in another optional embodiment provided by this application, N original sentences corresponding to each key point are generated according to the second information content set and the summary information corresponding to each key point, specifically including:

[0280] Each piece of information in the second set of information is segmented into sentences to obtain the set of original sentences.

[0281] For each of the M key points, obtain the similarity between the summary information corresponding to the key point and each original sentence in the set of original sentences;

[0282] For each of the M key points, extract the top N original sentences from the set of original sentences that have the highest similarity to the summary information corresponding to the key point.

[0283] In one or more embodiments, a method for finding original sentences based on key points is introduced. As described in the foregoing embodiments, based on a second set of information content, summary information corresponding to each key point is generated by calling an LLM (Limited Language Management) function. Further, the summary information generated by the LLM for each key point is encoded to obtain a corresponding summary embedding vector (e.g., a 1500-dimensional vector). Thus, the most similar N original sentences are found using M summary embedding vectors.

[0284] Specifically, firstly, each piece of information in the second information content set is segmented into sentences, and short sentences are merged and duplicate statements are removed to obtain the original sentence set. Based on this, each original sentence in the original sentence set is encoded to obtain the corresponding sentence embedding vector (e.g., a 1500-dimensional vector). Assume that the original sentence set includes W original sentences, that is, there are a total of W sentence embedding vectors.

[0285] For each key point, calculate the cosine similarity between the summary embedding vector and the W sentence embedding vectors, i.e.:

[0286]

[0287] Where Sim(t,s) represents the cosine similarity between the t-th summary information and the s-th original sentence. tThis represents the embedding vector corresponding to the t-th abstract information. s Let represent the sentence embedding vector corresponding to the s-th original sentence.

[0288] Based on this, for each summary information, its corresponding N original sentences can be obtained in the following way:

[0289]

[0290] Where s′ represents the original sentence in the set of original sentences that meets the requirements. S′ represents the set of N original sentences. s represents the s-th original sentence in the set of original sentences. This represents the top N original sentences with the highest cosine similarity. Sen represents the set of original sentences. def represents the definition.

[0291] It should be noted that in practical applications, cosine distance, Euclidean distance, Manhattan distance, or Jaccard coefficient can also be used to evaluate similarity. Generally, the smaller the cosine distance, the greater the similarity. Similarly, the smaller the Euclidean distance, the greater the similarity. The smaller the Manhattan distance, the greater the similarity. And the larger the Jaccard coefficient, the greater the similarity.

[0292] Furthermore, this application provides a method for finding original sentences based on key points. Using the above method, an LLM can generate a second set of information content summaries for each key point. However, LLM is prone to errors; therefore, by specifying a generation format, the LLM outputs summary information according to key points, and then finds the most relevant original sentences for each key point, assisting the user in judging the reasonableness of the summary information generated by the LLM.

[0293] Optionally, in the above Figure 6 In addition to one or more corresponding embodiments, another optional embodiment provided in this application may further include:

[0294] The receiving terminal sends a key configuration request, which carries M key identifiers;

[0295] Based on the key configuration request, M key points are determined, where each of the M key points has a one-to-one correspondence with a key point identifier among the M key point identifiers.

[0296] In one or more embodiments, a method for supporting user-configurable key points is described. As can be seen from the foregoing embodiments, users can also set M key points according to actual needs. The following description uses the collection of information content from the gaming industry as an example.

[0297] Specifically, for ease of understanding, please refer to Figure 11 , Figure 11This is a schematic diagram of an interface for selecting key points in an embodiment of this application. As shown in the figure, the key points available for users to select in the gaming industry include, but are not limited to, game selling points, advantages and disadvantages, game potential, game performance, game sales, user reviews, user scale, and developer's future plans. Specifically, 1101 indicates the option corresponding to "game selling points". 1102 indicates the option corresponding to "advantages and disadvantages". 1103 indicates the option corresponding to "game potential". 1104 indicates the option corresponding to "game performance". 1105 indicates the option corresponding to "game sales". 1106 indicates the option corresponding to "user reviews". 1107 indicates the option corresponding to "user scale". 1108 indicates the option corresponding to "developer's future plans". 1109 indicates a confirmation control.

[0298] For example, suppose the user selects both the option indicated by 1101 and the option indicated by 1102, and then clicks the confirmation control indicated by 1109, thus triggering a key point configuration request. This request carries a key point identifier (e.g., 0001) corresponding to "game selling points" and a key point identifier (e.g., 0001) corresponding to "advantages and disadvantages." Therefore, the M key points are determined to include "game selling points" and "advantages and disadvantages."

[0299] Furthermore, this application embodiment provides a method that supports user-configurable key points. Through this method, users can configure corresponding key points according to their actual needs, and generate summary content corresponding to each key point based on the user-configured key points. This accurately meets personalized needs.

[0300] Optionally, in the above Figure 6 In addition to one or more corresponding embodiments, another optional embodiment provided in this application may further include:

[0301] Obtain the image to be processed;

[0302] Based on the image to be processed, information content is obtained through a graph-to-text model;

[0303] It may also include:

[0304] Get the video to be processed;

[0305] Based on the video to be processed, information content is obtained through a video description generation model;

[0306] It may also include:

[0307] Acquire the speech to be processed;

[0308] Information content is obtained based on the speech to be processed through a speech recognition model.

[0309] In one or more embodiments, a method for generating information content is described. As can be seen from the foregoing embodiments, the information content is in text form. In practical applications, collected materials such as images, videos, and audio can also be converted into information content in text form. The implementation method for generating information content will be described below.

[0310] (1) Generating information content based on images;

[0311] Specifically, first, a convolutional neural network (CNN) can be used to extract features from the image to obtain its low-level features. Then, higher-level semantic features are abstracted to understand the image's content. Next, the extracted image features are encoded to obtain a high-dimensional vector containing the image's key information. Finally, the image-to-text model uses the encoded vector and a decoder to generate text describing the image's content, thus obtaining the information content.

[0312] It should be noted that the image-text model involved in this application may adopt the show attend and tell (SAT) model, or the neural baby talk (NBT) model, or the image captioning with visual attention (ICA) model, etc.

[0313] (2) Generate information content based on video;

[0314] Specifically, first, the video to be processed is segmented into keyframes or segments, and visual and audio features are extracted. Then, multimodal information such as visual, audio, and motion is integrated to capture spatiotemporal relationships. Next, a video description generation model is used to analyze the temporal dependencies of video frames, thereby generating a coherent natural language description, i.e., obtaining the information content.

[0315] It should be noted that the video description generation model involved in this application may adopt a unified vision-language model (UniVL), or a temporal generative adversarial network (TempGAN), or a sequence to sequence–video to text (S2VT) model, etc.

[0316] (3) Generate information content based on speech;

[0317] Specifically, first, the speech to be processed undergoes noise reduction, framing, and normalization, followed by the extraction of acoustic features (e.g., Mel-frequency cepstral coefficients). Then, the acoustic features are input into a speech recognition model, which outputs the corresponding text sequence. Finally, a language model is used to adjust and optimize the recognition results, thus obtaining the information content.

[0318] It should be noted that the speech recognition model involved in this application may be a transformer-based automatic speech recognition (Transformer-based ASR) model, or a listen attend and spell (LAS) model, or a deep speech recognition system, etc.

[0319] Secondly, this application provides a method for generating information content. Through this method, video, images, and audio are converted into text-based information content. Information content can be obtained from various types of materials, without being limited by a single channel, thus more comprehensively and diversely meeting the diverse needs of information.

[0320] Optionally, in the above Figure 6 In addition to one or more corresponding embodiments, another optional embodiment provided in this application may further include:

[0321] A website for publishing information;

[0322] If the publishing site of the information content successfully matches the preset site, the information content will be added to the first information content set.

[0323] In one or more embodiments, a method for obtaining a first set of information content is described. As can be seen from the foregoing embodiments, information content typically originates from different websites (e.g., public accounts, official websites, Weibo, social media platforms, etc.). Based on this, information content crawled from preset websites can be added to the first set of information content.

[0324] Specifically, taking the collection of information from the gaming industry as an example, the corresponding preset sites include, but are not limited to, game official accounts, game websites, game forums, and game official Weibo accounts. That is, the information content crawled through these preset sites all belongs to the first set of information content. Among them, game official accounts are official accounts operated by game companies through social media platforms, used to publish information and introduce activities. Game websites are authoritative information platforms established by the game company, used to provide downloads, updates, and customer support. Game official Weibo accounts are accounts opened by game companies on Weibo, used for dynamic posting and interaction with users. Game forums are online discussion platforms where players exchange strategies and share experiences.

[0325] Secondly, this application provides a method for obtaining a first set of information content. Using this method, information content can be crawled from multiple information publishing platforms and added to the first set of information content. This allows for the acquisition of information content covering multiple fields and types, facilitating comprehensive research and analysis.

[0326] Optionally, in the above Figure 6 In addition to one or more corresponding embodiments, another optional embodiment provided in this application may further include:

[0327] Obtain a collection of candidate information content;

[0328] For each piece of information in the candidate information content set, if the title or summary of the information content meets the information matching conditions, the information content is added to the first information content set;

[0329] Among them, at least one of the following conditions must be met for information matching:

[0330] The title of the news content includes target keywords;

[0331] The summary of the information content includes target keywords;

[0332] The title of the news content includes seed words;

[0333] The summary of the information content includes seed words.

[0334] In one or more embodiments, another method for obtaining the first set of information content is described. As can be seen from the foregoing embodiments, information content can be crawled from different websites, thereby obtaining a candidate set of information content. Based on this, information content that meets the information matching conditions is added to the first set of information content.

[0335] Specifically, taking the collection of information content from the gaming industry as an example, one implementation method is to add information content whose titles include target words to a first information content set, where the target words can be game names. Another implementation method is to add information content whose summaries include target words to the first information content set. Yet another implementation method is to add information content whose titles include seed words to the first information content set, where seed words can be game aliases. Seed words represent words used as initial input in natural language processing or data mining, used to expand or identify them as target words. Another implementation method is to add information content whose summaries include seed words to the first information content set.

[0336] Secondly, this application provides another method for obtaining the first set of information content. Using this method, information content related to specific words can be directly extracted. This reduces interference from irrelevant information and makes the information more targeted. Furthermore, it saves time and effort in filtering large amounts of information, improving information acquisition efficiency.

[0337] Based on the above embodiments, the following will be combined with Figure 12 This section introduces the interactive process. Please refer to [link / reference]. Figure 12 , Figure 12 This is a schematic diagram of an interaction flow of the information push method in an embodiment of this application, as shown in the figure. Specifically:

[0338] In step S1, the server crawls information content from hundreds of websites, obtaining daily incremental media articles. These websites include, but are not limited to, WeChat official accounts, official websites, Weibo, and other social media platforms.

[0339] In step S2, the server constructs a configuration table, which includes a preset time score, a preset matching score, a preset attention score, a preset header, and corresponding weights. The weights are coefficients that decay over time.

[0340] In step S3, the terminal sends an information viewing request to the server.

[0341] In step S4, the server responds to the information viewing request and calculates the comprehensive ranking score for each piece of information. The information content is then sorted in descending order according to the comprehensive ranking score, and a sequence of information information in the form of an information stream is generated.

[0342] In step S5, the server pushes a sequence of information to the terminal.

[0343] In step S6, the terminal sends a summary viewing request to the server.

[0344] In step S7, the server responds to the summary viewing request and filters several news items based on the object requested by the user. Then, the LLM large model is invoked to generate intelligent summaries from multiple news items. The intelligent summary includes summary information obtained based on "textual key points" and several original sentences obtained based on "key point text retrieval."

[0345] In step S8, the server pushes summary information of each key point and N original sentences corresponding to each key point to the terminal.

[0346] The information push device in this application is described in detail below. Please refer to [link / reference]. Figure 13 , Figure 13 This is a schematic diagram of one embodiment of the information push device in this application. The information push device 130 includes:

[0347] The acquisition module 1301 is used to respond to the information viewing request and acquire the comprehensive ranking score corresponding to each piece of information in the first information content set. The comprehensive ranking score is positively correlated with the time decay score and the matching decay score. The time decay score is negatively correlated with the amount of time the information content has been published. The matching decay score is negatively correlated with the amount of time the information content has been published. The matching decay score is positively correlated with the number of keywords matched by the information content.

[0348] The sorting module 1302 is used to sort the first set of information content according to the comprehensive sorting score corresponding to each information content, in descending order of the comprehensive sorting score, to obtain the information content sequence;

[0349] The acquisition module 1301 is also used to acquire information information corresponding to each piece of information in the information content sequence, wherein the information information includes metadata corresponding to the information content, and the metadata is used to describe the attribute features related to the information content;

[0350] The push module 1303 is used to push information information sequences consisting of information information corresponding to each information content, wherein each information information in the information information sequence has a one-to-one correspondence with the information content in the same position in the information content sequence.

[0351] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application, the information push device 130 further includes a calculation module 1304 and a determination module 1305;

[0352] The acquisition module 1301 is also used to acquire the publication time of each piece of information in the first set of information content;

[0353] The calculation module 1304 is used to calculate the time decay score of each piece of information in the first set of information content based on the amount of time the information content has been published and a preset time score.

[0354] The calculation module 1304 is also used to, for each piece of information in the first set of information content, if the information content is successfully matched with at least one keyword in the keyword set, sum up the preset matching scores corresponding to each keyword of the at least one keyword to obtain the matching score of the information content.

[0355] The calculation module 1304 is also used to calculate the matching decay score of each piece of information in the first set of information content based on the amount of time the information content has been published and the matching score.

[0356] The determination module 1305 is used to determine the comprehensive ranking score of each piece of information in the first set of information content based on the time decay score and the matching decay score of the information content.

[0357] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0358] The acquisition module 1301 is also used to acquire the preset header part of the information content if the subject object of the information content matches any object in the header object set for each information content in the first information content set. The header object set includes the top K objects with the largest header index, where K is an integer greater than or equal to 1.

[0359] The calculation module 1304 is also used to calculate the header decay score of each piece of information in the first set of information content based on the amount of time the information content has been published and the preset header portion.

[0360] The determination module 1305 is specifically used to determine the comprehensive ranking score of each piece of information in the first set of information content based on the time decay score, matching decay score, and header decay score of the information content.

[0361] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0362] The acquisition module 1301 is specifically used to respond to information viewing requests and, for each piece of information in the first information content set, acquire the publication time of the information content, wherein the information viewing request carries the target account identifier;

[0363] For each piece of information in the first set of information content, the time decay score of the information content is calculated based on the published time and the preset time score.

[0364] For each piece of information in the first set of information content, if the information content matches at least one keyword in the keyword set, the preset matching scores corresponding to each keyword of the at least one keyword are summed to obtain the matching score of the information content.

[0365] For each piece of information in the first set of information content, the matching decay score of the information content is calculated based on the amount of time the information content has been published and the matching score.

[0366] For each piece of information in the first set of information content, if the subject of the information content matches the target account's associated follower, the preset follower score for the information content is obtained.

[0367] For each piece of information in the first set of information content, the attention decay score of the information content is calculated based on the amount of time the information content has been published and the preset attention score;

[0368] For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, and attention decay score.

[0369] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0370] The acquisition module 1301 is also used to respond to the information viewing request. For each piece of information in the first information content set, if the subject object of the information content matches any object in the header object set, the preset header part of the information content is acquired. The header object set includes the top K objects with the largest header index, where K is an integer greater than or equal to 1.

[0371] The calculation module 1304 is also used to calculate the header decay score of each piece of information in the first set of information content based on the amount of time the information content has been published and the preset header portion.

[0372] The acquisition module 1301 is specifically used to determine the comprehensive ranking score of each piece of information in the first set of information based on the time decay score, matching decay score, attention decay score and head decay score of the information content.

[0373] Optionally, in the above Figure 13Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0374] The acquisition module 1301 is also used to acquire a candidate object set, wherein the candidate object set includes T objects, and T is an integer greater than or equal to K;

[0375] The determination module 1305 is also used to determine the head index of each object in the candidate object set based on at least one of the following: user activity, revenue, retention rate and user rating score.

[0376] The determination module 1305 is also used to select the top K objects with the largest head index in the candidate object set as the head object set based on the head index of each object in the candidate object set.

[0377] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0378] The acquisition module 1301 is also used to respond to information viewing requests. For each piece of information in the first information content set, if the industry type of the information content matches the industry type associated with the target account identifier, the preset industry score of the information content is acquired.

[0379] The determination module 1305 is also used to determine the industry decay score of each piece of information in the first information content set, based on the amount of time the information content has been published and the preset industry score.

[0380] The acquisition module 1301 is specifically used to determine the comprehensive ranking score of each piece of information in the first set of information content based on the information content's time decay score, matching decay score, attention decay score, top decay score, and industry decay score.

[0381] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0382] The acquisition module 1301 is specifically used to receive information viewing requests sent by the terminal;

[0383] Based on the information viewing request, obtain the comprehensive ranking score corresponding to each piece of information in the first set of information content;

[0384] The push module 1303 is specifically used to send a sequence of information information consisting of information information corresponding to each piece of information to the terminal so that the terminal can display the sequence of information information.

[0385] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application, the information push device 130 further includes a receiving module 1306 and a sending module 1307.

[0386] The receiving module 1306 is used to receive a detail viewing request for the target information sent by the terminal after pushing the information information sequence consisting of information information corresponding to each information content. The target information is any one of the information information in the information information sequence.

[0387] The sending module 1307 is used to send the information content corresponding to the target information to the terminal according to the details viewing request, so that the terminal can display the information content corresponding to the target information.

[0388] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0389] The receiving module 1306 is further configured to receive an information filtering request sent by the terminal after performing a push operation on the information information sequence consisting of information information corresponding to each information content, wherein the information filtering request carries at least one piece of information for constructing information filtering conditions.

[0390] The determination module 1305 is also used to determine all information information that meets the information filtering conditions from the information information sequence according to the information filtering request;

[0391] The sorting module 1302 is also used to sort all the information information that meets the information filtering conditions in descending order of the comprehensive sorting score, so as to obtain the information information subsequence;

[0392] The sending module 1307 is also used to send a subsequence of information information to the terminal so that the terminal can display the subsequence of information information.

[0393] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0394] The receiving module 1306 is also configured to receive a summary viewing request sent by the terminal, wherein the summary viewing request carries at least one piece of information for constructing summary generation conditions;

[0395] The acquisition module 1301 is also used to acquire a second set of information content that meets the summary generation conditions based on the summary viewing request;

[0396] The acquisition module 1301 is also used to generate summary information corresponding to each of the M key points according to the second information content set, where M and N are both integers greater than or equal to 1;

[0397] The acquisition module 1301 is also used to generate N original sentences corresponding to each key point based on the second information content set and the summary information corresponding to each key point;

[0398] The sending module 1307 is also used to send the summary information corresponding to each key point and N original sentences to the terminal, so that the terminal can display the summary information corresponding to each key point and N original sentences.

[0399] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0400] The acquisition module 1301 is specifically used to segment each piece of information in the second information content set into sentences to obtain the original sentence set;

[0401] For each of the M key points, obtain the similarity between the summary information corresponding to the key point and each original sentence in the set of original sentences;

[0402] For each of the M key points, extract the top N original sentences from the set of original sentences that have the highest similarity to the summary information corresponding to the key point.

[0403] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0404] The receiving module 1306 is also used to receive a key configuration request sent by the terminal, wherein the key configuration request carries M key identifiers;

[0405] The determination module 1305 is also used to determine M key points based on the key point configuration request, wherein the key points among the M key points have a one-to-one correspondence with the key point identifiers among the M key point identifiers.

[0406] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0407] The acquisition module 1301 is also used to acquire the image to be processed;

[0408] The acquisition module 1301 is also used to acquire information content based on the image to be processed through a graph-to-text model;

[0409] The acquisition module 1301 is also used to acquire the video to be processed;

[0410] The acquisition module 1301 is also used to acquire information content based on the video to be processed by a video description generation model;

[0411] The acquisition module 1301 is also used to acquire the voice to be processed;

[0412] The acquisition module 1301 is also used to acquire information content based on the speech to be processed through a speech recognition model.

[0413] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0414] The acquisition module 1301 is also used to acquire the publishing sites of information content;

[0415] The acquisition module 1301 is also used to add the information content to the first information content set if the publishing site of the published information content successfully matches the preset site.

[0416] Optionally, in the above Figure 13 Based on the corresponding embodiments, in another embodiment of the information push device 130 provided in this application,

[0417] The acquisition module 1301 is also used to acquire a set of candidate information content;

[0418] The acquisition module 1301 is also used to add the information content to the first information content set if the title or summary of the information content meets the information matching conditions for each information content in the candidate information content set.

[0419] Among them, at least one of the following conditions must be met for information matching:

[0420] The title of the news content includes target keywords;

[0421] The summary of the information content includes target keywords;

[0422] The title of the news content includes seed words;

[0423] The summary of the information content includes seed words.

[0424] Figure 14This is a schematic diagram of a computer device structure provided in an embodiment of this application. The computer device 1400 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 1422 (e.g., one or more processors) and a memory 1432, and one or more storage media 1430 (e.g., one or more mass storage devices) for storing application programs 1442 or data 1444. The memory 1432 and storage media 1430 can be temporary or persistent storage. The program stored in the storage media 1430 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the computer device. Furthermore, the CPU 1422 may be configured to communicate with the storage media 1430 and execute the series of instruction operations in the storage media 1430 on the computer device 1400.

[0425] Computer device 1400 may also include one or more power supplies 1426, one or more wired or wireless network interfaces 1450, one or more input / output interfaces 1458, and / or one or more operating systems 1441, such as Windows Server. TM Mac OS X TM Unix TM Linux TM FreeBSD TM etc.

[0426] The steps performed by the computer device in the above embodiments can be based on this Figure 14 The computer device structure shown.

[0427] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the methods described in the foregoing embodiments.

[0428] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the methods described in the foregoing embodiments.

[0429] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0430] In this application embodiment, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0431] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0432] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0433] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0434] If the integrated unit is implemented as 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 technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a server or terminal device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing computer programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0435] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for pushing information, characterized in that, include: In response to an information viewing request, a comprehensive ranking score is obtained for each piece of information in the first information content set. The comprehensive ranking score is positively correlated with the time decay score and the matching decay score. The time decay score is negatively correlated with the amount of time the information content has been published and the matching decay score is negatively correlated with the amount of time the information content has been published. The matching decay score is positively correlated with the number of keywords matched by the information content. Based on the comprehensive ranking score corresponding to each piece of information, the first set of information content is sorted in descending order of comprehensive ranking score to obtain the information content sequence; Obtain information information corresponding to each piece of information in the information content sequence, wherein the information information includes metadata corresponding to the information content, and the metadata is used to describe attribute features related to the information content; A push operation is performed on the information information sequence consisting of the information information corresponding to each information content, wherein each information information in the information information sequence has a one-to-one correspondence with the information content in the same position in the information content sequence.

2. The push method according to claim 1, characterized in that, The method further includes: For each piece of information in the first set of information content, obtain the amount of time the information content has been published; For each piece of information in the first set of information content, the time decay score of the information content is calculated based on the published time and a preset time score. For each piece of information in the first set of information content, if the information content matches at least one keyword in the keyword set, the preset matching scores corresponding to each of the at least one keyword are summed to obtain the matching score of the information content. For each piece of information in the first set of information content, the matching decay score of the information content is calculated based on the amount of time the information content has been published and the matching score. For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the time decay score and the matching decay score of the information content.

3. The push method according to claim 2, characterized in that, The method further includes: For each piece of information in the first set of information content, if the subject object of the information content matches any object in the set of header objects, then the preset header part of the information content is obtained, wherein the set of header objects includes the top K objects with the largest header index, and K is an integer greater than or equal to 1; For each piece of information in the first set of information content, the header decay score of the information content is calculated based on the amount of time the information content has been published and the preset header portion. The step of determining a comprehensive ranking score for each piece of information in the first set of information content, based on its time decay score and matching decay score, includes: For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, and header decay score.

4. The push method according to claim 1, characterized in that, The step of responding to an information viewing request by obtaining the comprehensive ranking score corresponding to each piece of information in the first information content set includes: In response to the information viewing request, for each piece of information in the first information content set, the publication time of the information content is obtained, wherein the information viewing request carries a target account identifier; For each piece of information in the first set of information content, the time decay score of the information content is calculated based on the published time and a preset time score. For each piece of information in the first set of information content, if the information content matches at least one keyword in the keyword set, the preset matching scores corresponding to each of the at least one keyword are summed to obtain the matching score of the information content. For each piece of information in the first set of information content, the matching decay score of the information content is calculated based on the amount of time the information content has been published and the matching score. For each piece of information in the first set of information content, if the subject of the information content matches the follow object associated with the target account identifier, then the preset follow score of the information content is obtained. For each piece of information in the first set of information content, the attention decay score of the information content is calculated based on the amount of time the information content has been published and the preset attention score. For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, and attention decay score.

5. The push method according to claim 4, characterized in that, The method further includes: In response to the information viewing request, for each piece of information in the first information content set, if the subject object of the information content matches any object in the header object set, then the preset header part of the information content is obtained, wherein the header object set includes the top K objects with the largest header index, and K is an integer greater than or equal to 1; For each piece of information in the first set of information content, the header decay score of the information content is calculated based on the amount of time the information content has been published and the preset header portion. For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, and attention decay score, including: For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, attention decay score, and header decay score.

6. The push method according to claim 3 or 5, characterized in that, The method further includes: Obtain a candidate object set, wherein the candidate object set includes T objects, and T is an integer greater than or equal to K; For each object in the candidate object set, the head index of the object is determined based on at least one of the following: user activity, revenue, retention rate, and user rating score. Based on the head index of each object in the candidate object set, the top K objects with the largest head index in the candidate object set are taken as the head object set.

7. The push method according to claim 5 or 6, characterized in that, The method further includes: In response to the information viewing request, for each piece of information in the first information content set, if the industry type of the information content matches the industry type associated with the target account identifier, then the preset industry score of the information content is obtained. For each piece of information in the first set of information content, the industry decay score of the information content is determined based on the amount of time the information content has been published and a preset industry score. For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, and attention decay score, including: For each piece of information in the first set of information content, a comprehensive ranking score is determined based on the information content's time decay score, matching decay score, attention decay score, top-ranking decay score, and industry decay score.

8. The push method according to any one of claims 1 to 7, characterized in that, The step of responding to an information viewing request by obtaining the comprehensive ranking score corresponding to each piece of information in the first information content set includes: The information viewing request sent by the receiving terminal; Based on the information viewing request, obtain the comprehensive ranking score corresponding to each piece of information in the first information content set; The operation of pushing the information information sequence composed of the information information corresponding to each information content includes: Send a sequence of information information consisting of information information corresponding to each piece of information to the terminal so that the terminal can display the sequence of information information.

9. The push method according to any one of claims 1 to 8, characterized in that, After performing the push operation on the information information sequence composed of the information information corresponding to each information content, the method further includes: A receiving terminal sends a request to view details of target information, wherein the target information is any one of the information in the information sequence; Based on the details viewing request, the information content corresponding to the target information is sent to the terminal so that the terminal can display the information content corresponding to the target information.

10. The push method according to any one of claims 1 to 9, characterized in that, After performing the push operation on the information information sequence composed of the information information corresponding to each information content, the method further includes: The receiving terminal sends an information filtering request, wherein the information filtering request carries at least one piece of information for constructing information filtering conditions; Based on the information filtering request, determine all information information that meets the information filtering conditions from the information information sequence; All information that meets the information filtering criteria are sorted in descending order of comprehensive ranking score to obtain information subsequences; The information subsequence is sent to the terminal so that the terminal displays the information subsequence.

11. The push method according to any one of claims 1 to 10, characterized in that, The method further includes: A summary viewing request sent by a receiving terminal, wherein the summary viewing request carries at least one piece of information for constructing summary generation conditions; Based on the summary viewing request, obtain a second set of information content that meets the summary generation conditions; Based on the second set of information content, generate summary information for each of the M key points, where M and N are both integers greater than or equal to 1; Based on the second set of information content and the summary information corresponding to each key point, N original sentences corresponding to each key point are generated; The terminal is sent with a summary of each key point and N original sentences, so that the terminal can display the summary of each key point and N original sentences.

12. The push method according to claim 11, characterized in that, The step of generating N original sentences corresponding to each key point based on the second information content set and the summary information corresponding to each key point includes: Each piece of information in the second set of information content is segmented into sentences to obtain the set of original sentences; For each of the M key points, obtain the similarity between the summary information corresponding to the key point and each original sentence in the set of original sentences; For each of the M key points, the top N original sentences with the highest similarity to the summary information corresponding to the key point are obtained from the set of original sentences.

13. The push method according to claim 11 or 12, characterized in that, The method further includes: The receiving terminal sends a key configuration request, wherein the key configuration request carries M key identifiers; Based on the key configuration request, the M key points are determined, wherein the key points among the M key points have a one-to-one correspondence with the key point identifiers among the M key point identifiers.

14. The push method according to any one of claims 1 to 13, characterized in that, The method further includes: Obtain the image to be processed; Based on the image to be processed, information content is obtained through a graph-to-text model; The method further includes: Get the video to be processed; Based on the video to be processed, information content is obtained through a video description generation model; The method further includes: Acquire the speech to be processed; Based on the speech to be processed, information content is obtained through a speech recognition model.

15. The push method according to any one of claims 1 to 14, characterized in that, The method further includes: A website for publishing information; If the publishing site of the published information content successfully matches a preset site, the information content is added to the first information content set.

16. The push method according to any one of claims 1 to 14, characterized in that, The method further includes: Obtain a collection of candidate information content; For each piece of information in the candidate information content set, if the title or summary of the information content meets the information matching conditions, then the information content is added to the first information content set. The conditions for satisfying the information matching criteria include at least one of the following: The title of the information content includes target keywords; The summary of the information content includes target words; The title of the information content includes seed words; The summary of the information content includes seed words.

17. An information push device, characterized in that, include: The acquisition module is used to respond to an information viewing request and acquire the comprehensive ranking score corresponding to each piece of information in the first information content set. The comprehensive ranking score is positively correlated with the time decay score and positively correlated with the matching decay score. The time decay score is negatively correlated with the amount of time the information content has been published and the matching decay score is negatively correlated with the amount of time the information content has been published. Furthermore, the matching decay score is positively correlated with the number of keywords matched by the information content. The sorting module is used to sort the first set of information content according to the comprehensive sorting score corresponding to each information content, in descending order of the comprehensive sorting score, to obtain the information content sequence; The acquisition module is further configured to acquire information information corresponding to each piece of information in the information content sequence, wherein the information information includes metadata corresponding to the information content, and the metadata is used to describe attribute features related to the information content; The push module is used to push the information information sequence consisting of the information information corresponding to each information content, wherein each information information in the information information sequence has a one-to-one correspondence with the information content in the same position in the information content sequence.

18. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the information push method according to any one of claims 1 to 16.

19. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the information push method according to any one of claims 1 to 16.

20. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the information push method according to any one of claims 1 to 16.

Images