Multimedia content push method and device

Abstract

The present application discloses a method and device for pushing multimedia content, which belongs to the field of computer technology. The method includes: determining the entity object associated with the multimedia content; obtaining the popularity information of the entity object; determining the target timeliness information of the multimedia content based on the popularity information of the entity object; and pushing the multimedia content according to the target timeliness information. The embodiment of the present application first determines the timeliness of the multimedia content and the associated entity object, then determines the popularity of the entity object, and then combines the timeliness of the multimedia content and the popularity of the associated entity object to update the timeliness of the multimedia content to improve the accuracy of the timeliness of the multimedia content, and finally pushes the multimedia content according to the updated timeliness to reduce expired content in the information flow, reduce the operating pressure of the server, and effectively improve resource utilization.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method and device for pushing multimedia content. Background Art

[0002] With the development of Internet technology, more and more information flow content services have emerged, through which users can receive various news and information.

[0003] To improve users' information acquisition experience, various information platforms often push content to users based on certain information push standards. For example, content from information flow services may be pushed to users at a fixed time limit.

[0004] However, the timeliness of content in current information flows is often inaccurate, resulting in expired content in the information flow and low resource utilization. Summary of the Invention

[0005] The embodiments of the present application provide a method and apparatus for pushing multimedia content, which improve the push rate of multimedia content and enhance the integration degree of compression methods.

[0006] In one aspect, an embodiment of the present application provides a method for pushing multimedia content, the method comprising:

[0007] Determining entity objects associated with multimedia content;

[0008] Obtaining the heat information of the entity object;

[0009] Determining target timeliness information of the multimedia content based on the popularity information of the entity object;

[0010] The multimedia content is pushed according to the target timeliness information.

[0011] On the other hand, an embodiment of the present application provides a device for pushing multimedia content, the device comprising:

[0012] An entity object determination module, configured to determine entity objects associated with multimedia content;

[0013] A heat information acquisition module, used to acquire heat information of the entity object;

[0014] a timeliness information determination module, configured to determine target timeliness information of the multimedia content based on the popularity information of the entity object;

[0015] The content push module is used to push the multimedia content according to the target timeliness information.

[0016] On the other hand, an embodiment of the present application provides a computer device, which includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the above-mentioned multimedia content push method.

[0017] On the other hand, an embodiment of the present application provides a computer-readable storage medium, which stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement the above-mentioned multimedia content push method.

[0018] In another aspect, embodiments of the present application provide a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the above-described method for pushing multimedia content.

[0019] The technical solutions provided in the embodiments of the present application can bring the following beneficial effects:

[0020] By first determining the timeliness of multimedia content and the physical objects associated with the multimedia content, then determining the popularity of the physical objects associated with the multimedia content, and then combining the timeliness of the multimedia content and the popularity of the physical objects associated with the multimedia content, the timeliness of the multimedia content is updated to improve the accuracy of the timeliness of the multimedia content. Finally, the multimedia content is pushed according to the updated timeliness to reduce expired content in the information flow, reduce server operating pressure, and effectively improve resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0022] Figure 1 is a schematic diagram of an application program operating environment provided by an embodiment of the present application;

[0023] Figure 2 This is a flowchart of a method for pushing multimedia content provided by one embodiment of the present application;

[0024] Figure 3is a flowchart of a method for pushing multimedia content provided by another embodiment of the present application;

[0025] Figure 4 A schematic diagram of a display page for an entity object is shown as an example;

[0026] Figure 5 The flowchart of determining the timeliness of an article is shown as an example;

[0027] Figure 6 A schematic diagram exemplarily shows determining variety shows and film and television works associated with an article;

[0028] Figure 7 This is a block diagram of a multimedia content push device provided by one embodiment of the present application;

[0029] Figure 8 This is a structural block diagram of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0030] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0031] The technical solution of this application relates to the fields of artificial intelligence technology and cloud technology, which are introduced and explained below.

[0032] Artificial Intelligence (AI) refers to the theories, methods, techniques, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, to perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that seeks to understand the essence of intelligence and produce new intelligent machines that can respond in a manner similar to human intelligence. AI also involves studying the design principles and implementation methods of various intelligent machines, enabling them to possess the capabilities of perception, reasoning, and decision-making.

[0033] Artificial intelligence (AI) technology is a comprehensive discipline encompassing a wide range of fields, encompassing both hardware and software technologies. Foundational AI technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0034] Machine learning (ML) is a multidisciplinary field that encompasses probability theory, statistics, approximation theory, convex analysis, and algorithmic complexity theory. It specifically studies how computers can simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is at the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications span all areas of AI. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and self-learning.

[0035] Natural language processing (NLP) is a key area of research in computer science and artificial intelligence. It studies the theories and methods that enable effective communication between humans and computers using natural language. Natural language processing (NLP) integrates linguistics, computer science, and mathematics. Therefore, research in this field involves natural language—the language we use in everyday life—and is closely linked to the study of linguistics. Natural language processing technologies typically include text processing, semantic understanding, machine translation, robotic question answering, and knowledge graphs.

[0036] Computer vision (CV) is the science of making machines "see." Specifically, it refers to machine vision techniques such as using cameras and computers to replace the human eye in identifying, tracking, and measuring objects. This involves further processing the images, transforming them into images more suitable for human observation or transmission to instruments. As a scientific discipline, computer vision studies related theories and technologies, attempting to build artificial intelligence systems capable of extracting information from images or multidimensional data. Computer vision technologies typically include image processing, image recognition, image semantic understanding, image retrieval, video processing, video semantic understanding, video content / behavior recognition, 3D object reconstruction, virtual reality, augmented reality, simultaneous localization and mapping, and common biometric recognition technologies such as facial recognition and fingerprint recognition.

[0037] The following is a brief introduction to the relevant terms and nouns that may be involved in the embodiments of the present application to facilitate understanding by those skilled in the art.

[0038] Regular expressions, also known as regular expressions (RE), are a concept in computer science. Regular expressions are often used to search for and replace text that matches a certain pattern (rule). Many programming languages support string operations using regular expressions. Regular expressions use predefined specific characters and combinations of these specific characters to form a "regular string," which is used to express a filtering logic for a string. Given a regular expression and another string, the following objectives can be achieved: 1. Determine whether the given string meets the filtering logic of the regular expression (called a "match"); 2. Use the regular expression to obtain the specific part of the string we want.

[0039] The TextRank algorithm is a graph-based text ranking algorithm. It breaks text into components (words, sentences) and builds a graph model. It then uses a voting mechanism to rank the most important components of the text. It can extract keywords and summarize content using only the information in a single document. TextRank does not require prior training on multiple documents, making it widely used due to its simplicity and effectiveness.

[0040] Word embedding is a general term for a set of language modeling and feature learning techniques in embedded natural language processing, in which words or phrases from a vocabulary are mapped to vectors of real numbers. Conceptually, it involves mathematically embedding each word from a one-dimensional space into a continuous vector space with lower dimensions. Methods for generating this mapping include neural networks, dimensionality reduction using word co-occurrence matrices, probabilistic models, interpretable knowledge base approaches, and explicit representations of terms in the context in which they appear. When used as the underlying input representation, word and phrase embeddings have been shown to improve the performance of natural language processing tasks such as grammatical parsing and sentiment analysis. Word embeddings have well-defined semantic properties and are a common way to represent word features. The value of each dimension of a word embedding represents a feature with a certain semantic and grammatical interpretation. Therefore, each dimension of a word embedding can be referred to as a word feature.

[0041] Word2Vec is a related model used to generate word vectors and is one of the word embedding techniques used to generate vectors from words. This is probably obvious from the name itself. Word2Vec is a shallow neural network with only two layers, and therefore does not qualify as a deep learning model. It takes a text corpus as input and generates vectors as output. These vectors are called feature vectors for the words in the input corpus. It converts the corpus into numerical data that can be understood by deep neural networks. Word2Vec aims to understand the probability of two or more words appearing together, thereby grouping words with similar meanings together to form clusters in vector space. Like any other machine learning or deep learning model, Word2Vec becomes increasingly effective by learning from past data and past words. Therefore, given sufficient data and context, it can accurately guess the meaning of a word based on past events and context, much like how we understand language.

[0042] Please refer to Figure 1 , which shows a schematic diagram of an application program running environment provided by an embodiment of the present application. The application program running environment may include: a terminal 10 and a server 20.

[0043] The terminal 10 may be an electronic device such as a mobile phone, a tablet computer, a game console, an e-book reader, a multimedia player, a wearable device, a PC (Personal Computer), etc. A client of an application program may be installed in the terminal 10 .

[0044] In the embodiment of the present application, the above-mentioned application can be any application that can provide information flow content services. Typically, the application is a content recommendation application. Of course, in addition to content recommendation applications, other types of applications can also provide information flow content services. For example, news applications, social applications, interactive entertainment applications, browser applications, shopping applications, content sharing applications, virtual reality (VR) applications, augmented reality (AR) applications, etc., which are not limited in the embodiment of the present application. In addition, for different applications, the content pushed will also be different, and the corresponding functions will also be different. This can be pre-configured according to actual needs, which is not limited in the embodiment of the present application. Optionally, a client of the above-mentioned application is running in the terminal 10. In some embodiments, the above-mentioned information flow content service covers many vertical content such as variety shows, movies, news, finance, sports, entertainment, games, etc., and users can enjoy content services in many forms such as articles, pictures, small videos, short videos, live broadcasts, special topics, columns, etc. through the above-mentioned information flow content service.

[0045] The server 20 is used to provide background services for the client of the application in the terminal 10. For example, the server 20 can be the background server of the aforementioned application. The server 20 can be a single server, a server cluster consisting of multiple servers, or a cloud computing service center. Optionally, the server 20 provides background services for the application in multiple terminals 10 simultaneously.

[0046] Optionally, the terminal 10 and the server 20 may communicate with each other via a network 30 .

[0047] Please refer to Figure 2 , which shows a flowchart of a method for pushing multimedia content provided by an embodiment of the present application. The method can be applied to a computer device, which refers to an electronic device with data calculation and processing capabilities. For example, the execution subject of each step can be Figure 1 The server 20 in the application running environment is shown. The method may include the following steps (210-240).

[0048] Step 210: Determine the entity object associated with the multimedia content.

[0049] Multimedia refers to the integration of various media, including text, audio, video, and images. Multimedia content refers to information or material carried and transmitted by various media. In some potential application scenarios, such as information streaming services, multimedia content is used for user reading, viewing, and information transmission.

[0050] Entity objects are distinguishable and identifiable objects or things. Entity objects can include specific people or things; abstract events; or intangible entities, such as film and television works or variety shows.

[0051] Optionally, the entity object associated with the multimedia content is determined based on the specific content of the multimedia content. For example, if the specific content of an online article is to introduce a film or TV work, then the film or TV work is the entity object associated with the multimedia content.

[0052] Optionally, the entity object associated with the multimedia content is determined based on the tag information of the multimedia content. For example, if the tag of an online article is a film or TV work, then the film or TV work is the entity object associated with the multimedia content.

[0053] Optionally, entity objects associated with multimedia content are determined through entity recognition processing. The above-mentioned entity recognition technology includes keyword extraction processing, regular extraction processing, etc., which are not limited in the embodiments of the present application.

[0054] Step 220: Obtain the heat information of the entity object.

[0055] Popularity information reflects the level of public interest. In real life, there are often many hot topics of public concern, which can involve one or more entities. For example, a new film or television work or variety show often attracts widespread public attention. Popularity information can reflect the public's level of interest in the new film or television work or variety show.

[0056] Optionally, the popularity information includes the popularity type of the entity object. For example, the popularity type includes expired content, recent popular content, and current popular content. Alternatively, the popularity type of the entity object can be determined using popularity-related data. This popularity-related data is data that reflects the popularity of the entity object. For example, the popularity-related data may include, but is not limited to, the number of multimedia content associated with the entity object, exposure rate, click-through rate, like rate, and pageviews.

[0057] Optionally, the popularity information includes a popularity value, which reflects the current popularity of the entity object. The popularity value can be determined based on data such as the number, exposure rate, click-through rate, appreciation rate, and page views of multimedia content associated with the entity object.

[0058] Step 230 : Determine target timeliness information of the multimedia content based on the popularity information of the entity object.

[0059] The target timeliness information represents the target push duration of the multimedia content. The above-mentioned target timeliness information can be the actual timeliness information of the multimedia content, or the timeliness information in a specific application scenario, or the timeliness information obtained by modifying the initial timeliness information. The above-mentioned target push duration refers to the recommended duration set for the multimedia content, or the upper limit of the push duration of the multimedia content. Multimedia content has timeliness. Due to the differences in multimedia content, the timeliness of different multimedia content is different. The above-mentioned timeliness can be reflected by the target push duration of the multimedia content. For example, in the information flow content service scenario, it is necessary to push the multimedia content in the scenario in a targeted manner. The above-mentioned target duration can represent the longest push duration of the multimedia content. Once the cumulative push duration exceeds the target push duration, it means that the multimedia content is expired content, the user has low interest in expired content, and it may be that the user has already seen content similar to the expired content, so the push of expired multimedia content can be stopped.

[0060] Optionally, the timeliness information includes a timeliness type, which refers to a type of multimedia content with a common timeliness. Multimedia content of the same timeliness type has the same or similar timeliness. Optionally, the timeliness type corresponds to the push duration. For example, the push durations of multimedia content of the same timeliness type are the same, or the push durations of multimedia content of the same timeliness type are in the same duration interval, and the shortest push duration of multimedia content of the same timeliness type is greater than or equal to the minimum value of the duration interval, and the longest push duration of multimedia content of the same timeliness type is less than or equal to the maximum value of the duration interval.

[0061] Optionally, the timeliness type of the multimedia content associated with the entity object is determined based on the popularity type of the entity object. There is a corresponding relationship between the popularity type and the timeliness type, and the timeliness type of the multimedia content associated with the entity object is determined based on the corresponding relationship between the popularity type and the timeliness type.

[0062] Optionally, a target push duration for multimedia content associated with the entity object is determined based on the entity object's popularity type. There is a corresponding relationship between the aforementioned popularity type and the target push duration, and based on the corresponding relationship between the popularity type and the target push duration, the timeliness type of the multimedia content associated with the entity object is determined.

[0063] Optionally, the popularity information of the physical object and the feature information of the multimedia content are input into a trained machine learning model, which then outputs the timeliness type or target push duration of the multimedia content. By combining the popularity of the physical object to determine the timeliness of the multimedia content, the accuracy of timeliness determination is improved.

[0064] In an exemplary embodiment, the method further comprises the following steps:

[0065] Step 250: Acquire initial timeliness information of the multimedia content.

[0066] The initial timeliness information represents the initial push duration of the multimedia content.

[0067] Optionally, the initial timeliness information of the multimedia content is determined by a timeliness classification model. The timeliness classification model can be a pre-trained machine learning model. The timeliness classification model can determine the timeliness type of the multimedia content and can also directly determine the push duration of the multimedia content.

[0068] Optionally, the initial timeliness information of the multimedia content is determined based on the multimedia content's tag information. The tag information is used to reflect the theme of the multimedia content, such as a film or television work, variety show, news hotspot, or hotly debated topic. In one possible implementation, a mapping relationship exists between the tag information and the timeliness information. The mapping relationship can serve as a basis for determining the timeliness information, for example, determining which tag corresponds to which timeliness type, or corresponds to a specific target push duration.

[0069] Optionally, the initial expiration information of the multimedia content is determined by its category. These categories may include movies, TV series, variety shows, finance, entertainment, and social events. The expiration information of the multimedia content can be determined based on the correspondence between the category and the expiration time. For example, if the multimedia content is a movie, the default expiration time corresponding to the movie category will be used as the expiration information of the multimedia content.

[0070] In a possible implementation, target timeliness information of the multimedia content is determined based on the popularity information of the physical object and the initial timeliness information of the multimedia content.

[0071] In the information flow content service scenario, a timeliness determination strategy will be set to establish a correspondence between the content timeliness and the content category. By determining the category corresponding to the content in the information flow, and then determining the timeliness of the content based on the category corresponding to the content in the information flow and the correspondence between the content timeliness and the content category. However, there is often such a situation that the popularity of the entity object associated with the multimedia content has long declined and has become expired hot content, but because the target push time corresponding to the category to which the multimedia content belongs is long, the system will still push the multimedia content to the user for a long time, and the user may receive content that has already been read, affecting the user experience. The embodiment of the present application combines the timeliness of the multimedia content and the popularity of the entity object associated with the multimedia content to update the timeliness of the multimedia content, so as to more accurately determine the timeliness of the multimedia content.

[0072] The method provided in the embodiment of the present application includes determining whether the popularity information of the entity object meets the preset popularity conditions, and whether the timeliness information of the multimedia content meets the preset timeliness conditions, and then updating the timeliness information of the multimedia content based on the judgment results of the two to obtain the target timeliness information of the multimedia content.

[0073] The preset heat condition is used to filter out entity objects of target heat. Optionally, the preset heat condition includes that the heat-related data meets a heat threshold condition. Optionally, the preset heat condition includes that the heat type is a target heat type, where the target heat type is a heat type selected based on the application scenario. Optionally, the preset heat condition includes that the heat value is greater than or equal to a heat value threshold.

[0074] The above-mentioned heat threshold conditions are used to distinguish entity objects of different heat. Optionally, the heat threshold conditions include that the size relationship between the heat association data and the heat threshold corresponding to the heat association data satisfies the preset rules. Heat association data includes but is not limited to the number, exposure rate, click-through rate, appreciation rate, and page views of multimedia content. The above-mentioned heat threshold is a general term for at least one threshold corresponding to at least one heat association data. Accordingly, the heat threshold conditions include but are not limited to at least one of the number of multimedia content being greater than or equal to the quantity threshold; the exposure rate being greater than or equal to the exposure rate threshold; the click-through rate being greater than or equal to the click-through rate threshold; the appreciation rate being greater than or equal to the appreciation rate threshold; and the page views being greater than or equal to the page views threshold. The above-mentioned heat threshold conditions can be combined, and the embodiments of the present application are not limited to this. The above-mentioned quantity threshold is a threshold for evaluating the popularity of multimedia content from the dimension of the quantity of multimedia content; the above-mentioned exposure rate threshold is a threshold for evaluating the popularity of multimedia content from the dimension of the exposure rate of multimedia content among users; the click-through rate threshold is a threshold for evaluating the popularity of multimedia content from the dimension of the click-through rate of multimedia content; the appreciation rate threshold is a threshold for evaluating the popularity of multimedia content from the dimension of the appreciation rate of multimedia content; and the view volume threshold is a threshold for evaluating the popularity of multimedia content from the dimension of the view volume of multimedia content.

[0075] Optionally, one type of heat association data corresponds to one heat threshold, or multiple types of heat association data correspond to one heat threshold, or one type of heat association data corresponds to multiple heat thresholds, which is not limited in the embodiment of the present application.

[0076] The above-mentioned preset timeliness conditions are used to filter out entity objects with target timeliness. Optionally, the preset timeliness conditions include that the timeliness type is a target timeliness type. Optionally, the preset timeliness conditions include that the target push duration is greater than or equal to the target duration threshold. The above-mentioned target duration threshold can be the longest push duration of multimedia content, for example, the target duration threshold is 7 days. The above-mentioned preset timeliness conditions can be adjusted according to the specific application scenario, and are not limited to the comparison in the embodiment of the present application.

[0077] In an exemplary embodiment, when the popularity information of the entity object meets the preset popularity condition and the timeliness information of the multimedia content meets the preset timeliness condition, the initial timeliness information of the multimedia content is modified to obtain the actual timeliness information of the multimedia content, and the target recommendation duration corresponding to the actual timeliness information is less than the target recommendation duration corresponding to the initial timeliness information.

[0078] Optionally, the entity object whose popularity information meets the preset popularity conditions belongs to the recent hot type. Optionally, the multimedia content whose timeliness information meets the preset timeliness conditions belongs to the long timeliness type. In the case that the multimedia content is of the long timeliness type, and the entity object associated with the multimedia content belongs to the recent hot type, the timeliness type of the multimedia content is changed from the long timeliness type to the medium timeliness type or the short timeliness type, or the target push duration of the multimedia content is shortened. The basis for distinguishing the above-mentioned long timeliness type, medium timeliness type and short timeliness type is that their corresponding target push durations are different. For example, the target push duration corresponding to the long timeliness type is 7 days, the target push duration corresponding to the medium timeliness type is 3 days, and the target push duration corresponding to the short timeliness type is 1 day.

[0079] In an exemplary embodiment, the above step 230 may be implemented as follows.

[0080] When the popularity type of the entity object is a recent hot type and the initial push duration corresponding to the initial timeliness information is greater than or equal to the target duration threshold, the initial timeliness information of the multimedia content is modified to obtain the target timeliness information of the multimedia content.

[0081] The proportion of multimedia content associated with entity objects of recently popular types in the content database is higher than the target proportion threshold. The target duration threshold can be the maximum push duration of multimedia content, for example, the target duration threshold is 7 days. The target proportion threshold is a threshold for assessing the popularity of entity objects from the perspective of the proportion of multimedia content associated with the entity objects in the content database, and can be a percentage. The target recommendation duration corresponding to the target timeliness information is less than the initial recommendation duration corresponding to the initial timeliness information.

[0082] The above-mentioned recent hot types can be determined by the heat association data of the entity object. In a possible implementation, the heat association data is the proportion of multimedia content associated with the entity object in the content database. The data statistical scope of the above-mentioned content database can be multimedia content in the entire network. For example, multimedia content information is captured from the entire network, and a data record is generated for each multimedia content, recording the network address, content type, associated entity object and other information of the multimedia content. The data statistical scope of the above-mentioned content database can also be multimedia content in the information flow content service, or multimedia content outside the information flow content service. The heat of the entity object can be represented by comparing the proportion of multimedia content associated with the entity object in the content database with the target proportion threshold. The proportion of multimedia content associated with entity objects of the recent hot type in the content database will be higher than the proportion of multimedia content associated with the hot entity object in the content database. Therefore, according to the actual scenario, a suitable threshold can be selected as the target proportion threshold to determine the heat type of the entity object to represent the heat information of the entity object.

[0083] Optionally, the initial push duration corresponding to the initial validity information of the multimedia content is modified to a target push duration that is less than the initial push duration, thereby obtaining the target validity information of the multimedia content. Optionally, the above-mentioned initial validity information includes a validity type. In the case where the multimedia content is of a long validity type, the long validity type of the multimedia content is modified to a medium validity type or a short validity type, thereby obtaining the target validity information of the multimedia content. The above-mentioned long validity type, medium validity type, and short validity type correspond to different push durations, respectively. For example, the push duration of the long validity type is 7 days, the push duration of the medium validity type is 4 days, and the push duration of the short validity type is 2 days.

[0084] Optionally, the target push duration of the multimedia content is determined based on the difference between the initial push duration and a target duration threshold. For example, the target push duration corresponds to the difference between the initial push duration and the target duration threshold. The target push duration of the multimedia content is determined based on the corresponding relationship between the target push duration and the difference between the initial push duration and the target duration threshold. Optionally, the target duration threshold is used as the target push duration for the multimedia content.

[0085] In one possible implementation, if the entity object's popularity type is a recent hot topic and the initial push duration corresponding to the initial timeliness information is greater than or equal to the target duration threshold, the initial push duration is modified based on the type of multimedia content to obtain a target push duration for the multimedia content. Optionally, there is a correspondence between the target push duration and the type of multimedia content, and this correspondence can be pre-set based on the business scenario. For example, if the multimedia content type is a film, television, or variety show article, the corresponding target push duration is 2 days.

[0086] In one possible implementation, if the entity object's popularity type is a recent hot topic and the initial push duration corresponding to the initial timeliness information is greater than or equal to the target duration threshold, the initial push duration is modified based on the content of the multimedia content to obtain a target push duration for the multimedia content. Optionally, there is a correspondence between the target push duration and the content of the multimedia content, and this correspondence can be pre-set based on the business scenario. For example, if the multimedia content is an introduction to a film, television, or variety show, the corresponding target push duration is 3 days.

[0087] In a possible implementation, when the popularity type of the entity object is a recent hot type and the initial push duration corresponding to the initial timeliness information is greater than or equal to the target duration threshold, the initial push duration of the multimedia content is modified to a fixed target push duration.

[0088] Schematically, by marking the popularity type of an entity object as a recent hot type, it is indicated that the popularity of this entity object has decreased, or that this entity object has become an expired hot topic. Multimedia content associated with recently popular entity objects has a longer initial push duration corresponding to the category to which it belongs, so the system often pushes such multimedia content according to the longer initial push duration, such as variety show and film articles that provide plot commentary. However, because the entity object is already a recent hot type, users may receive duplicate content about the entity object or related content that they already know, affecting the user experience. Therefore, the embodiment of the present application shortens the push duration of long-term multimedia content for recently popular entity objects to ensure the timeliness accuracy of the multimedia content. Taking film and television variety articles as an example, the embodiment of the present application combines the latest film and television variety works information and uses entity recognition technology to determine the film and television variety works involved in the film and television variety articles. To a certain extent, it ensures the timeliness accuracy of articles that are more inclined to explanation and introduction but are related to recent popular variety shows and films, avoiding the serious expiration of articles at the user level caused by such articles being left on the recommendation side for a long time, thereby improving the user experience in terms of timeliness to a certain extent.

[0089] Step 240: Push the multimedia content according to the target timeliness information.

[0090] Multimedia content is pushed according to the target push duration corresponding to the target timeliness information. In an illustrative application scenario, such as an information flow content service, there are many multimedia contents. First, the multimedia content will be screened to determine the recommended multimedia content and the non-recommended multimedia content. The recommended multimedia content determined above will be entered into the content recommendation library and pushed to the user first. In this case, the above-mentioned target push duration can also be understood as the upper limit of the duration of the recommended multimedia content in the content recommendation library. Once the duration of the multimedia content in the content recommendation library, that is, (cumulative push duration), reaches the upper limit of the duration, that is, (target push duration), the multimedia content will be removed from the content recommendation library, so that the push of the multimedia content will be stopped to save computing resources.

[0091] In an exemplary embodiment, if the cumulative push duration corresponding to the multimedia content does not reach the target recommended duration corresponding to the initial timeliness information, the multimedia content is pushed. If the cumulative push duration corresponding to the multimedia content reaches the target recommended duration corresponding to the initial timeliness information, the multimedia content is stopped from being pushed.

[0092] In one possible implementation, when multimedia content is added to the content recommendation library, the system will push the multimedia content in the content recommendation library to the client. When the cumulative push time reaches the initial recommendation time, the multimedia content will be removed from the content recommendation library to stop pushing the multimedia content.

[0093] In another possible implementation, multimedia content whose cumulative push duration has not reached the initial recommended duration is marked, and the system pushes the marked multimedia content to the client. When the cumulative push duration reaches the initial recommended duration, the mark of the multimedia content is deleted, thereby stopping the push of the multimedia content.

[0094] In an exemplary embodiment, if the cumulative push duration corresponding to the multimedia content does not reach the target recommended duration corresponding to the target timeliness information, the multimedia content is pushed. If the cumulative push duration corresponding to the multimedia content reaches the target recommended duration corresponding to the target timeliness information, the multimedia content is stopped from being pushed.

[0095] In one possible implementation, when multimedia content is added to the content recommendation library, the system will push the multimedia content in the content recommendation library to the client. When the cumulative push time reaches the target recommendation time, the multimedia content will be removed from the content recommendation library to stop pushing the multimedia content.

[0096] In another possible implementation, multimedia content whose cumulative push duration has not reached the target recommended duration is marked, and the system pushes the marked multimedia content to the client. When the cumulative push duration reaches the target recommended duration, the mark of the multimedia content is deleted, thereby stopping the push of the multimedia content.

[0097] To sum up, the technical solution provided by the embodiment of the present application first determines the timeliness of the multimedia content and the physical objects associated with the multimedia content, then determines the popularity of the physical objects associated with the multimedia content, and then combines the timeliness of the multimedia content and the popularity of the physical objects associated with the multimedia content to update the timeliness of the multimedia content to improve the accuracy of the timeliness of the multimedia content, and finally pushes the multimedia content according to the updated timeliness to reduce expired content in the information flow, reduce the operating pressure of the server, and effectively improve resource utilization.

[0098] Please refer to Figure 3 , which shows a flowchart of a method for pushing multimedia content provided by another embodiment of the present application. The method can be applied to a computer device, which refers to an electronic device with data calculation and processing capabilities. For example, the execution subject of each step can be Figure 1 The server 20 in the application running environment is shown. The method may include the following steps (310-380).

[0099] Step 310: Determine a feature vector corresponding to the multimedia content.

[0100] The above feature vectors represent content features of multimedia content.

[0101] In an exemplary embodiment, the above step 310 can be implemented through the following steps ( 311 - 314 ).

[0102] Step 311: Acquire text information corresponding to the multimedia content.

[0103] According to the type of multimedia content, corresponding text information of the multimedia content is obtained.

[0104] Optionally, the multimedia content is text content, and accordingly, the above text information includes a title and a body.

[0105] Step 312: perform word segmentation processing on the text information to obtain a word segmentation result.

[0106] The text in the text information is segmented to obtain a segmentation result. The above-mentioned segmentation process is used to divide the input text into characters and words. Optionally, the segmentation granularity of the above-mentioned segmentation process can be defined according to the actual application scenario, which is not limited by the embodiment of the present application. Optionally, the segmentation process is implemented by a segmentation algorithm, for example, a dictionary-based rule matching method and a statistical-based machine learning method, which is not limited by the embodiment of the present application.

[0107] In the case where the multimedia content is text content, the title is segmented to obtain a title segmentation result; the text is segmented to obtain a text segmentation result.

[0108] Step 313: Determine the word vector corresponding to the word in the word segmentation result.

[0109] Based on the correspondence between words and word vectors in the word vector library, the word vector corresponding to the word in the word segmentation result is determined. The above-mentioned word vector library is generated by a pre-trained word vector model. Optionally, the above-mentioned word vector model is a Word2Vec model, and the training data of the above-mentioned Word2Vec model includes text content in the information flow content service. In a possible real-time method, the Word2Vec model is first trained using 1 million historical articles in the multimedia content library to generate a 100-dimensional word vector library. The word vector library can reach a scale of 2 million.

[0110] When the multimedia content is text content, the word vectors corresponding to the words in the title word segmentation result are determined; and the word vectors corresponding to the words in the text word segmentation result are determined.

[0111] Step 314: Determine a feature vector corresponding to the multimedia content based on the word vector corresponding to the word.

[0112] In a possible implementation, word vectors are accumulated to obtain accumulated word vectors; and the accumulated word vectors are normalized to obtain feature vectors corresponding to the multimedia content.

[0113] When the multimedia content is text content, the word vectors corresponding to the words in the title word segmentation results are accumulated to obtain the accumulated word vectors; the accumulated word vectors are normalized to obtain the title word vector; the word vectors corresponding to the words in the text word segmentation results are accumulated to obtain the accumulated word vector; the accumulated word vectors are normalized to obtain the text word vector; the title word vector and the text word vector are concatenated to form a text content feature vector.

[0114] Optionally, the normalization process mentioned above refers to mapping the data of each dimension in the accumulated word vector to a range of 0 to 1 or a range of -1 to 1. For example, the maximum and minimum values in the accumulated word vector are obtained, and the difference between the vector value and the minimum value in each dimension is divided by the difference between the maximum and minimum values to obtain the normalized vector value in each dimension, thereby completing the normalization process.

[0115] Step 320: Input the feature vector corresponding to the multimedia content into the timeliness classification model, and output the initial timeliness information of the multimedia content through the timeliness classification model.

[0116] Optionally, the above-mentioned time-effect classification model is a pre-trained machine learning model. Optionally, the training data of the above-mentioned time-effect classification model is the feature vector corresponding to the labeled multimedia content. The above-mentioned multimedia content can be the historical multimedia content in the information flow content service. The generation process of the feature vector of the multimedia content can refer to the above and will not be repeated here. The label information of the feature vector corresponding to the above-mentioned multimedia content can be the time-effect information of the multimedia content, such as the time-effect type or the target recommendation duration. The above-mentioned time-effect classification model is trained based on the feature vector corresponding to the labeled multimedia content.

[0117] In one possible implementation, the above-mentioned time-effectiveness classification model is a binary classification model, and the label of the feature vector corresponding to the above-mentioned multimedia content is the time-effectiveness type of the multimedia content, including a long time-effectiveness type and a short time-effectiveness type. Optionally, the time-effectiveness classification model is an Xgboost model. Xgboost is an optimized distributed gradient boosting library. The Xgboost model can more accurately determine the time-effectiveness type of multimedia content. Optionally, the time-effectiveness classification model is a logistic regression (LR) model. Optionally, the time-effectiveness classification model is a support vector machine (SVM).

[0118] The feature vector corresponding to the multimedia content is input into the timeliness classification model, and the timeliness classification model outputs the timeliness type of the multimedia content, or the target recommendation duration.

[0119] In an exemplary embodiment, step 210 in the previous embodiment may be implemented through steps 310 - 320 described above.

[0120] Step 330: perform information extraction processing on the multimedia content to obtain feature information of the multimedia content.

[0121] Optionally, the information extraction processing method is determined according to the form of the multimedia content, and then the information extraction processing corresponding to the form of the multimedia content is performed on the multimedia content to obtain feature information of the multimedia content.

[0122] In one possible implementation, the multimedia content is text content, such as an article, and the information extraction processing corresponding to the text content can be word segmentation processing, keyword extraction processing, or regular expression-based extraction processing. The feature information of the text content, such as keywords in the text content, is obtained through the above information extraction processing method.

[0123] In one possible implementation, the multimedia content is image content, such as pictures and videos, and the information extraction processing corresponding to the image content can be image recognition processing, image feature extraction, image text recognition, image attribute information extraction, etc. The feature information of the image content is obtained through the above information extraction processing method, such as key frame images, characters, objects, buildings and text information in the image content.

[0124] In one possible implementation, the multimedia content is audio content, such as songs or recordings, and the information extraction processing corresponding to the audio content can be audio recognition processing, audio-to-text conversion processing, and the like. The characteristic information of the audio content is obtained through the above-mentioned information extraction processing method, such as text information corresponding to the audio content.

[0125] In an exemplary embodiment, keyword extraction processing is performed on multimedia content to obtain keywords corresponding to the multimedia content.

[0126] Obtaining the text information corresponding to the multimedia content: The method for obtaining the text information corresponding to the multimedia content is described above and will not be repeated here.

[0127] Perform keyword extraction on the text information to obtain keywords corresponding to the multimedia content.

[0128] In an exemplary embodiment, weight information of at least one word in text information corresponding to multimedia content is determined; and based on the weight information of the at least one word, a keyword corresponding to the multimedia content is determined.

[0129] The weight information reflects the importance of a word in the text. Optionally, the weight information includes a weight value, which is positively correlated with the importance of the word in the text. Optionally, the weight information includes the degree of association between words, which is positively correlated with the importance of the word in the text.

[0130] In an exemplary embodiment, the keywords corresponding to the multimedia content include title keywords and text keywords, and the title keywords and text keywords can be obtained in the following manner.

[0131] Get the title content and body content in the text information corresponding to the multimedia content.

[0132] Perform regular extraction processing on the title content to obtain the title keywords. The above regular extraction processing refers to the processing based on obtaining the keywords of a specific part from the character string through regular expressions. In one possible implementation, since the title text content is relatively small, the keywords of a specific part can be obtained from the character string through regular expressions. For example, pattern 1 = (recently released|recently popular|soon to be released|soon to be presented|wonderfully presented|currently on the air) XXXX, pattern 2 = '《XXXX》', etc., where XXXX is the keyword of the specific part that the regular expression wants to obtain. Some obvious and easy-to-extract keywords are extracted through regular expressions as a backup. These keywords may reflect the entity objects associated with the multimedia content.

[0133] Determine weight information of at least one word in the text content. The weight information reflects the importance of the word in the text content.

[0134] Based on the weight information of the at least one word, a text keyword is determined.

[0135] In one possible implementation, multimedia content is textual. The content authors often use a wide variety of techniques during the writing and construction process, making it difficult to extract clean keywords using a standardized strategy. In this case, a keyword extraction process based on textual topic identification, such as the TextRank algorithm, can be used to extract keywords from the text.

[0136] Optionally, the title keywords and the text keywords are filtered to obtain filtered keywords.

[0137] In a possible real-time method, the screening method is to compare the title keywords and the text keywords with the preset keywords, retain the words in the title keywords and the text keywords that are similar or identical to the preset keywords, and obtain the screened keywords.

[0138] Step 340: Determine the entity object associated with the multimedia content based on the feature information of the multimedia content.

[0139] In an exemplary embodiment, entity objects associated with the multimedia content are determined based on keywords corresponding to the multimedia content.

[0140] In an exemplary embodiment, the embodiment of the present application further includes the following steps.

[0141] Step 390: Obtain entity object information.

[0142] The above-mentioned entity object information includes an entity object identifier and characteristic information associated with the entity object, such as the entity object name, the entity object abbreviation, or characteristic words associated with the entity object.

[0143] In a possible implementation, entity object information may be collected regularly from certain external professional websites, and the collected entity object information may be subjected to data integration and normalization processing to obtain an entity object information table, as shown in Table 1. Figure 1 The entity object information table shown is only for reference. The specific content in the table can be formulated according to the actual scenario, and the embodiment of the present application does not limit this. In an example, Figure 4 As shown in FIG, a schematic diagram of a display page of an entity object is exemplified, wherein the entity object 41 is a film or television work, and the entity object name is the film or television work name 42.

[0144] Table 1

[0145] Entity object name Alias time Hot XXXX Group XXXXX 2020 / X1 / 13 Wind XXX Quiet Heavy XXX Terminal 202X / 11 / 06 Hot XXXXX home XXXXXX driver 2020 / 11 / XX

[0146] Determine an entity object associated with the multimedia content based on the feature information of the multimedia content and the entity object information. Match the feature information of the multimedia content with the entity object information. If a match is found, determine that the entity object is an associated entity object for the multimedia content. Optionally, the number of entity objects associated with the multimedia content can be one or more.

[0147] Optionally, entity objects associated with the multimedia content are determined based on the filtered keywords.

[0148] The filtered keywords are matched with the keywords in the entity object information of each entity object. If the match is successful, the entity object associated with the multimedia content can be determined.

[0149] In an exemplary embodiment, step 220 in the previous embodiment may be implemented through steps 330 - 340 described above.

[0150] Step 350: Obtain the popularity associated data of the entity object.

[0151] Optionally, at least one kind of popularity-related data of the entity object is obtained, such as the number, exposure rate, click-through rate, appreciation rate, page views and other popularity-related data of the multimedia content associated with the entity object.

[0152] In an exemplary embodiment, the above step 350 can be implemented through the following steps ( 351 - 353 ).

[0153] Step 351: Obtain the number of multimedia contents associated with the entity object among the multimedia contents entering the content recommendation library within the target time period.

[0154] Step 352: Obtain the total amount of multimedia content that enters the content recommendation library within the target period.

[0155] Step 353 : determining the popularity association data of the entity object according to the number of multimedia contents associated with the entity object and the total amount of multimedia contents.

[0156] The proportion of multimedia content associated with the entity object in the content database is used as the popularity association data of the entity object, reflecting the proportion of multimedia content associated with the entity object in the content database during the target period. The above proportion can reflect the popularity information of the entity object.

[0157] Step 360 : Determine the popularity information of the entity object based on the popularity association data of the entity object and the popularity threshold condition corresponding to the popularity association data.

[0158] If the proportion of multimedia content associated with the entity object in the content database is greater than or equal to the target proportion threshold, it is determined that the popularity type of the entity object is a recent popular type.

[0159] The heat association data may be compared with a heat threshold in a heat threshold condition corresponding to the heat association data, and the heat information of the entity object, such as the heat type of the entity object, may be determined according to the comparison result.

[0160] Optionally, the above-mentioned heat-related data include heat-related data such as the number, exposure rate, click-through rate, appreciation rate, and page views of multimedia content associated with the entity object; correspondingly, the heat threshold conditions include heat threshold conditions such as the number of multimedia content associated with the entity object reaches the number threshold, the exposure rate reaches the exposure rate threshold, the click-through rate reaches the click rate threshold, the appreciation rate reaches the appreciation rate threshold, and the page views reach the page views threshold.

[0161] In an exemplary embodiment, step 230 in the previous embodiment can be implemented through steps 350 - 360 as described above.

[0162] In one possible implementation, the multimedia content is multimedia content that enters the content recommendation library in real time, and the popularity information of the entity object can be stored in the entity object information table. The steps 350-360 can be executed in advance to determine the popularity information of each entity object in the entity object information table, and the popularity information of each entity object in the entity object information table can be updated. In this way, after the entity object of the multimedia content is identified, the popularity information of the entity object can be obtained immediately without further calculation, thereby improving the running speed.

[0163] Step 370 : Determine target timeliness information of the multimedia content based on the popularity information of the entity object and the initial timeliness information of the multimedia content.

[0164] Optionally, entity objects whose popularity information meets preset popularity conditions are classified as a recent hot type. Optionally, multimedia content whose timeliness information meets preset timeliness conditions is classified as a long timeliness type. If the multimedia content is of a long timeliness type and the entity object associated with the multimedia content is of a recent hot type, the timeliness type of the multimedia content is changed from a long timeliness type to a medium timeliness type or a short timeliness type, or the target push duration of the multimedia content is shortened.

[0165] Step 380: Push multimedia content according to the target timeliness information.

[0166] For the explanation of steps 370 and 380, please refer to the introduction in the previous embodiment and will not be repeated here.

[0167] In summary, the technical solution provided by the embodiment of the present application determines the timeliness of multimedia content through a timeliness classification model, extracts entity objects associated with the multimedia content through information extraction processing, and then updates the timeliness of the multimedia content in combination with the popularity of the entity objects, and corrects the timeliness of multimedia content that has been determined to be long-term to medium-short-term, thereby preventing such multimedia content from being pushed for a long time in the information flow service, affecting the user experience, improving the accuracy of the timeliness of the multimedia content, reducing expired content in the information flow, reducing the operating pressure of the server, and effectively improving resource utilization.

[0168] Below, the method provided by this application is explained in conjunction with specific scenarios. The embodiment provided by this application can solve the timeliness expiration problem of film and television variety articles by matching the identified entity objects with the external film and television variety works library in the information flow content service scenario, identify the expired film and television variety articles in the recommendation pool (i.e., the content recommendation library) and remove them in time to optimize the user experience. Different from the traditional timeliness identification strategy based on keywords, dates, etc., the timeliness classification model, entity recognition, film and television works library matching and other strategies are used to mark the articles with appropriate timeliness results.

[0169] Generally speaking, we first use a timeliness classification model to assign articles to short or long timeliness. We then use entity recognition strategies to discover film, television, and variety show content related to these articles. Finally, by matching these entities with an external database of film, television, and variety show content, we recall film, television, and variety show articles that have been assigned long timeliness and reclassify them as medium- to short-term timeliness articles. This prevents these articles from being recommended long-term, impacting user experience. Detailed instructions are provided below.

[0170] 1. Get the target article content. The target article is the article that enters the content recommendation library in real time.

[0171] In one example, Figure 5 As shown, it exemplarily shows a flow chart of determining the timeliness of an article. Figure 5 A brief flow of the following steps 2-5 is shown in FIG.

[0172] 2. Perform word segmentation on the title to obtain the title segmentation result; determine the word vectors corresponding to the words in the title segmentation result; accumulate the word vectors corresponding to the words in the title segmentation result to obtain the accumulated word vectors; normalize the accumulated word vectors to obtain the title word vector.

[0173] 3. Perform word segmentation on the main text to obtain the main text word segmentation result; determine the word vectors corresponding to the words in the main text word segmentation result; accumulate the word vectors corresponding to the words in the main text word segmentation result to obtain the accumulated word vectors; normalize the accumulated word vectors to obtain the main text word vector.

[0174] 4. Concatenate the title word vector and the body text word vector to form the article feature vector corresponding to the target article.

[0175] 5. Input the article feature vector into the long / short timeliness classification model, which then outputs the timeliness type of the target article. Optionally, the timeliness type includes a long timeliness type and a short timeliness type. Optionally, the long / short timeliness classification model is a binary classification model.

[0176] In one example, if Figure 6 As shown, it exemplarily shows a schematic diagram of determining variety shows and film and television works associated with an article. Figure 6 A brief flow of the following steps 6-9 is shown in FIG.

[0177] 6. Perform regular expression extraction on the title of the target article to obtain the title keywords.

[0178] 7. Perform keyword extraction on the main text of the target article to obtain the main text keywords.

[0179] 8. Filter the title keywords and text keywords to obtain the filtered keywords.

[0180] 9. Based on the filtered keywords, determine the target film, television and variety show works associated with the target article.

[0181] 10. Count the total number N of film, television and variety show articles that enter the content recommendation library during the target period, where N is a positive integer.

[0182] 11. Count the number of film and television variety show articles that are associated with the target film and television variety show in the content recommendation library within the target period, M, where M is a positive integer. Optionally, the target period is the past month.

[0183] 12. If M / N is greater than or equal to the target ratio threshold, the target film and television variety show is determined to be a recent popular film and television variety show. Optionally, the target ratio threshold is 0.02. The above process of determining the popularity type of the target film and television variety show can be performed in advance. The popularity information of the target film and television variety show can be recorded and updated in an external film and television variety show library for direct access.

[0184] 13. If the target film, television or variety show work associated with the target article is a recently popular film, television or variety show work, and the target article is of the long-term validity type, change the validity type of the target article to a short-term validity type.

[0185] To sum up, the technical solution provided by the embodiment of the present application determines the timeliness of film and television variety articles through a long / short timeliness classification model, extracts film and television variety works related to the film and television variety articles through information extraction processing, and then updates the timeliness of the film and television variety articles based on the popularity of the film and television variety works. The timeliness of the film and television variety articles that have been determined to have a long timeliness is corrected to a medium or short timeliness, thereby avoiding such expired film and television variety articles from being pushed for a long time in the information flow service, affecting the user experience, improving the accuracy of the timeliness of film and television variety articles, reducing expired content in the information flow, reducing server operating pressure, and effectively improving resource utilization.

[0186] The following are device embodiments of the present application, which can be used to implement the method embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the present application.

[0187] Please refer to Figure 7 , which shows a block diagram of a multimedia content push device provided by an embodiment of the present application. The device has the function of implementing the above-mentioned multimedia content push method, and the function can be implemented by hardware or by hardware executing corresponding software. The device can be a computer device or can be set in a computer device. The device 700 may include: an entity object determination module 710, a heat information acquisition module 720, a timeliness information determination module 730 and a content push module 740.

[0188] The entity object determination module 710 is configured to determine entity objects associated with multimedia content.

[0189] The popularity information acquisition module 720 is used to acquire the popularity information of the entity object.

[0190] The timeliness information determination module 730 is configured to determine target timeliness information of the multimedia content based on the popularity information of the entity object.

[0191] The content push module 740 is configured to push the multimedia content according to the target timeliness information.

[0192] In an exemplary embodiment, the apparatus further includes: an initial time-effectiveness acquisition module.

[0193] An initial timeliness acquisition module, configured to acquire initial timeliness information of the multimedia content;

[0194] The popularity information includes a popularity type, and the timeliness information determination module 740 is used to:

[0195] When the popularity type of the entity object is a recent hot type and the target push duration corresponding to the initial timeliness information is greater than or equal to a target duration threshold, the initial timeliness information of the multimedia content is modified to obtain the target timeliness information of the multimedia content, the proportion of multimedia content associated with the entity object of the recent hot type in the content recommendation library is greater than or equal to the target proportion threshold, and the target recommendation duration corresponding to the target timeliness information is less than the initial recommendation duration corresponding to the initial timeliness information.

[0196] In an exemplary embodiment, the initial time acquisition module is used to:

[0197] Obtaining text information corresponding to the multimedia content;

[0198] Performing word segmentation processing on the text information to obtain a word segmentation result;

[0199] Determine the word vector corresponding to the word in the word segmentation result;

[0200] Determining a feature vector corresponding to the multimedia content based on the word vector corresponding to the word;

[0201] The feature vector corresponding to the multimedia content is input into a time-effectiveness classification model, and the initial time-effectiveness information of the multimedia content is output through the time-effectiveness classification model.

[0202] In an exemplary embodiment, the content push module 750 is configured to:

[0203] Pushing the multimedia content when the cumulative push duration corresponding to the multimedia content does not reach the target recommended duration corresponding to the target timeliness information;

[0204] When the accumulated push duration corresponding to the multimedia content reaches the target recommendation duration corresponding to the target timeliness information, the multimedia content is stopped from being pushed.

[0205] In an exemplary embodiment, the heat information acquisition module 730 is used to:

[0206] Obtaining the hotness associated data of the entity object;

[0207] Based on the heat association data of the entity object and a heat threshold condition corresponding to the heat association data, heat information of the entity object is determined.

[0208] In an exemplary embodiment, the popularity information acquisition module 730 includes: a quantity acquisition unit, a total amount acquisition unit, and a data determination unit.

[0209] The quantity acquisition unit is used to acquire the quantity of multimedia content associated with the entity object from the multimedia content entering the content recommendation library within a target time period.

[0210] The total amount acquisition unit is used to acquire the total amount of multimedia content entering the content recommendation library within the target time period.

[0211] The data determining unit is configured to determine the popularity associated data of the entity object according to the number of multimedia contents associated with the entity object and the total amount of the multimedia contents.

[0212] In an exemplary embodiment, the entity object determination module 720 includes: a keyword extraction unit and an entity object determination unit.

[0213] The keyword extraction unit is used to perform keyword extraction processing on the multimedia content to obtain keywords corresponding to the multimedia content.

[0214] The entity object determining unit is configured to determine the entity object associated with the multimedia content according to the keyword corresponding to the multimedia content.

[0215] In an exemplary embodiment, the keyword extraction unit includes a weight determination subunit and a keyword determination subunit.

[0216] The weight determination subunit is configured to determine weight information of at least one word in the text information corresponding to the multimedia content, wherein the weight information reflects the importance of the word in the text information.

[0217] The keyword determination subunit is configured to determine a keyword corresponding to the multimedia content based on the weight information of the at least one word.

[0218] In an exemplary embodiment, the keywords corresponding to the multimedia content include title keywords and text keywords, and the keyword extraction unit further includes: a content acquisition subunit and a keyword extraction subunit.

[0219] The content acquisition subunit is used to acquire the title content and the body content in the text information corresponding to the multimedia content.

[0220] The keyword extraction subunit is used to perform regular extraction processing on the title information to obtain the title keywords.

[0221] The weight determination subunit is further configured to determine weight information of at least one word in the main text content, wherein the weight information reflects the importance of the word in the main text content.

[0222] The keyword determination subunit is further configured to determine the text keyword based on the weight information of the at least one word.

[0223] To sum up, the technical solution provided by the embodiment of the present application first determines the timeliness of the multimedia content and the physical objects associated with the multimedia content, then determines the popularity of the physical objects associated with the multimedia content, and then combines the timeliness of the multimedia content and the popularity of the physical objects associated with the multimedia content to update the timeliness of the multimedia content to improve the accuracy of the timeliness of the multimedia content, and finally pushes the multimedia content according to the updated timeliness to reduce expired content in the information flow, reduce the operating pressure of the server, and effectively improve resource utilization.

[0224] It should be noted that the apparatus provided in the above embodiments, when implementing its functions, is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0225] Please refer to Figure 8 , which shows a block diagram of a computer device provided by an embodiment of the present application. The computer device may be a server for executing the above-mentioned method for pushing multimedia content. Specifically:

[0226] Computer device 800 includes a central processing unit (CPU) 801, a system memory 804 including a random access memory (RAM) 802 and a read-only memory (ROM) 803, and a system bus 805 connecting system memory 804 and CPU 801. Computer device 800 also includes a basic input / output system (I / O system) 806 that facilitates information transfer between various components within the computer, and a mass storage device 807 for storing an operating system 813, application programs 814, and other program modules 812.

[0227] The basic input / output system 806 includes a display 808 for displaying information and an input device 809, such as a mouse and keyboard, for user input. Both the display 808 and the input device 809 are connected to the central processing unit 801 via an input / output controller 810 connected to the system bus 805. The basic input / output system 806 may also include an input / output controller 810 for receiving and processing input from a variety of other devices, such as a keyboard, mouse, or electronic stylus. Similarly, the input / output controller 810 also provides output to a display screen, printer, or other types of output devices.

[0228] The mass storage device 807 is connected to the central processing unit 801 via a mass storage controller (not shown) connected to the system bus 805. The mass storage device 807 and its associated computer-readable media provide non-volatile storage for the computer device 800. In other words, the mass storage device 807 may include a computer-readable medium (not shown) such as a hard disk or a CD-ROM (Compact Disc Read-Only Memory) drive.

[0229] Without loss of generality, computer-readable media may include computer storage media and communication media. Computer storage media include volatile and non-volatile, removable and non-removable media implemented by any method or technology for storing information such as computer-readable instructions, data structures, program modules or other data. Computer storage media include RAM, ROM, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically Erasable Programmable Read Only Memory), flash memory or other solid-state storage technologies, CD-ROM, DVD (Digital Video Disc) or other optical storage, tape cassettes, magnetic tape, disk storage or other magnetic storage devices. Of course, those skilled in the art will appreciate that computer storage media are not limited to the above-mentioned ones. The above-mentioned system memory 804 and mass storage device 807 can be collectively referred to as memory.

[0230] According to various embodiments of the present application, the computer device 800 may also be connected to a remote computer on a network such as the Internet for operation. That is, the computer device 800 may be connected to a network 812 via a network interface unit 811 connected to the system bus 805, or the network interface unit 811 may be used to connect to other types of networks or remote computer systems (not shown).

[0231] The memory further includes a computer program, which is stored in the memory and configured to be executed by one or more processors to implement the above-mentioned method for pushing multimedia content.

[0232] In an exemplary embodiment, a computer-readable storage medium is also provided, in which at least one instruction, at least one program, a code set or an instruction set is stored. When the at least one instruction, the at least one program, the code set or the instruction set is executed by a processor, the method for pushing the above-mentioned multimedia content is implemented.

[0233] Optionally, the computer-readable storage medium may include: ROM (Read Only Memory), RAM (Random Access Memory), SSD (Solid State Drives), or an optical disk, etc. Among them, the random access memory may include ReRAM (Resistance Random Access Memory) and DRAM (Dynamic Random Access Memory).

[0234] In an exemplary embodiment, a computer program product or computer program is also provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the above-described method for pushing multimedia content.

[0235] It should be understood that the "multiple" mentioned in this article refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent three situations: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the previous and subsequent associated objects are in an "or" relationship. In addition, the step numbers described in this article only illustrate a possible execution sequence between the steps. In some other embodiments, the above steps may not be executed in the order of the numbers, such as two steps with different numbers are executed at the same time, or two steps with different numbers are executed in the opposite order to the diagram. The embodiments of the present application do not limit this.

[0236] The above description is merely an exemplary embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for pushing multimedia content, characterized in that: The method comprises: Determining entity objects associated with multimedia content; the entity objects are distinguishable and identifiable objects or things; Acquire initial timeliness information of the multimedia content, and acquire popularity information of the entity object; the popularity information includes a popularity type; If the popularity type of the entity object is a recent hot type, and the initial push duration corresponding to the initial timeliness information is greater than or equal to a target duration threshold, modifying the initial timeliness information of the multimedia content to obtain target timeliness information of the multimedia content, a proportion of multimedia content associated with the entity object of the recent hot type in the content database is greater than or equal to a target proportion threshold, and a target recommendation duration corresponding to the target timeliness information is less than an initial recommendation duration corresponding to the initial timeliness information; Pushing the multimedia content according to the target timeliness information; Pushing the multimedia content according to the target timeliness information includes: removing the multimedia content from the content database when the accumulated push duration corresponding to the multimedia content reaches the target recommendation duration corresponding to the target timeliness information.

2. The method according to claim 1, characterized in that The obtaining of the initial timeliness information of the multimedia content includes: Obtaining text information corresponding to the multimedia content; Performing word segmentation processing on the text information to obtain a word segmentation result; Determine the word vector corresponding to the word in the word segmentation result; Determining a feature vector corresponding to the multimedia content based on the word vector corresponding to the word; The feature vector corresponding to the multimedia content is input into a time-effectiveness classification model, and the initial time-effectiveness information of the multimedia content is determined by the time-effectiveness classification model.

3. The method according to claim 1, characterized in that The pushing of the multimedia content according to the target timeliness information further includes: When the accumulated push duration corresponding to the multimedia content does not reach the target recommended duration corresponding to the target timeliness information, the multimedia content is pushed.

4. The method according to claim 1, wherein The acquiring of the popularity information of the entity object includes: Obtaining the hotness associated data of the entity object; Based on the heat association data of the entity object and a heat threshold condition corresponding to the heat association data, heat information of the entity object is determined.

5. The method according to claim 4, characterized in that The acquiring of the popularity associated data of the entity object includes: Obtaining the number of multimedia contents associated with the entity object from the multimedia contents entering the content recommendation library within a target period; Obtaining the total amount of multimedia content entering the content recommendation library within the target time period; The popularity association data of the entity object is determined according to the number of multimedia contents associated with the entity object and the total amount of the multimedia contents.

6. The method according to claim 1, characterized in that The determining of the entity object associated with the multimedia content includes: Performing keyword extraction processing on the multimedia content to obtain keywords corresponding to the multimedia content; Determine entity objects associated with the multimedia content based on keywords corresponding to the multimedia content.

7. The method according to claim 6, characterized in that The performing keyword extraction processing on the multimedia content to obtain keywords corresponding to the multimedia content includes: Determining weight information of at least one word in the text information corresponding to the multimedia content, wherein the weight information reflects the importance of the word in the text information; Based on the weight information of the at least one word, a keyword corresponding to the multimedia content is determined.

8. The method according to claim 6, characterized in that The keywords corresponding to the multimedia content include title keywords and text keywords, and the keyword extraction process for the multimedia content to obtain the keywords corresponding to the multimedia content further includes: Obtaining the title content and the body content in the text information corresponding to the multimedia content; Performing regular expression extraction processing on the title content to obtain the title keywords; Determining weight information of at least one word in the text content, where the weight information reflects the importance of the word in the text content; The text keyword is determined based on the weight information of the at least one word.

9. A multimedia content push device, characterized in that: The device comprises: An entity object determination module, configured to determine entity objects associated with multimedia content; the entity objects are distinguishable and identifiable objects or things; A popularity information acquisition module, configured to acquire initial timeliness information of the multimedia content and popularity information of the entity object; the popularity information includes a popularity type; a timeliness information determination module configured to modify the initial timeliness information of the multimedia content to obtain target timeliness information of the multimedia content if the popularity type of the entity object is a recent hot type and the initial push duration corresponding to the initial timeliness information is greater than or equal to a target duration threshold, the proportion of multimedia content associated with the entity object of the recent hot type in the content database is greater than or equal to a target proportion threshold, and the target recommendation duration corresponding to the target timeliness information is less than the initial recommendation duration corresponding to the initial timeliness information; A content push module is used to push the multimedia content according to the target timeliness information; wherein, pushing the multimedia content according to the target timeliness information includes: when the cumulative push duration corresponding to the multimedia content reaches the target recommended duration corresponding to the target timeliness information, the multimedia content is removed from the content database.

10. A computer device, comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, and wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the method for pushing multimedia content according to any one of claims 1 to 8.

11. A computer-readable storage medium, wherein the storage medium stores at least one instruction, at least one program, a code set, or an instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the method for pushing multimedia content according to any one of claims 1 to 8.

Images