Recommendation Method, Device, Server, and Storage Medium for Text Information

Abstract

The present disclosure provides a method, apparatus, server, and storage medium for recommending text information, which can be applied to various scenarios such as cloud technology, artificial intelligence, intelligent transportation, and assisted driving. The method includes: obtaining a picture associated with the content of the text information to be recommended according to the text category label corresponding to the text information to be recommended; obtaining the text elements corresponding to the text information to be recommended; generating cover information corresponding to the text information to be recommended based on the picture and the text elements; and recommending the cover information to a target object. The present disclosure obtains a picture associated with the content of the text information to be recommended according to the text category label of the information to be recommended, and obtains the text elements corresponding to the recommended text information, and then combines the picture associated with the content of the text information to be recommended and the text elements into cover information. The cover information is not in a single text form, but in a form combining pictures and texts, with a richer information form and enhanced interest of the recommended information.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and particularly to a method, apparatus, server, and storage medium for recommending text information. Background Art

[0002] With the development of Internet technology, the user group is becoming younger and younger, and more and more underage users start to obtain text information from the Internet. Since the amount of information on the current Internet is extremely large and the content of text information is diverse, in order to improve the efficiency of underage users in obtaining text information, it is necessary to recommend text information to underage users.

[0003] When recommending text information to underage users, the related technologies mainly adopt the following methods: depicting the object data of underage users based on the demographic information, social relationships, preference habits, consumption behaviors, etc. of underage users; screening out text information that meets the requirements from the Internet according to the object data of the underage user; and recommending the text information to the underage user.

[0004] However, the form of the text information recommended by the related technologies is generally in the form of pure text, and the information form is relatively single. Summary of the Invention

[0005] Embodiments of the present disclosure provide a method, apparatus, server, and storage medium for recommending text information, which can enrich the form of the recommended information. The technical solutions are as follows:

[0006] On the one hand, a method for recommending text information is provided. The method includes:

[0007] Responding to an information acquisition request of a target object, and acquiring text information to be recommended;

[0008] Acquiring a text category label corresponding to the text information to be recommended;

[0009] According to the text category label, acquiring a picture associated with the content of the text information to be recommended;

[0010] Acquiring text elements corresponding to the text information to be recommended, where the text elements include at least one of a text title or a text abstract;

[0011] Based on the picture and the text elements, generating cover information corresponding to the text information to be recommended;

[0012] Recommending the cover information to the target object.

[0013] On the other hand, a device for recommending text information is provided. The device includes:

[0014] The first acquisition module is configured to acquire the text information to be recommended in response to an information acquisition request of a target object;

[0015] The first acquisition module is further configured to acquire a text category label corresponding to the text information to be recommended;

[0016] The first acquisition module is further configured to acquire a picture associated with the content of the text information to be recommended according to the text category label;

[0017] The first acquisition module is further configured to acquire text elements corresponding to the text information to be recommended, where the text elements include at least one of a text title or a text abstract;

[0018] The generation module is configured to generate cover information corresponding to the text information to be recommended based on the picture and the text elements;

[0019] The recommendation module is configured to recommend the cover information to the target object.

[0020] On the other hand, a server is provided, where the server includes a processor and a memory, and at least one program code is stored in the memory. The at least one program code is loaded and executed by the processor to implement the recommendation method of the text information on one hand.

[0021] On the other hand, a computer-readable storage medium is provided, where at least one program code is stored in the storage medium. The at least one program code is loaded and executed by a processor to implement the recommendation method of the text information on one hand.

[0022] On the other hand, a computer program product is provided, where the computer program product includes computer program code. The computer program code is stored in a computer-readable storage medium, and a processor of a server reads the computer program code from the computer-readable storage medium. The processor executes the computer program code so that the server executes the recommendation method of the text information on one hand.

[0023] The beneficial effects brought by the technical solution provided by the embodiments of the present disclosure are:

[0024] Obtain an image associated with the content of the text category label of the information to be recommended, where the image associated with the content of the information to be recommended can visually and vividly display the external manifestation of the object described by the information to be recommended to the user. Then, obtain the text elements corresponding to the recommended text information, where the text elements include at least one of the text title or text abstract of the information to be recommended, and can briefly display the core content of the information to be recommended to the user. Furthermore, generate cover information in the form of a combination of an image associated with the content of the information to be recommended and the text elements. Compared with the information in a single text form, the cover information has a richer information form and stronger interest, greatly increasing the probability that the user views the text information. Description of the Drawings

[0025] To more clearly illustrate the technical solutions in the embodiments of the present disclosure, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present disclosure. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0026] Figure 1 It is a schematic diagram of the implementation environment involved in a method for recommending text information provided by an embodiment of the present disclosure;

[0027] Figure 2 It is a flowchart of a method for recommending text information provided by an embodiment of the present disclosure;

[0028] Figure 3 It is a flowchart of another method for recommending text information provided by an embodiment of the present disclosure;

[0029] Figure 4 It is a schematic diagram of the associated image searched by an embodiment of the present disclosure;

[0030] Figure 5 It is a schematic diagram of the generation process of the cover information in an embodiment of the present disclosure;

[0031] Figure 6 It is the display effect diagram of the information provided by the related art;

[0032] Figure 7 It is a flowchart of a method for training a text category recognition model provided by an embodiment of the present disclosure;

[0033] Figure 8 It is a schematic diagram of a process for recommending text information provided by an embodiment of the present disclosure;

[0034] Figure 9 It is a schematic diagram of the structure of a device for recommending text information provided by an embodiment of the present disclosure;

[0035] Figure 10 A server for text information recommendation shown according to an exemplary embodiment. Detailed implementation manners

[0036] To make the objectives, technical solutions and advantages of the present disclosure clearer, the following will further describe the embodiments of the present disclosure in detail with reference to the accompanying drawings.

[0037] It can be understood that the terms "each", "multiple" and "any one" used in the embodiments of the present disclosure, multiple includes two or more, each refers to each one in the corresponding multiple, and any one refers to any one in the corresponding multiple. For example, multiple words include 10 words, and each word refers to each of these 10 words, and any one word refers to any one of the 10 words.

[0038] In the detailed implementation manners of the present disclosure, relevant data of users are involved. When the embodiments of the present disclosure are applied to specific products or technologies, user permission or consent needs to be obtained, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions.

[0039] First, before implementing the embodiments of the present disclosure, the nouns involved in the embodiments of the present disclosure are first explained.

[0040] Object data refers to a labeled portrait abstracted according to user demographic information, social relationships, preference habits, consumption behaviors and other information. The core work of constructing object data is to label users, and some of the labels are directly obtained from user behavior data, and some are mined through a series of algorithms or rules.

[0041] Text information classification refers to the process in which an electronic device automatically classifies input text information according to a certain category system through an algorithm. When the electronic device automatically classifies and marks a text information set according to a certain classification system or standard, a relationship model between text category features and text category labels can be found according to a sample text information set that has been labeled with text category labels. This relationship model is the text category recognition model described in the embodiments of the present disclosure, and then the learned relationship model is used to identify the text category of new text information.

[0042] OCR (Optical Character Recognition) recognition refers to the process in which an electronic device directly converts the text content on a picture or photo into editable text based on OCR technology.

[0043] A text abstract refers to the automatic extraction of a text abstract from an original document by an electronic device. This text abstract is a simple and coherent short passage that comprehensively and accurately reflects the central content of a document.

[0044] Introduce the technologies involved in the embodiments of the present disclosure.

[0045] Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a way similar to human intelligence. Artificial intelligence also studies the design principles and implementation methods of various intelligent machines to enable the machines to have the functions of perception, reasoning, and decision-making.

[0046] Artificial intelligence technology is an interdisciplinary subject with a wide range of fields involved, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. The software technologies of artificial intelligence mainly include several major directions such as computer vision technology, natural language processing technology, and machine learning / deep learning.

[0047] Computer Vision Technology (CV): Computer vision is a science that studies how to make machines "see". More specifically, it refers to using cameras and computers to replace human eyes for tasks such as target recognition and measurement in machine vision, and further performing image processing to make the computer-processed images more suitable for human eye observation or transmission to instruments for detection. As a scientific discipline, computer vision studies related theories and technologies and attempts to establish artificial intelligence systems that can obtain information from images or multi-dimensional data. Computer vision technology usually includes technologies such as image processing, image recognition, image semantic understanding, image retrieval, OCR, video processing, video semantic understanding, video content / behavior recognition, three-dimensional object reconstruction, 3D technology, virtual reality, augmented reality, simultaneous localization and mapping, and also includes common biometric recognition technologies such as face recognition and fingerprint recognition.

[0048] Natural Language Processing (NLP) is an important direction in the fields of computer science and artificial intelligence. It studies various theories and methods that can achieve effective communication between humans and computers in natural language. Natural language processing is a science that integrates linguistics, computer science, and mathematics. Therefore, the research in this field will involve natural language, that is, the language people use in daily life, so it has a close connection with the research of linguistics. Natural language processing technologies usually include text processing, semantic understanding, machine translation, robot question answering, knowledge graphs, and other technologies.

[0049] Machine Learning (ML) is an interdisciplinary subject that involves multiple disciplines such as probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve their own performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent, and its applications cover all fields of artificial intelligence. Machine learning and deep learning usually include technologies such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and rote learning.

[0050] Cloud technology refers to a hosting technology that unifies a series of resources such as hardware, software, and networks within a wide area network or local area network to achieve data computing, storage, processing, and sharing. Cloud technology is the general term for network technology, information technology, integration technology, management platform technology, application technology, etc. based on the cloud computing business model, which can form a resource pool, be used on demand, and be flexible and convenient.

[0051] Please refer to Figure 1 , which shows the implementation environment involved in the method for recommending text information provided by the embodiments of the present disclosure. The implementation environment includes: a terminal 101 and a server 102.

[0052] The terminal 101 can be an electronic device such as a smart phone, a tablet computer, a laptop computer, a desktop computer, an e-book reader, a multimedia playback device, a wearable device, a PC (Personal Computer, personal computer), a smart home appliance, a vehicle-mounted terminal, a smart voice interaction device, etc. A text information acquisition application is installed in the terminal 101. Based on this text information acquisition application, the terminal 101 can send an information acquisition request to the server 101, and then display the cover information of the text information recommended by the server 102 for this information acquisition request.

[0053] Server 102 is used to provide background services for the text information acquisition application in terminal 101. For example, server 102 can be the background server of the text information acquisition application. Server 102 can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. This server 102 is capable of receiving the information acquisition request sent by terminal 101 and acquiring the text information to be recommended based on this information acquisition request. A text category recognition model and a text element generation model are installed and run in server 102. This text category recognition model can identify the text category label corresponding to the text information to be recommended, and this text element generation model can generate text elements for the text information to be recommended. Server 101 can also acquire a picture associated with the content of the text information to be recommended based on the identified text category label, and then combine this picture with the generated text elements into cover information, so as to recommend this cover information to terminal 101.

[0054] Terminal 101 and server 102 can communicate through network 103. This network 103 can be a wired network or a wireless network.

[0055] An embodiment of the present disclosure provides a method for recommending text information, and this method is executed by the server shown above Figure 1 This embodiment of the present disclosure can be applied to various scenarios, including but not limited to various scenarios such as cloud technology, artificial intelligence, intelligent transportation, and assisted driving. Refer to Figure 2 , the method flow provided by the embodiment of the present disclosure includes:

[0056] 201. In response to the information acquisition request of the target object, acquire the text information to be recommended.

[0057] Among them, the target object is an object with a need for text information acquisition. The information acquisition request includes the object information of the target object, the information acquisition time, etc. The object information of this target object includes the account, age, gender, native place, address, education level, etc. of the target object. When receiving the information acquisition request of the target object, in response to this information acquisition request of the target object, the server acquires the information to be recommended from the network based on the object data corresponding to the target object.

[0058] 202. Acquire the text category label corresponding to the text information to be recommended.

[0059] Among them, the text category label is used to identify the category to which the text information to be recommended belongs, and the text category label can be animal science popularization, plant science popularization, games, emotions, etc.

[0060] 203. According to the text category label, obtain the pictures associated with the content of the text information to be recommended.

[0061] In the embodiments of the present disclosure, the text category label is actually the content keyword of the text information to be recommended, which can reflect the object described by the text information to be recommended. Based on this text category label, the server can obtain the pictures associated with the content of the text information to be recommended.

[0062] 204. Obtain the text elements corresponding to the text information to be recommended.

[0063] Among them, the text elements are used to reflect the core content of the text information to be recommended. The text elements include at least one of the text title or the text abstract. That is to say, the text elements can be the text title, the text elements can be the text abstract, and the text elements can also be the text title and the text abstract.

[0064] 205. Based on the pictures and text elements, generate the cover information corresponding to the text information to be recommended.

[0065] Based on the pictures associated with the content of the text information to be recommended and the text elements, the server combines the pictures with the text elements to generate the cover information corresponding to the recommended text information. The cover information is in the form of a combination of pictures and texts, and the information form is richer. For different contents included in the text elements, the way the server generates the cover information is also different. When the text element is the text title, the server generates the cover information corresponding to the text information to be recommended based on the text title and the pictures associated with the content of the text information to be recommended; when the text element is the text abstract, the server generates the cover information corresponding to the text information to be recommended based on the text abstract and the pictures associated with the content of the text information to be recommended; when the text element includes the text title and the text abstract, the server generates the cover information corresponding to the text information to be recommended based on the text title, the text abstract and the pictures associated with the content of the text information to be recommended.

[0066] 206. Recommend the cover information to the target object.

[0067] The server recommends the generated cover information to the target object to meet the target object's need to obtain text information. Compared with the text information in pure text form, the cover information has richer content and form, stronger interest, and increases the stickiness of the target object to the text information acquisition application.

[0068] The method provided by the embodiments of the present disclosure obtains a picture associated with the content of the to-be-recommended text information based on the text category label of the to-be-recommended information. The picture associated with the content of the to-be-recommended text information can visually and vividly display the external manifestation of the object described by the to-be-recommended text information to the user. Then, the text elements corresponding to the recommended text information are obtained. The text elements include at least one of the text title or text summary of the to-be-recommended text information, and can briefly display the core content of the to-be-recommended information to the user. Furthermore, a cover information in the form of a combination of a picture associated with the content of the to-be-recommended text information and text elements is generated. Compared with the information in a single text form, the cover information has a richer information form and stronger interest, and greatly improves the probability of the user viewing the text information.

[0069] The embodiments of the present disclosure provide a method for recommending text information, which is executed by the server shown above Figure 1 The embodiments of the present disclosure can be applied to various scenarios, including but not limited to cloud technology, artificial intelligence, intelligent transportation, assisted driving, etc. Refer to Figure 3 , the method flow provided by the embodiments of the present disclosure includes:

[0070] 301. In response to an information acquisition request of a target object, the server acquires to-be-recommended text information.

[0071] When detecting an open operation of a text information acquisition application by the target object, or detecting a refresh operation of the interface of the text information acquisition application by the target object, the terminal generates an information acquisition request of the target object, and then sends the information acquisition request to the server to acquire the to-be-recommended text information recommended by the server to the target object, and then meets the information acquisition requirement of the target object by displaying the recommended text information. Among them, the terminal includes but is not limited to a smart phone, a computer, a smart voice interaction device, a smart home appliance, a vehicle-mounted terminal, etc.

[0072] When receiving the information acquisition request of the target object, in response to the information acquisition request, the server acquires to-be-recommended text information for the target object. In a possible implementation manner, the server acquires the object data of the target object, and then based on the object data, acquires the to-be-recommended text information that the target object may be interested in from the network. In another possible implementation manner, the server can acquire the text information with a higher attention degree from the network as the to-be-recommended text information according to the attention information of the text information (such as click-through rate, repost volume, like volume, share volume, etc.). In another possible implementation manner, the server can acquire the latest published text information from the network as the to-be-recommended text information according to the publication time of the text information.

[0073] 302. The server acquires the text category label corresponding to the to-be-recommended text information.

[0074] In an embodiment of the present disclosure, a text category recognition model is installed and run on a server. The text category recognition model is used to identify the text category label corresponding to any text information. Based on the text category recognition model, the server can obtain the text category label corresponding to the text information to be recommended, so that not only can the text information to be recommended obtained be screened to avoid recommending text information inappropriate for the target object to the target object, but also based on the text category label corresponding to the text information to be recommended, pictures associated with the content of the recommended text information can be obtained, and then by generating cover information combining text and pictures, the interest of the recommended text information can be enhanced.

[0075] Based on the text category recognition model, when the server obtains the text category label corresponding to the information to be recommended, the following method can be adopted:

[0076] 3021. The server extracts text category features from the text information to be recommended.

[0077] Among them, the text category feature is a feature used to identify the category of text information. The server extracting text category features from the text information to be recommended includes the following steps:

[0078] The first step, the server preprocesses the text information to be recommended to remove meaningless information in the recommended text information.

[0079] Among them, the meaningless information includes punctuation marks, prepositions, etc.

[0080] The second step, the server performs word segmentation on the preprocessed text information to be recommended and identifies out-of-vocabulary words in it, obtaining the text information to be recommended after word segmentation.

[0081] Among them, the out-of-vocabulary word refers to a word that is not included in the word segmentation table but needs to be segmented, including various proper nouns (such as personal names, place names, company names, etc.), abbreviations, newly added words, etc.

[0082] The third step, the server extracts text category features from the text information to be recommended after word segmentation.

[0083] The server extracts text features from the text information to be recommended after word segmentation, and then selects features related to text classification from the extracted text features to achieve the purpose of reducing the dimension of text category features and reducing the amount of calculation, and then selects a suitable way to represent the selected features to obtain text category features. The suitable way can be a vector way, a matrix way, etc.

[0084] 3022. The server calls the text category recognition model to process the text category features to obtain text category labels.

[0085] Among them, the text category labels include first-level text category labels and second-level text category labels. The first-level text category labels include animal science popularization, plant science popularization, games, emotions, etc. The second-level text category labels are sub-labels of the first-level text category labels. For example, if the first-level text category label is animal science popularization, the second-level text category labels are tiger, rabbit, lion, owl, bear, etc.; another example, if the first-level text category label is plant science popularization, the second-level text category labels are dandelion, mulberry tree, sensitive plant, etc.

[0086] In the embodiments of the present disclosure, the text category recognition model is a relationship model that reflects the relationship between text category features and text category labels. The server processes the text category features of the text information to be recommended by calling the text category recognition model, and obtains the text category label corresponding to the text information to be recommended.

[0087] 303. The server obtains the object information of the target object. When the object information of the target object meets the preset age condition and the text category label belongs to the preset text category label, step 304 is executed.

[0088] To protect minor users and reduce the risk of minor users accessing bad information on the Internet and being victimized by criminal acts, when the text category label corresponding to the recommended text information is obtained, the server will also obtain the object information of the target object, and then judge whether the object information of the target object meets the preset age condition based on the object information of the target object. The preset age condition can be less than 18 years old, etc. When it is determined based on the object information of the target object that the target object meets the preset age condition, that is, the target object may be a minor user, the server will also screen the recommended text information based on the preset text category label. Among them, the preset text category label is the text category label corresponding to the text information allowed for the target object to view. The preset text category label can be animal science popularization, plant science popularization, etc. When the text category label belongs to the preset text category label, the server will execute step 304; when the text category label does not belong to the preset text category label, the server will no longer recommend the information to be recommended to the target object.

[0089] The following three points need to be explained:

[0090] First, when the server determines that the object information of the target object meets the preset age condition, even if the text category label does not belong to the preset text category label, whether to recommend the to-be-recommended text information to the target object also depends on the target object. If the target object enables the corresponding protection mode in the text information acquisition application, such as the minor user protection mode, then when the text category label does not belong to the preset text category label, the to-be-recommended text information will no longer be recommended to the target object; if the target object does not enable the corresponding protection mode in the text information acquisition application, then when the text category label does not belong to the preset text category label, the to-be-recommended text information can still be recommended to the target object.

[0091] Second, the above text category label includes the first-level text category label and the second-level text category label. When the server determines whether the text category label belongs to the preset text category label, it can match the first-level text category label with the preset text category label, or match the second-level text category label with the preset text category label. When the first-level text category label or the second-level text category label is the same as the preset text category label, or the first-level text category label or the second-level text category label is a sub-label of the preset text category label, it can be determined that the text category label belongs to the preset text category label.

[0092] Third, after obtaining the text category label corresponding to the to-be-recommended information and before generating the cover information corresponding to the to-be-recommended text information, it is determined whether the object information of the target object meets the preset age condition, so that when the text category label of the to-be-recommended text information does not belong to the preset category label, the picture associated with the content of the to-be-recommended text information will no longer be obtained. Of course, the server can also determine whether the object information of the target object meets the preset age condition after generating the cover information corresponding to the to-be-recommended text information and before recommending the cover information to the target object. The embodiments of the present disclosure do not make specific limitations on this.

[0093] 304. The server obtains the picture associated with the content of the to-be-recommended text information according to the text category label.

[0094] Based on the text category label corresponding to the to-be-recommended text information, the server obtains the picture associated with the content of the to-be-recommended text information. Specifically, it includes the following steps:

[0095] 3041. The server obtains at least one associated picture associated with the content of the text information according to the text category label.

[0096] The server obtains a Scrapy framework crawler written by a technician based on the Python programming language, and then, based on the text category label, uses the goose31 library of Python to initiate a query request to the search engine, enabling the search engine to query images based on this query request and return at least one associated image related to the content of the recommended text information to the server. Among them, goose31 is an article extractor written in the Java language, which is used to obtain web pages of news articles or article types, and then extract the main body of the article as well as all metadata and images.

[0097] Since the text category label includes a first-level text category label and a second-level text category label, for different levels of text category labels, the at least one associated image obtained by the server related to the content of the text information is also different.

[0098] In a possible implementation manner, when the text category label is a first-level text category label, the server obtains images containing the content indicated by the first-level text category label to obtain at least one associated image. The objects in the at least one associated image belong to the same first-level text category label, but are not necessarily the same object. For example, if the first-level text category label is animal science popularization, the images obtained based on the first-level text category label may be a picture of a tiger or a picture of a lion.

[0099] In another possible implementation manner, when the text category label is a second-level text category label, the server obtains images containing the content indicated by the second-level text category label to obtain at least one associated image. The objects in the at least one associated image belong to the same second-level text category label, but are not necessarily the same object. For example, if the second-level text category label is tiger, the images obtained based on the second-level text category label may be a picture of a South China tiger or a picture of a Northeast China tiger.

[0100] In order to more accurately reflect the content in the text information to be recommended, the text category label in the embodiments of the present disclosure may further include a third-level text category label, which is a sub-label of the second-level text category label. Based on this third-level text category label, the server obtains images containing the content indicated by the third-level text category label to obtain at least one associated image. The objects in the at least one associated image belong to the same third-level text category label and are the same object. For example, if the second-level text category label is bear and the third-level text category label is polar bear, the images obtained based on the third-level text category label are all pictures of polar bears.

[0101] 3042. The server identifies sensitive images containing sensitive information from at least one associated image.

[0102] Considering that the pictures on the network may contain sensitive information such as advertisements, website addresses, and uncivilized language, in order to prevent such sensitive information from interfering with the target object's viewing of the text information to be recommended, after the server obtains at least one associated picture, it will also identify whether at least one associated picture is a sensitive picture containing sensitive information. For each associated picture among at least one associated picture, when the server identifies whether the associated picture is a sensitive picture, the following several methods can be adopted:

[0103] In the first method, the server can identify whether the associated picture contains preset characters based on OCR technology. When the associated picture contains preset characters, the server identifies the associated picture as a sensitive picture. The preset characters can be advertisement words, website address characters, uncivilized language words, etc. For example, Figure 4 Among the six pictures shown in, the first picture, the second picture, and the third picture contain the word "advertisement", so it is determined that the first picture, the second picture, and the third picture are sensitive pictures.

[0104] In the second method, the server obtains the picture parameters of the associated picture, and then, when the picture parameters match the picture parameters of the sensitive picture, identifies the associated picture as a sensitive picture.

[0105] 3043. The server deletes the sensitive picture.

[0106] When any associated picture is identified as a sensitive picture, the server deletes the sensitive picture to reduce the interference of the sensitive picture to the user and improve the user's reading experience of the text information.

[0107] 3044. The server obtains pictures from the remaining associated pictures according to the attribute information of the remaining associated pictures.

[0108] Among them, the attribute information includes size, dimensions, contrast, clarity, etc. The server sorts the remaining associated pictures according to at least one of the attribute information of the remaining associated pictures, and selects high-quality associated pictures as the pictures associated with the content of the recommended text information.

[0109] 305. The server obtains the text elements corresponding to the text information to be recommended.

[0110] In the embodiments of the present disclosure, a text element generation model is installed and run in the server. The text element generation model is used to generate corresponding text elements for any text information. Based on the text element generation model, the server can obtain the text elements corresponding to the text information to be recommended. Specifically, it includes the following steps:

[0111] In the first step, the server extracts the text element features from the text information to be recommended.

[0112] Among them, the text element feature is the feature of the text element that identifies the text information.

[0113] In the second step, the server calls the text element generation model to process the text element feature and obtain the text element.

[0114] When the server calls the text element generation model to process the text element feature according to the text element and the text information to be recommended and obtains the text element, the following situations are included:

[0115] In the first case, when the text element includes a text title and the text information to be recommended does not include a text title, the server calls the text element generation model to process the text element feature and obtain the text title corresponding to the text information to be recommended.

[0116] In the second case, when the text element includes a text abstract and the text information to be recommended does not include a text abstract, the server calls the text element generation model to process the text element feature and obtain the text abstract corresponding to the text information to be recommended.

[0117] In the third case, when the text element includes a text title and a text abstract, if the text information to be recommended includes a text title but does not include a text abstract, the server calls the text element generation model to process the text element feature and obtain the text abstract corresponding to the text information to be recommended; if the text information to be recommended does not include a text title but includes a text abstract, the server calls the text element generation model to process the text element feature and obtain the text title corresponding to the text information to be recommended; if the text information to be recommended does not include a text title nor a text abstract, the server calls the text element generation model to process the text element feature and obtain the text title and text abstract corresponding to the text information to be recommended.

[0118] It should be noted that the above process of obtaining the picture associated with the content of the text information to be recommended and the process of obtaining the text title and text abstract corresponding to the text information to be recommended can be executed successively or synchronously. The embodiments of the present disclosure do not limit the execution order of the above two processes.

[0119] 306. Based on the picture and the text element, the server generates the cover information corresponding to the text information to be recommended.

[0120] In the embodiments of the present disclosure, the text element includes at least one of a text title or a text abstract. For different contents included in the text element, when the server generates the cover information corresponding to the text information to be recommended based on the picture associated with the content of the text information to be recommended and the text element, the following situations are included:

[0121] In the first case, the text element includes a text title.

[0122] In this case, the server determines a specified position on the picture for adding the text title according to the text features of the text title and the picture features of the picture associated with the content of the text to be recommended, and then adds the text title to the specified position on the picture to obtain the cover information.

[0123] The second case: the text element includes a text abstract.

[0124] In this case, the server determines the display areas of the picture and the text abstract on the cover information to be generated, and then adds the picture and the text abstract to the corresponding display areas respectively to obtain the cover information.

[0125] The third case: the text elements include a text title and a text abstract.

[0126] In this case, the server determines the display areas of the picture and the text abstract on the cover information to be generated, and then adds the picture and the text abstract to the corresponding display areas respectively. Then, the server determines a specified position on the picture for adding the text title according to the text features of the text title and the picture features of the picture, and then adds the text title to the specified position on the picture to obtain the cover information.

[0127] Figure 5 The generation process of the above cover information is shown. See Figure 5 , the server obtains the text information to be recommended, extracts the text category features of the text information to be recommended, and then calls the text category recognition model to process the text category features to obtain the text category label of the text information to be recommended. The server extracts the text element features of the text information to be recommended, and then calls the text element generation model to process the text element features to obtain the text elements of the text information to be recommended. Then, based on the text category label, the server obtains a picture associated with the content of the text information to be recommended, and then combines the picture and the text elements to generate the cover information corresponding to the text information to be recommended.

[0128] 307. The server recommends the cover information to the target object.

[0129] When the server recommends the cover information to the target object, it can determine the recommendation order of the cover information according to at least one of the attention information or the release time of the information to be recommended, and then recommend the cover information to the target object according to the recommendation order. Among them, the attention information includes the number of likes, the number of shares, the number of forwards, the click-through rate, etc.

[0130] In a possible implementation, the server may determine the recommended order of the cover information according to the concerned information. For example, the server may determine the recommended order of the cover information in descending order of the number of likes; the server may also determine the recommended order of the cover information in descending order of the number of shares; the server may also determine the recommended order of the cover information in descending order of the number of forwards; the server may also determine the recommended order of the cover information in descending order of the click-through rate. Of course, the server may also determine the recommended order of the cover information based on at least two of the number of likes, the number of shares, the number of forwards, and the click-through rate, which will not be elaborated here.

[0131] In another possible implementation, the server may determine the recommended order of the cover information according to the release time. For example, the server may determine the recommended order of the cover information in ascending order of the release time.

[0132] In another possible implementation, the server may determine the recommended order of the cover information according to the concerned information and the release time. For example, the server may set weight values for each piece of information included in the concerned information and the release time, calculate the recommended scores of each piece of text information to be recommended based on the set weight values, and then determine the recommended order of the cover information corresponding to the text information to be recommended in descending order of the recommended scores.

[0133] Figure 6 The display effect of the information recommended by the related art is shown. The cover information recommended by the embodiments of the present disclosure is in the form of a combination of pictures and texts, which is more readable and interesting.

[0134] The method provided by the embodiments of the present disclosure obtains a picture associated with the content of the text information to be recommended based on the text category label of the information to be recommended. The picture associated with the content of the text information to be recommended can visually and vividly show the external manifestation of the object described by the text information to be recommended to the user. Then, the text elements corresponding to the recommended text information are obtained. The text elements include at least one of the text title or the text summary of the text information to be recommended, which can briefly show the core content of the information to be recommended to the user. Furthermore, a cover information in the form of a combination of pictures and texts is generated by combining the picture associated with the content of the text information to be recommended and the text elements. Compared with the information in a single text form, the cover information has a richer information form and stronger interest, which greatly increases the probability of the user viewing the text information.

[0135] In addition, for underage users, the method provided by the embodiments of the present disclosure screens the obtained information to be recommended based on the preset text category tags and the text category tags of the information to be recommended, and screens out the text content suitable for underage users to view, reducing the risk of underage users being exposed to bad information on the Internet and being violated by criminal acts, and creating a good Internet environment for underage users.

[0136] The embodiments of the present disclosure provide a method for training a text category recognition model. Taking the server executing the embodiments of the present disclosure as an example, the server can be Figure 1 the server shown in Figure 1 At this time, the server shown in Figure 1 will execute two processes: model training and text information recommendation. The server can also be Figure 1 a server other than the server shown in Figure 7 At this time, the server shown in

[0137] 801. The server obtains a plurality of sample text information.

[0138] Among them, the sample text information includes short text information, such as sentences, text titles, product reviews, etc., and also includes long text information, such as articles, etc. Each sample text information is labeled with a sample text category tag, which can be set by technicians. For example, technicians can set it according to the category of text information that objects meeting the preset age conditions can view. The sample text category tag includes a first-level sample text category tag and a second-level sample text category tag. The first-level sample text category tag includes animal science popularization, plant science popularization, games, emotions, etc. The second-level sample text category tag is a sub-tag of the first-level sample text category tag. For example, the second-level sample text category tags under animal science popularization include tigers, lions, bears, owls, etc.; for another example, the second-level sample text category tags under plant science popularization include dandelions, mulberry trees, sensitive plants, etc.

[0139] When the quantity of the obtained sample text information accumulates to the sample magnitude required for model training,

[0140] the server will start training the text category recognition model by executing step 802.

[0141] 802. The server extracts the sample text category features of a plurality of sample text information to obtain a plurality of sample text category features.

[0142] When the server extracts the sample text category features of a plurality of sample text information to obtain a plurality of sample text category features, the following steps are included:

[0143] Step 1: The server preprocesses multiple sample text messages to remove meaningless information in the multiple sample text messages.

[0144] Step 2: The server performs word segmentation on the multiple preprocessed sample text messages to obtain multiple sample text messages after word segmentation.

[0145] Step 3: The server extracts multiple sample text category features from the multiple sample text messages after word segmentation.

[0146] 803. Based on the multiple sample text category features, the server trains the initial text category recognition model to obtain a text category recognition model.

[0147] The server sets initial model parameters for the initial text category recognition model, inputs the multiple sample text category features into the initial text category recognition model, and outputs the recognition results of the multiple sample text messages. The server inputs the recognition results of the multiple sample text messages and their labeled sample text annotation labels into a pre-constructed target loss function to obtain the function value of the target loss function. When the function value of the target loss function does not meet the threshold condition, the server continuously adjusts the model parameters of the text category recognition model until the function value of the target loss function meets the threshold condition. The server obtains the model parameters when the threshold condition is met, and then takes the text category recognition model corresponding to the model parameters as the trained text category recognition model. Among them, the threshold condition can be set according to the processing accuracy of the server.

[0148] Further, after the text category recognition model is trained, the server also tests the accuracy and coverage rate of the text category recognition model by itself. When the accuracy and coverage rate of the text category recognition model reach the set indicators, the server can launch the text category recognition model by turning on the switch; when the accuracy and coverage rate of the text category recognition model do not reach the set indicators, the server needs to continue training the accuracy and coverage rate until the accuracy and coverage rate of the text category recognition model reach the set indicators.

[0149] In the embodiments of the present disclosure, by pre-training the text category recognition model, in the scenario of text information recommendation, the text category label of the text information to be recommended can be quickly recognized based on the trained text category recognition model. On the one hand, the text information to be recommended can be screened based on the text category label, and on the other hand, the cover information corresponding to the text information to be recommended can be generated based on the text category label, which improves the generation speed of the cover information and enhances the fit between the cover information and the text information to be recommended.

[0150]

[0151] ​For the recommendation process of the text information provided by the embodiments of the present disclosure, the following will take Figure 8 as an example for detailed description.

[0152] See Figure 8 , the server pre-trains a text category recognition model based on the labeled sample text information in advance. When receiving an information acquisition request sent by a user and obtaining the text information to be recommended, the server extracts the text category features of the text information to be recommended, and then calls the text category recognition model to process the text category features to obtain the text category label corresponding to the text information to be recommended. If the text category label does not belong to the non-recommended label (that is, does not belong to the preset text category label), the server will no longer recommend the text information to be recommended; if the text category label belongs to the recommended label (that is, belongs to the preset text category label), the server will, according to the text category label, obtain the associated pictures related to the content of the text information to be recommended, filter the obtained associated pictures, delete the sensitive pictures including sensitive information such as advertisements, and then sort the remaining associated pictures according to the attribute information of the remaining associated pictures, and obtain the pictures with better quality from them. The server also detects whether the text information to be recommended contains a text title. If the text information to be recommended contains a text title, it calls a text element generation model to generate a text summary for the text information to be recommended; if the text information to be recommended contains a text title, it calls a text element generation model to generate a text summary for the text information to be recommended. Then, the server generates an intelligent information cover based on the obtained pictures, text summary and text title, and then recommends the information cover to the user.

[0153] See Figure 9 , the embodiments of the present disclosure provide a text information recommendation device, which can be applied to various scenarios, including but not limited to various scenarios such as cloud technology, artificial intelligence, intelligent transportation, and assisted driving. The device includes:

[0154] A first acquisition module 1101, configured to acquire the text information to be recommended in response to an information acquisition request of a target object;

[0155] The first acquisition module 1101 is further configured to acquire the text category label corresponding to the text information to be recommended;

[0156] The first acquisition module 1101 is further configured to acquire the pictures associated with the content of the text information to be recommended according to the text category label;

[0157] The first acquisition module 1101 is further configured to acquire the text elements corresponding to the text information to be recommended, where the text elements include at least one of a text title or a text summary;

[0158] A generation module 1102, configured to generate cover information corresponding to the text information to be recommended based on pictures and text elements;

[0159] A recommendation module 1103, configured to recommend the cover information to a target object.

[0160] In another embodiment of the present disclosure, a first acquisition module 301 is configured to extract text category features from the text information to be recommended; call a text category recognition model to process the text category features to obtain a text category label, where the text category recognition model is used to recognize the text category label corresponding to any text information.

[0161] In another embodiment of the present disclosure, the apparatus for training a text category recognition model includes:

[0162] A second acquisition module, configured to acquire a plurality of sample text information, and each of the plurality of sample text information is labeled with a sample text category label;

[0163] An extraction module, configured to extract sample text category features of the plurality of sample text information to obtain a plurality of sample text category features;

[0164] A training module, configured to train an initial text category recognition model based on the plurality of sample text category features to obtain a text category recognition model.

[0165] In another embodiment of the present disclosure, the apparatus for training a text category recognition model further includes:

[0166] A preprocessing module, configured to preprocess the plurality of sample text information to remove meaningless information in the plurality of sample text information;

[0167] A word segmentation module, configured to perform word segmentation processing on the plurality of preprocessed sample text information to obtain a plurality of segmented sample text information;

[0168] An extraction module, configured to extract a plurality of sample text category features from the plurality of segmented sample text information.

[0169] In another embodiment of the present disclosure, a first acquisition module 1101 is configured to acquire at least one associated picture associated with the content of the text information according to the text category label; identify a sensitive picture containing sensitive information from the at least one associated picture; delete the sensitive picture; and acquire the picture from the remaining associated pictures according to the attribute information of the remaining associated pictures.

[0170] In another embodiment of the present disclosure, the text category label includes a first-level text category label and a second-level text category label. The first acquisition module 1101 is configured to acquire pictures containing the content indicated by the first-level text category label to obtain at least one associated picture; or,

[0171] The first acquisition module 1101 is configured to acquire pictures containing the content indicated by the second-level text category label, and obtain at least one associated picture.

[0172] In another embodiment of the present disclosure, for any one of the at least one associated picture, when the associated picture contains a preset character, the first acquisition module 1101 is configured to identify the associated picture as a sensitive picture; or,

[0173] The first acquisition module 1101 is configured to acquire the picture parameters of the associated picture, and when the picture parameters match the picture parameters of the sensitive picture, identify the associated picture as a sensitive picture.

[0174] In another embodiment of the present disclosure, the first acquisition module 1101 is further configured to acquire the object information of the target object;

[0175] The first acquisition module 1101 is further configured to, when the object information of the target object meets the preset age condition and the text category label belongs to the preset text category label, acquire a picture associated with the content of the text information to be recommended according to the text category label, where the preset text category label is the text category label corresponding to the text information allowed for the target object to view.

[0176] In another embodiment of the present disclosure, the first acquisition module 1101 is configured to extract text element features from the text information to be recommended; call a text element generation model to process the text element features to obtain text elements, where the text element generation model is used to generate corresponding text elements for any text information.

[0177] In another embodiment of the present disclosure, when the text element includes a text title, the generation module 1102 is configured to add the text title to a specified position of the picture to obtain cover information, where the specified position is determined according to the text feature of the text title and the picture feature of the picture; or,

[0178] When the text element includes a text abstract, the generation module 1102 is configured to determine the display areas of the picture and the text abstract on the cover information to be generated, and add the picture and the text abstract to the corresponding display areas respectively to obtain cover information; or,

[0179] When the text element includes a text title and a text abstract, the generation module 1102 is configured to determine the display areas of the picture and the text abstract on the cover information to be generated, add the picture and the text abstract to the corresponding display areas respectively, and add the text title to a specified position of the picture to obtain cover information.

[0180] In another embodiment of the present disclosure, the recommendation module 1103 is configured to determine the recommendation order of the cover information according to at least one of the attention information or the release time of the information to be recommended; and recommend the cover information to the target object according to the recommendation order.

[0181] In another embodiment of the present disclosure, when the object information of the target object meets the preset age condition, the first acquisition module 1101 is further configured to acquire the label type to which the text category label belongs.

[0182] The recommendation module 1103 is further configured to recommend the cover information to the target object when the text category label belongs to a preset text category label, where the preset text category label is the text category label corresponding to the text information that the target object is allowed to view.

[0183] In summary, the device provided by the embodiments of the present disclosure obtains a picture associated with the content of the to-be-recommended text information based on the text category label of the to-be-recommended information. The picture associated with the content of the to-be-recommended text information can intuitively and vividly show the external manifestation of the object described by the to-be-recommended text information to the user. Then, the text elements corresponding to the recommended text information are obtained. The text elements include at least one of the text title or the text summary of the to-be-recommended text information, which can briefly show the core content of the to-be-recommended information to the user. Furthermore, by combining the picture associated with the content of the to-be-recommended text information and the text elements, a cover information in the form of a combination of text and picture is generated. Compared with the information in a single text form, the cover information has a richer information form and stronger interest, greatly increasing the probability that the user views the text information.

[0184] In addition, for minor users, the method provided by the embodiments of the present disclosure screens the obtained to-be-recommended information based on the preset text category label and the text category label of the to-be-recommended information, and screens out the text content suitable for minor users to view, reducing the risk that minor users are exposed to bad information on the Internet and are victimized by criminal acts, and creating a good network environment for minor users.

[0185] Figure 10 is a server for recommending text information shown according to an exemplary embodiment. Refer to Figure 10 , the server 1200 includes a processing component 1222, which further includes one or more processors, and memory resources represented by a memory 1232 for storing instructions executable by the processing component 1222, such as application programs. The application programs stored in the memory 1232 may include one or more modules each corresponding to a set of instructions. In addition, the processing component 1222 is configured to execute instructions to perform the functions performed by the server in the above-mentioned text information recommendation method.

[0186] The server 1200 may also include a power supply component 1226 configured to perform power management of the server 1200, a wired or wireless network interface 1250 configured to connect the server 1200 to a network, and an input / output (I / O) interface 1258. The server 1200 may operate based on an operating system stored in the memory 1232, such as Windows Server TM , Mac OS X TM , Unix TM , Linux TM , FreeBSD TM or the like.

[0187] Embodiments of the present disclosure provide a computer-readable storage medium storing at least one program code, which is loaded and executed by a processor to implement a method for recommending text information. The computer-readable storage medium may be non-transitory. For example, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, magnetic tape, a floppy disk, and an optical data storage device, etc.

[0188] Embodiments of the present disclosure provide a computer program product including computer program code stored in a computer-readable storage medium. The processor of the server reads the computer program code from the computer-readable storage medium, and the processor executes the computer program code to cause the server to execute the method for recommending text information.

[0189] Those of ordinary skill in the art can understand that all or part of the steps to implement the above embodiments can be completed by hardware or by a program instructing relevant hardware. The program can be stored in a computer-readable storage medium, and the above-mentioned storage medium can be a read-only memory, a magnetic disk, or an optical disc, etc.

[0190] The above are only optional embodiments of the present disclosure and are not intended to limit the present disclosure. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present disclosure shall be included within the protection scope of the present disclosure.

Claims

1. A method for recommending text information, characterized in that The method includes: In response to an information acquisition request of a target object, based on the object information of the target object included in the information acquisition request, acquire to-be-recommended text information, where the to-be-recommended text information is text-type information; Acquire a text category label corresponding to the to-be-recommended text information, where the text category label includes multi-level text category labels, and a high-level text category label is a sub-label of a low-level text category label; When the object information of the target object meets a preset age condition and the text category label belongs to a preset text category label, according to the highest-level text category label in the text category label, acquire a picture associated with the content of the to-be-recommended text information from the network, where the preset text category label is a text category label corresponding to text information allowed to be viewed by the target object; acquire text elements corresponding to the to-be-recommended text information, where the text elements include at least one of a text title or a text abstract; when the text elements include the text title, add the text title to a specified position of the picture to generate cover information corresponding to the to-be-recommended text information; recommend the cover information to the target object; When the object information of the target object meets the preset age condition and the text category label does not belong to the preset text category label, determine whether the target object has enabled a protection mode; if the protection mode has been enabled, do not recommend the to-be-recommended text information to the target object, and if the protection mode has not been enabled, continue to perform the step of recommending the to-be-recommended text information to the target object.

2. The method according to claim 1, characterized in that, The acquiring the text category label corresponding to the to-be-recommended text information includes: Extract text category features from the to-be-recommended text information; Call a text category recognition model to process the text category features to obtain the text category label, where the text category recognition model is used to recognize a text category label corresponding to any text information.

3. The method according to claim 2, wherein The training process of the text category recognition model is: Acquire a plurality of sample text information, where the plurality of sample text information are all labeled with sample text category labels; Extract sample text category features of the plurality of sample text information to obtain a plurality of sample text category features; Based on the plurality of sample text category features, train an initial text category recognition model to obtain the text category recognition model.

4. The method according to claim 3, characterized in that, Before the extracting the sample text category features of the plurality of sample text information to obtain a plurality of sample text category features, it further includes: Perform preprocessing on the plurality of sample text information to remove meaningless information in the plurality of sample text information; Perform word segmentation processing on the plurality of preprocessed sample text information to obtain a plurality of word-segmented sample text information; The extracting the sample text category features of the plurality of sample text information to obtain a plurality of sample text category features includes: Extract a plurality of sample text category features from the plurality of word-segmented sample text information.

5. The method according to claim 1, characterized in that, The acquiring the picture associated with the content of the to-be-recommended text information from the network according to the highest-level text category label in the text category label includes: Obtain at least one associated picture associated with the content of the to-be-recommended text information according to the text category label; Identify a sensitive picture containing sensitive information from the at least one associated picture; Delete the sensitive picture; Obtain the picture from the remaining associated pictures according to the attribute information of the remaining associated pictures.

6. The method according to claim 5, wherein The text category label includes a first-level text category label and a second-level text category label. The step of obtaining at least one associated picture associated with the content of the to-be-recommended text information according to the text category label includes: Obtain pictures containing the content indicated by the second-level text category label to obtain the at least one associated picture.

7. The method according to claim 5, characterized in that, The step of identifying a sensitive picture containing sensitive information from the at least one associated picture includes: For any one of the at least one associated pictures, when the associated picture contains a preset character, identify the associated picture as a sensitive picture; or Obtain the picture parameters of the associated picture, and when the picture parameters match the picture parameters of a sensitive picture, identify the associated picture as a sensitive picture.

8. The method according to claim 1, wherein Before obtaining a picture associated with the content of the to-be-recommended text information from the network according to the highest-level text category label in the text category label when the object information of the target object meets the preset age condition and the text category label belongs to a preset text category label, it further includes: Obtain the object information of the target object.

9. The method according to claim 1, characterized in that, The step of obtaining the text elements corresponding to the to-be-recommended text information includes: Extract text element features from the to-be-recommended text information; Call a text element generation model to process the text element features to obtain the text elements. The text element generation model is used to generate corresponding text elements for any text information.

10. The method according to claim 1, characterized in that The method further includes: When the text elements include a text summary, determine the display areas of the picture and the text summary on the to-be-generated cover information, and add the picture and the text summary to the corresponding display areas respectively to obtain the cover information; or When the text elements include a text title and a text summary, determine the display areas of the picture and the text summary on the to-be-generated cover information, add the picture and the text summary to the corresponding display areas respectively, and add the text title to the specified position of the picture to obtain the cover information.

11. The method according to any one of claims 1 to 10, characterized in that, The step of recommending the cover information to the target object includes: Determine the recommendation order of the cover information according to at least one of the attention information or the release time of the to-be-recommended text information; Recommend the cover information to the target object according to the recommendation order.

12. A recommendation device for text information, characterized in that, The device includes: A first acquisition module, configured to, in response to an information acquisition request of a target object, acquire to-be-recommended text information based on the object information of the target object included in the information acquisition request, where the to-be-recommended text information is text-type information; The first acquisition module is further configured to acquire a text category label corresponding to the text information to be recommended, where the text category label includes multi-level text category labels, and the high-level text category label is a sub-label of the low-level text category label; The first acquisition module is further configured to, when the object information of the target object meets a preset age condition and the text category label belongs to a preset text category label, obtain, according to the highest-level text category label in the text category label, a picture associated with the content of the text information to be recommended from the network, where the preset text category label is a text category label corresponding to text information allowed to be viewed by the target object; The first acquisition module is further configured to acquire text elements corresponding to the text information to be recommended, where the text elements include at least one of a text title or a text abstract; The generation module is configured to, when the text elements include the text title, add the text title to a specified position of the picture to generate cover information corresponding to the text information to be recommended; The recommendation module is configured to recommend the cover information to the target object; A module for performing the following steps: when the object information of the target object meets the preset age condition and the text category label does not belong to the preset text category label, determine whether the target object has enabled a protection mode; if the protection mode has been enabled, do not recommend the text information to be recommended to the target object, and if the protection mode has not been enabled, continue to perform the step of recommending the text information to be recommended to the target object.

13. The device according to claim 12, characterized in that, The first acquisition module is configured to: Extract text category features from the text information to be recommended; Call a text category recognition model to process the text category features to obtain the text category label, where the text category recognition model is used to recognize the text category label corresponding to any text information.

14. The device according to claim 13, wherein An apparatus for training a text category recognition model includes: A second acquisition module for acquiring a plurality of sample text information, where the plurality of sample text information are all labeled with sample text category labels; An extraction module for extracting sample text category features of the plurality of sample text information to obtain a plurality of sample text category features; A training module for training an initial text category recognition model based on the plurality of sample text category features to obtain the text category recognition model.

15. The device according to claim 14, characterized in that, The apparatus for training a text category recognition model further includes: A preprocessing module for preprocessing the plurality of sample text information to remove meaningless information in the plurality of sample text information; A word segmentation module for performing word segmentation processing on the plurality of preprocessed sample text information to obtain a plurality of word-segmented sample text information; The extraction module for extracting a plurality of sample text category features from the plurality of word-segmented sample text information.

16. The device according to claim 12, characterized in that, The first acquisition module is configured to: Obtain at least one associated picture associated with the content of the text information to be recommended according to the text category label; Identify sensitive pictures containing sensitive information from the at least one associated picture; Delete the sensitive pictures; Obtain the picture from the remaining associated pictures according to the attribute information of the remaining associated pictures.

17. The device according to claim 16, characterized in that, The text category label includes a first-level text category label and a second-level text category label. The first obtaining module is configured to: Obtain pictures containing the content indicated by the second-level text category label to obtain the at least one associated picture.

18. The device according to claim 16, wherein The first obtaining module is configured to: For any one of the at least one associated picture, when the associated picture contains a preset character, identify the associated picture as a sensitive picture; or, Obtain the picture parameters of the associated picture, and when the picture parameters match the picture parameters of the sensitive picture, identify the associated picture as a sensitive picture.

19. The device according to claim 12, wherein The first obtaining module is further configured to obtain the object information of the target object.

20. The device according to claim 12, characterized in that, The first obtaining module is configured to: Extract text element features from the text information to be recommended; Call a text element generation model to process the text element features to obtain the text elements. The text element generation model is used to generate corresponding text elements for any text information.

21. The device according to claim 12, characterized in that, The generation module is further configured to: When the text element includes a text summary, determine the display areas of the picture and the text summary on the cover information to be generated, and add the picture and the text summary to the corresponding display areas respectively to obtain the cover information; Or, When the text element includes a text title and a text summary, determine the display areas of the picture and the text summary on the cover information to be generated, add the picture and the text summary to the corresponding display areas respectively, and add the text title to the specified position of the picture to obtain the cover information.

22. The device according to any one of claims 12 to 21, characterized in that, The recommendation module is configured to: Determine the recommendation order of the cover information according to at least one of the attention information or the release time of the text information to be recommended; Recommend the cover information to the target object according to the recommendation order.

23. A server, characterized in that, The server includes a processor and a memory. At least one program code is stored in the memory, and the at least one program code is loaded and executed by the processor to implement the text information recommendation method according to any one of claims 1 to 11.

24. A computer-readable storage medium, characterized in that, At least one program code is stored in the storage medium, and the at least one program code is loaded and executed by a processor to implement the text information recommendation method according to any one of claims 1 to 11.

25. A computer program product, characterized in that, The computer program product includes computer program code. The computer program code is stored in a computer-readable storage medium. The processor of the server reads the computer program code from the computer-readable storage medium, and the processor executes the computer program code to enable the server to execute the text information recommendation method according to any one of claims 1 to 11.

Images