Information recommendation method, electronic equipment, computer readable storage medium and product

Abstract

The invention relates to an information recommendation method, electronic equipment, a computer readable storage medium and a product, and relates to the technical field of computers. The information recommendation method comprises the steps that one or more topics and content corresponding to each topic are generated according to multiple answers generated by an intelligent agent in the conversation process with one or more first users, and the multiple answers are authorized by the one or more first users and do not involve personal information of the one or more first users; determining a target topic to be recommended from the one or more topics; generating to-be-recommended target content based on the content corresponding to the target topic; and displaying the target content. Therefore, the utilization rate of computing resources configured for the intelligent agent is improved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technologies, and particularly to an information recommendation method, an electronic device, a computer-readable storage medium, and a product. Background Art

[0002] With the development of artificial intelligence technologies, users can obtain information by conversing with an agent. For example, a user can ask the agent questions, or instruct the agent to create text or images, etc. The agent can reply with a response message according to the message sent by the user to provide the generated content to the user. Summary of the Invention

[0003] According to some embodiments of the present disclosure, an information recommendation method is provided, including: generating one or more topics and content corresponding to each topic according to multiple answers generated by an agent during a conversation with one or more first users, the multiple answers being authorized by the one or more first users and not involving personal information of the one or more first users; determining a target topic to be recommended from the one or more topics; generating target content to be recommended based on the content corresponding to the target topic; and presenting the target content.

[0004] According to some embodiments of the present disclosure, an electronic device is provided, including: a memory; and a processor coupled to the memory, the processor being configured to execute the method according to any one of the embodiments of the present disclosure based on instructions stored in the memory.

[0005] According to some embodiments of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored, and the program, when executed by a processor, executes the method according to any one of the embodiments of the present disclosure.

[0006] According to some embodiments of the present disclosure, a computer program product is provided, which, when running on a computer, causes the computer to implement the method according to any one of the embodiments of the present disclosure.

[0007] Other features, aspects, and advantages of the present disclosure will become clear through the following detailed description of the exemplary embodiments of the present disclosure with reference to the accompanying drawings. Brief Description of the Drawings

[0008] Embodiments of the present disclosure will be described below with reference to the accompanying drawings. It should be understood that the drawings in the following description only relate to some embodiments of the present disclosure and do not constitute a limitation to the present disclosure. In the drawings:

[0009] Figure 1 A flowchart showing an information recommendation method according to some embodiments of the present disclosure is shown.

[0010] Figure 2 The flowchart shows a topic generation method according to some embodiments of the present disclosure.

[0011] Figure 3 The flowchart shows a method for determining a target topic according to some embodiments of the present disclosure.

[0012] Figure 4 The flowchart shows a method for generating target content according to some embodiments of the present disclosure.

[0013] Figure 5 The schematic diagram shows a dialogue interface according to some embodiments of the present disclosure.

[0014] Figure 6 The schematic diagram shows a recommendation interface according to some embodiments of the present disclosure.

[0015] Figure 7 The schematic diagram shows the structure of an information recommendation device according to some embodiments of the present disclosure.

[0016] Figure 8 The block diagram shows an electronic device according to some embodiments of the present disclosure.

[0017] Figure 9 The block diagram shows an electronic device according to some other embodiments of the present disclosure. Detailed implementation manners

[0018] Next, the technical solutions in the embodiments of the present disclosure will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present disclosure. It should be understood that the present disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein.

[0019] It should be understood that the steps described in the method embodiments of the present disclosure can be executed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this regard. Unless otherwise specifically stated, the relative arrangements of the components and steps set forth in these embodiments should be construed as merely exemplary and do not limit the scope of the present disclosure.

[0020] The term "including" and its variants used in the present disclosure mean an open term that includes at least the subsequent elements / features, but does not exclude other elements / features, that is, "including but not limited to". The term "based on" means "at least partially based on".

[0021] It should be noted that concepts such as "first" and "second" mentioned in this disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependent relationships. Unless otherwise specified, concepts such as "first" and "second" are not intended to imply that the objects so described must be in a given order in terms of time, space, ranking or any other way.

[0022] It should be noted that the modification of "one" and "multiple" mentioned in this disclosure is illustrative rather than restrictive. Those skilled in the art should understand that unless clearly specified otherwise in the context, it should be understood as "one or more".

[0023] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are only for illustrative purposes and are not used to limit the scope of these messages or information.

[0024] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties. And the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions, and corresponding operation entrances are provided for users to choose to authorize or refuse.

[0025] The embodiments of this disclosure will be described in detail below with reference to the accompanying drawings, but this disclosure is not limited to these specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. In addition, in one or more embodiments, specific features, structures or characteristics can be combined in any suitable manner that will be clear to those of ordinary skill in the art from this disclosure.

[0026] An agent can be an intelligent object relying on a machine learning model (such as a neural network model), and the machine learning model can be a large language model (LLM) or a foundation model. The agent can generate content for replying to a message based on the message sent by the user. The machine learning model can be a generative model, which is used to output target content based on the input information. The input information of the generative model includes the processing basis during the generation process of the generative model, such as which information to refer to for the generation process, the requirements for the output target content, and so on. The generative model includes, for example, a model for generating based on text or a model for generating based on images. The output of the generative model can include text, images, or a combination of both. Of course, the input or output of the generative model can also be data of other modalities, such as audio, video, or a combination of multiple types of data. The generative model can be a single-modal model, such as a model for generating text from text (abbreviated as "text-to-text model") or a model for generating images from images (abbreviated as "image-to-image model"); or, the generative model can also be a cross-modal model, that is, a model where the input and output belong to different modalities, such as a model for generating images from text (abbreviated as "text-to-image model"); or, the input of the generative model can include multiple modalities, and the output can also include multiple modalities.

[0027] During the interaction process between the agent and the user, a large amount of reply content will be generated. Currently, during the interaction between the agent and the user, the agent is still in a relatively passive state. The agent is more inclined to be a "tool" for the user to respond to various needs of the user. Since the agent relies on a large amount of computing resources to run, if it is only used to provide replies to the user, on the one hand, the resources of the agent cannot be more fully utilized; on the other hand, users who do not understand the capabilities of the agent may have less interest in interacting with the agent, resulting in a low usage rate of the agent and also wasting resources.

[0028] To improve the resource utilization rate of the agent, the agent can be made to actively push content to the user. Generally, when the agent responds to an instruction sent by the user, it can search for content related to the user's instruction in the agent's knowledge base or on the network, and use a machine learning model to process the obtained content, such as screening, splicing, and organizing language, to obtain high-quality content that can respond to the user. If this part of the content is only used as a response to the current user who issued the instruction, this part of the data cannot be fully utilized. Therefore, the present disclosure provides an information recommendation method that uses the answers generated by the agent during past interactions to generate content and recommends the generated content to the user. In this way, both the generated content of the agent can be fully utilized, and an active interaction method between the agent and the user can be provided.

[0029] Figure 1 FIG. shows a flowchart of an information recommendation method according to some embodiments of the present disclosure. As Figure 1 shown, the information recommendation method of this embodiment includes steps S11 to S17.

[0030] In step S11, based on a plurality of answers generated by the agent during the process of conversing with one or more first users, one or more topics and content corresponding to each topic are generated. These plurality of answers are authorized by one or more first users and do not involve the personal information of one or more first users.

[0031] The first user is a user who has had a conversation with the agent. In the conversation, the user can send questions or instructions, and the agent can send answers in response to the content sent by the user. The conversation between the first user and the agent can be about any content. Therefore, the content generated by the agent may involve one or more topics.

[0032] After obtaining these answers, they can be classified by topic. Thus, the answers generated by the agent during the conversation can be sorted out to form a knowledge base or information base under each topic. The process of sorting out these answers, that is, step S11, can be executed regularly or in response to the total amount of information or data volume of the answers reaching a threshold. Thus, new content generated by the agent can continuously be produced under each topic.

[0033] For example, in a historical conversation, User 1 instructs the agent to generate a travelogue of City A, User 2 instructs the agent to introduce the cuisine of City A, and User 3 instructs the agent to generate a badminton training plan. Thus, two topics, namely "Travel in City A" and "Badminton Training", can be generated. Under the topic of "Travel in City A", it includes the travelogue of City A generated by the agent in response to User 1's instruction and the introduction of the cuisine of City A generated in response to User 2's instruction. Under the topic of "Badminton Training", it includes the badminton training plan generated by the agent in response to User 3.

[0034] In step S13, determine the target topic to be recommended from one or more topics.

[0035] After sorting out one or more topics and the answer content under each topic based on the historical answers of the agent, recommendations can be made based on at least one of these topics. For example, the target topic matching the characteristics can be determined based on the characteristics of the user receiving the recommendation; or, the answer content of each topic can be processed to determine the content quality or topic quality, and the topic with higher quality can be used as the target topic and widely recommended to the user.

[0036] In step S15, generate the target content to be recommended based on the content corresponding to the target topic.

[0037] After determining the target topic, the content under the target topic can be directly used as the target content to be recommended, or the content under the target topic can be further processed to generate the target content. This processing process is, for example, using a machine learning model to process the content under the target topic and the processing instruction. The processing instruction is, for example, specifying the sub-topic under this topic and generating new content based on the sub-topic and the input content. The processing instruction can also be expansion, abbreviation, rewriting, etc.

[0038] The target content to be recommended can be of various types, such as text, image, graphic content (i.e., an article with illustrations), video, etc.

[0039] In step S17, display the target content.

[0040] If steps S11 to S15 are executed on the server, the target content can be sent by the server to the terminal of the second user for display. If steps S11 to S15 are executed locally on the terminal, it is directly displayed on the terminal.

[0041] When presenting the target content, it can be achieved through an agent. That is, the agent can actively push the target content to the second user. The second user can be the same as or different from the aforementioned first user. When presenting the target content, it can be presented through the dialogue interface between the agent and the second user, or through the content recommendation page in the application. The present disclosure does not limit this.

[0042] The above embodiments can classify the answers generated by the agent during the historical dialogue process according to topics, generate new content based on the answers under the target topic, and then recommend the newly generated content to the user. Thus, it is possible to make full use of the systematic and high-quality content once generated by the agent, improve the utilization rate of data, and the generated target content can also have higher quality. Moreover, this active recommendation method of the agent also improves the interactivity between the user and the agent, thereby improving the utilization rate of the agent and thus the utilization rate of the computing resources configured for the agent.

[0043] The process of classifying multiple answers in the historical dialogue according to topics can be achieved by clustering. The following refers to Figure 2 Embodiments describing the method for generating topics and their content of the present disclosure.

[0044] Figure 2 FIG. shows a flowchart of a topic generation method according to some embodiments of the present disclosure. As Figure 2 shown, the topic generation method of this embodiment includes steps S111 to S115.

[0045] In step S111, multiple answers generated by the agent during the dialogue with one or more first users are clustered to obtain one or more categories of answers.

[0046] For example, a vector can be generated for each answer, and these vectors are clustered based on the distance between the vectors. According to the result of clustering the vectors, the result of clustering the answers can be determined. That is, the category to which a certain answer belongs is the same as the category to which the vector of this answer belongs. The answers in each category can be regarded as answers to the same topic.

[0047] The vector of each answer can be determined by the vectors corresponding to any one or more elements such as characters, words, images, etc. in the answer. For example, calculations such as splicing and weighting the vectors of multiple elements can be performed, and the present disclosure will not elaborate on this.

[0048] Before clustering, if necessary, some similar answers can also be merged to reduce the data processing volume, improve the processing efficiency, and save storage space.

[0049] In step S113, for each category, extract a theme from the responses of that category and determine it as the topic of that category, or extract a theme from the questions corresponding to the responses of that category and determine it as the topic of that category.

[0050] That is, the specific content of the theme can be extracted from the response itself or from the question corresponding to the response. For example, each category corresponds to one or more responses included in that category, or one or more questions corresponding to one or more responses of that category. A topic analysis model can be used to process these responses or these questions to extract a common theme. Alternatively, the topic can be determined based on these themes or the common statements or keywords of these responses.

[0051] The topic of each category can match any one of the responses under that category. Thus, the topic can represent the common core content of these responses.

[0052] In step S115, for each category, determine the responses of that category as the content corresponding to the topic of that category.

[0053] After determining the topic, the responses in the category corresponding to the topic can be associated with the topic. That is, these responses can be used as the associated content of this topic. Or it can also be understood that a content library based on this topic is established, and the content in the content library is these topics.

[0054] Through the above processing, one or more topics and the content associated with each topic can be obtained, and these contents are the responses generated by the agent in the historical conversation. By means of clustering, the topics involved in the agent's responses can be accurately mined, and these responses can be accurately classified under each topic. Thus, the target content matching the target topic can be generated more accurately.

[0055] When determining the target topic, it can be carried out from the attributes and characteristics of the topic itself, or from the perspective of the object to be recommended. Several ways to determine the target topic are exemplarily determined below.

[0056] In some embodiments, determining the target topic to be recommended from one or more topics includes: for each of the one or more topics, in response to the number of times the agent answers regarding that topic being greater than a threshold, determining that topic as an alternative topic; and determining the target topic to be recommended from one or more alternative topics.

[0057] The number of times the agent answers regarding a topic reflects the number of questions the user asks about that topic. When the number of questions is relatively high, it indicates that more users are interested in that topic. Therefore, such a topic might be a popular topic or a topic that easily raises questions among users. Consequently, this type of topic can be included in the candidate topics. Then, the candidate topic can be directly determined as the target topic, or based on other screening methods, the target topic can be selected from these candidate topics.

[0058] This method can make recommendations based on the popularity of the topic, making the recommended target content more likely to arouse the interest of the users who receive the recommendation. As a result, the probability of the target content being viewed increases significantly, improving the accuracy and effectiveness of the recommendation. In this way, it can also ensure that the resources consumed in the process of generating content and making recommendations are not wasted, enhancing the utilization rate of resources.

[0059] In some embodiments, determining the target topic to be recommended from one or more topics includes: determining the field to which each topic in the one or more topics belongs; determining the interaction information of the content in the same field as the field to which each topic belongs; and determining the target topic to be recommended based on the interaction information corresponding to each topic.

[0060] The field to which each topic belongs can be determined according to the preset correspondence between topics and fields. Alternatively, the technical field can also be directly determined based on the semantic analysis result of the topic, and this semantic analysis process can be completed using a natural language processing model. Then, the content in the same field as this topic can be determined, and this content can be from a content database, where the content and the interaction information involved in the content are recorded. Interaction information is the information about the interaction between the users who have viewed the content and the content, such as the number of views, the number of likes, the number of comments, the number of collections, etc. Of course, the emotional tendency of the users towards the content in this field, such as whether they are interested or not, can also be determined according to the content of the comments. Based on the interaction information, the recommendation degree of each field can be determined, and this recommendation degree can reflect the degree of interest of the users in this field. The interaction information corresponding to each topic is the interaction information of the field corresponding to each topic. Thus, the degree of interest of the users in this field also reflects the popularity of this topic. Therefore, the target topic to be recommended can be determined based on this.

[0061] As needed, candidate topics can also be determined based on the interaction information corresponding to each topic, and then the target topic can be determined from the candidate topics. That is, when determining the target topic, other screening conditions can also be referred to.

[0062] This method can make recommendations based on the interaction information in the field to which the topic belongs. The interaction information in the same field reflects the potential interaction information of the content of this topic. Therefore, making recommendations with this as a reference can improve the accuracy and effectiveness of the recommendations, and indirectly improve the utilization rate of resources.

[0063] In some embodiments, determining the target topic to be recommended from one or more topics includes: for each of the one or more topics, determining the timeliness of the topic according to the content corresponding to the topic; and determining the topic whose timeliness meets the timeliness condition as the target topic to be recommended.

[0064] The content corresponding to the topic itself may contain timeliness information, such as the time when certain events occur, the effective period of certain events, etc. Or, it is also possible to search for the corresponding time based on the content corresponding to the topic as the timeliness information. Those skilled in the art can also use other methods to determine the timeliness of the topic.

[0065] That the timeliness meets the timeliness condition may mean that the occurrence time of the content does not exceed a threshold from the current time, or it may mean that the effective time of the content has not expired, etc. Thus, the content that has recently occurred or has not expired can be recommended to users in a timely manner. Similarly, in this method, it is also possible to first determine the alternative topics and then determine the target topic from the alternative topics. That is, when determining the target topic, other screening conditions can also be referred to.

[0066] This method can make recommendations using the timeliness of the topic, and users tend to browse content with strong timeliness more easily. Therefore, making recommendations with this as a reference can improve the accuracy and effectiveness of the recommendations, and indirectly improve the utilization rate of resources.

[0067] The above several methods make recommendations from the perspective of the topic itself, and the target content determined by this method can be recommended to multiple users. That is, these contents can be widely covered to a user group. In addition, personalized recommendations can also be made for each user. Hereinafter, the user to receive the recommendation will be referred to as the second user. When making recommendations for the second user, recommendations can be made based on the recommendation strategy of the second user under the authorization of the second user. This disclosure exemplarily introduces a recommendation method.

[0068] Figure 3 The flowchart shows the method for determining the target topic according to some embodiments of the present disclosure. As Figure 3 shown, the method for determining the target topic in this embodiment includes steps S131 to S135.

[0069] In step S131, with the authorization of the second user, the historical conversation data between the second user and the intelligent agent is divided into multiple paragraphs, and the conversations in each paragraph belong to the same topic.

[0070] The historical conversation between the second user and the intelligent agent is long text data. By splitting these long texts, one or more topics discussed by the user in the conversation can be determined. For example, these conversations can be clustered to split the conversations. When clustering, referring to the foregoing embodiments, the text of the conversation can be converted into a vector for processing.

[0071] In step S133, for each paragraph, according to the feedback of the second user on the message from the intelligent agent in the paragraph, the recommendation degree of the topic is determined.

[0072] In each paragraph, the user can give feedback by sending a message (i.e., text feedback), or can give feedback through some operations (such as liking, collecting, etc.). Through these feedbacks of the user, the emotional tendency of the user towards the topic (such as affirmative, negative, neutral) can be determined. Or, the degree of interest of the user in the topic can be determined. For example, if the user sends a lot of messages about the topic, the user is more interested in the topic.

[0073] In some embodiments, the feedback of the second user on the message from the intelligent agent includes the message from the second user. Determining the recommendation degree of the topic includes: for the conversation in each paragraph, performing semantic analysis on the message from the second user in the paragraph to determine at least one of the number of times of asking questions about the topic and the evaluation information; according to the semantic analysis result, determining the recommendation degree of the topic. For example, through semantic analysis, it can be determined whether the message sent by the user is a follow-up question to the answer provided by the intelligent agent, and whether the message sent by the user expresses an evaluation of the answer of the intelligent agent.

[0074] The semantic analysis result can be understood as including at least one of the number of times of asking questions and the evaluation information. The number of times the user asks questions is positively correlated with the recommendation degree of the topic. The positive degree of the user's evaluation is also positively correlated with the recommendation degree of the topic.

[0075] In step S135, according to the topics with a recommendation degree higher than the recommendation condition, and the matching result between the topics with a recommendation degree higher than the recommendation condition and one or more topics, the target topic to be recommended to the second user is determined.

[0076] For the topics with a high recommendation degree, it can be considered that they are topics that the user is more likely to browse or watch after receiving relevant content recommendations. Based on this topic, related topics can be determined, and the user is very likely to be interested in the content under this topic.

[0077] Based on the long conversation text between the user and the agent, the above embodiments mine the user's interest points. Thus, with the user's authorization, the generated conversation information can be utilized more fully. Moreover, the recommended target topics will be more matched with the user's interests or concerns, improving the accuracy and effectiveness of the recommendation, and indirectly enhancing the utilization rate of resources.

[0078] When generating the target content, a machine learning model can be utilized. Figure 4 The flowchart of the target content generation method according to some embodiments of the present disclosure is shown. As Figure 4 shown, the target content generation method of this embodiment includes steps S151 to S153.

[0079] In step S151, according to the target topic and the content corresponding to the target topic, a prompt is determined, and this prompt is used to indicate generating content that matches the target topic based on the content corresponding to the target topic.

[0080] For example, the target topic and content can be filled into a preset prompt information template to generate the prompt. The prompt can also include processing instructions to indicate what kind of processing the machine learning model needs to perform.

[0081] In step S153, the prompt is input into the machine learning model to obtain the target content to be recommended output by the machine learning model.

[0082] The machine learning model can be an LLM model or a base model. Through the powerful processing ability of the machine learning model, the output content can be obtained efficiently. The output content can match the target topic. And since the generated target content is based on the content corresponding to the target topic, and this content is already the answer provided by the agent to the user before, during the generation process, the amount of additional information obtained by the agent is less, which can also improve the generation efficiency. Of course, if needed, the agent can also perform additional information acquisition, such as accessing the information library, online searching, etc., to further improve the quality and richness of the generated target content.

[0083] The above way of generating content using the machine learning model is just an example. According to needs, other generation methods can also be adopted. For example, the content under this topic can also be directly de-duplicated and spliced simply to generate new target content.

[0084] The generated target content can have the same data type as the content under the topic, or a different data type.

[0085] For example, if the target content includes graphic content, graphic content can be generated using the text corresponding to the topic. That is to say, the target content to be recommended can include: generating an image corresponding to the text based on the text in the content corresponding to the target topic; generating the graphic content to be recommended based on the text in the content corresponding to the target topic and the generated image. When generating an image using text, a "text-to-image" model can be used. Thus, some text content can be presented in the form of an image, improving the readability and richness of the generated target content, so as to increase the success rate of recommendation.

[0086] Also for example, if the target content includes video, video can be generated using the text or image corresponding to the topic. That is to say, the target content to be recommended can include: generating the video to be recommended based on the text and image in the content corresponding to the target topic. For example, the text can be used as supplementary information for the image, generating more images based on the original content's image, and generating a video based on the original image and the generated images. Thus, video content can be generated efficiently. Of course, in the case where the original content only includes text, multiple images can also be generated first based on the text, and then a video can be generated based on the multiple images.

[0087] Through the above two exemplary methods, during the process of generating the target content, the data type of the original content can be converted, thus obtaining a target content with a richer form to increase the success rate of recommendation.

[0088] Before presenting the target content, the target content can be recommended to the user through a content stream (Feed). The content stream has multiple presentation methods. One is to display each content in the content stream in the main part of the interface, such as full-screen display of the currently playing content. The user can switch between different contents in the content stream through a switching operation (such as swiping up and down, or swiping left and right). The other is to display the entrances of multiple contents in a waterfall flow manner on the interface, including the target content. In response to the user's trigger operation on one of the entrances, the corresponding content is displayed. For example, in response to the trigger operation on the entrance (such as the cover) of the target content, the target content is displayed. When the user swipes up and down or left and right the interface of the waterfall flow, the entrances of more content recommended for the user can be displayed.

[0089] The entrances of the content stream can also be set in multiple interfaces such as the conversation interface or the recommendation interface.

[0090] In some embodiments, presenting target content includes: displaying an entry of a content stream in the dialogue interface of the agent; in response to a trigger operation on the entry, displaying the content stream, where the content stream includes the target content. The number of target contents can be one or more. The content stream may only include target contents determined by the embodiments of the present disclosure, or may also include recommended contents determined by using other recommendation strategies.

[0091] For example, when the user jumps from other interfaces to the dialogue interface, one or more contents actively recommended by the agent can be displayed. Of course, when the user has not opened the dialogue interface, if the user authorizes the function of the agent to actively push, the information of the target content can also be pushed to the user through a notification message. After the user triggers the notification, the dialogue interface is opened, and the entry of the content stream is displayed.

[0092] By using the dialogue interface to recommend target content to the user, the user can actively understand the generation ability of the agent before the dialogue, which is convenient for promoting the dialogue between the user and the agent and improving the utilization rate of the agent.

[0093] The agent can also partially actively trigger the recommendation process. For example, generating the target content to be recommended includes: displaying the recommended target topic; in response to a trigger operation on the target topic, generating the target content to be recommended based on the content corresponding to the target topic. That is, the target topic can be first presented to the user, and when the user triggers the topic, the generation process of the target content is then executed. This method can generate target content on demand and saves computing resources. The recommended target topic can be displayed in the dialogue interface between the user and the agent. For example, when the user jumps from other interfaces to the dialogue interface, or when the agent sends a reply message to the user, a trigger control including the target topic is displayed.

[0094] In some embodiments, presenting target content includes: displaying an entry of one or more contents in the recommended content stream in the recommendation interface, where the one or more contents include the target content; in response to a trigger on the entry of the target content, displaying the target content. The entry includes information such as the cover, name, author, etc. of the recommended content (including the target content), so as to facilitate the user to preview each content and select the content of interest to watch. In the recommendation interface, in addition to the above-mentioned target content generated by the agent, it may also include any one or more of the contents published by real users, the contents generated by the platform, the contents generated by other agents, etc. Thus, on the recommendation page, the user can receive contents from multiple sources, expanding the types of information obtained by the user.

[0095] After the user browses the content, questions can be asked based on the target content. Suppose a second user receives the recommended target content. In some embodiments, a first message sent by the second user to the agent is displayed, and the message is associated with the target content; a second message for replying to the first message is generated based on the first message and the content corresponding to the target topic; and the second message sent by the agent is displayed. On the display or playback interface of the target content, a message sending control, a thumbnail of the conversation interface, or an entry to the conversation interface can be set to facilitate the user to send a message to the agent. Thus, after the user browses the target content, the user can continue to ask one or more rounds of questions about the content, so that the user can efficiently obtain more information related to the target content. The information generated by the agent subsequently can also be generated based on the content related to the target topic. Of course, it can also be generated by referring to the content provided by other information sources.

[0096] Figure 5 FIG. shows a schematic diagram of a conversation interface according to some embodiments of the present disclosure. As Figure 5 shown, in the conversation interface 5, user 1 sent a message 51 "What are the interesting places in City A", and agent A sent a message 52 to user 1 as an answer, and the answer included some travel guides for City A. In the case where other users also have similar questions, agent A sorted out these contents and generated target content related to the tourism in City A. The target content is, for example, video 1 generated by the agent based on the answers including message 52.

[0097] Figure 6 FIG. shows a schematic diagram of a recommendation interface according to some embodiments of the present disclosure. As Figure 6 shown, in the recommendation interface 6, multiple contents recommended to a certain user are displayed, including an entry 61 for video 1. Thus, by triggering 61, the user can browse video 1 generated by the agent.

[0098] The above introduces the method embodiments of the present disclosure. Next, the apparatus for executing the above methods is further described.

[0099] Figure 7 FIG. shows a schematic structural diagram of an information recommendation apparatus according to some embodiments of the present disclosure. As Figure 7As shown, the information recommendation device 7 of this embodiment includes: a topic generation module 71 configured to generate one or more topics and content corresponding to each topic according to multiple answers generated by the agent during the conversation with one or more first users, where the multiple answers are authorized by the one or more first users and do not involve the personal information of the one or more first users; a determination module 72 configured to determine a target topic to be recommended from the one or more topics; a content generation module 73 configured to generate target content to be recommended based on the content corresponding to the target topic; and a display module 74 configured to display the target content.

[0100] The above embodiment can make full use of the systematic and high-quality content generated by the agent, improve the utilization rate of data, and the generated target content can also have higher quality. Moreover, this active recommendation method of the agent also improves the interactivity between the user and the agent, thus improving the utilization rate of the agent, and further improving the utilization rate of the computing resources configured for the agent.

[0101] In some embodiments, the topic generation module 71 is configured to: cluster multiple answers generated by the agent during the conversation with one or more first users to obtain one or more categories of answers; for each category, extract a theme from the answers of the category and determine it as the topic of the category, or extract a theme from the questions corresponding to the answers of the category and determine it as the topic of the category; for each category, determine the answers of the category as the content corresponding to the topic of the category.

[0102] In some embodiments, the determination module 72 is configured to: for each of the one or more topics, in response to the number of answers of the agent to the topic being greater than a threshold, determine the topic as an alternative topic; determine the target topic to be recommended from the one or more alternative topics.

[0103] In some embodiments, the determination module 72 is configured to: determine the field to which each topic in the one or more topics belongs; determine the interaction information of the content in the same field as the field to which each topic belongs; determine the target topic to be recommended according to the interaction information corresponding to each topic.

[0104] In some embodiments, the determination module 72 is configured to: for each topic in the one or more topics, determine the timeliness of the topic according to the content corresponding to the topic; determine the topic whose timeliness meets the timeliness condition as the target topic to be recommended.

[0105] In some embodiments, the determination module 72 is configured to: with the authorization of the second user, divide the historical conversation data between the second user and the intelligent agent into multiple paragraphs, and the conversation in each paragraph belongs to the same topic; for each paragraph, determine the recommendation degree of the topic based on the feedback of the second user on the message from the intelligent agent in the paragraph; determine the target topic to be recommended to the second user based on the matching results between the topics with a recommendation degree higher than the recommendation condition and the topics with a recommendation degree higher than the recommendation condition and one or more topics.

[0106] In some embodiments, the second user's feedback on the message from the intelligent agent includes a message from the second user, and the determination module 72 is configured to: for each paragraph of dialogue, perform semantic analysis on the message from the second user in the paragraph to determine at least one of the number of follow-up questions and evaluation information on the topic; and determine the recommendation degree of the topic based on the semantic analysis results.

[0107] In some embodiments, the content generation module 73 is configured to: determine prompt information based on the target topic and the content corresponding to the target topic, the prompt information is used to indicate that content matching the target topic is generated based on the content corresponding to the target topic; input the prompt information into the machine learning model to obtain the target content to be recommended output by the machine learning model.

[0108] In some embodiments, the content generation module 73 is configured to: generate an image corresponding to the text based on the text in the content corresponding to the target topic; generate graphic content to be recommended based on the text in the content corresponding to the target topic and the generated image.

[0109] In some embodiments, the content generation module 73 is configured to generate a video to be recommended based on text and images in the content corresponding to the target topic.

[0110] In some embodiments, the display module 74 is configured to display a first message sent by a second user to the intelligent entity, wherein the message is associated with the target content; the device 7 also includes a message generation module, which is configured to generate a second message for replying to the first message based on the first message and the content corresponding to the target topic; the display module 74 is configured to display the second message sent by the intelligent entity.

[0111] In some embodiments, the display module 74 is configured to: display an entrance to the content flow in the dialogue interface of the intelligent agent; and display the content flow in response to a triggering operation on the entrance, wherein the content flow includes the target content.

[0112] In some embodiments, the content generation module 73 is configured to: display a recommended target topic; and generate target content to be recommended based on content corresponding to the target topic in response to a trigger operation on the target topic.

[0113] In some embodiments, the display module 74 is configured to: in the recommendation interface, display the entry of one or more contents in the recommended content stream, where the one or more contents include the target content; and in response to the triggering of the entry of the target content, display the target content.

[0114] Figure 8 The block diagram of an electronic device according to some embodiments of the present disclosure is shown.

[0115] The memory 81 is used to store one or more computer-readable instructions. The memory 81 may include any combination of various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory, including but not limited to random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), read-only memory (ROM), and flash memory. The memory 81 may store, for example, an operating system, application programs, a boot loader, a database, and other programs, and may also store various application programs and various data.

[0116] The processor 82 is used to run the computer-readable instructions to implement the method described in any of the foregoing embodiments. For the specific implementation of each step of the method, reference may be made to the above embodiments, and the repeated parts will not be elaborated herein.

[0117] The processor 82 may be configured to execute the steps of the foregoing embodiments. The processor 82 may be embodied as various processing devices, such as a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The central processing unit (CPU) may be of the X86 or ARM architecture, etc.

[0118] The processor 82 and the memory 81 may communicate with each other directly or indirectly. For example, the processor 82 and the memory 81 may communicate through a network. The network may include a wireless network, a wired network, and / or any combination of a wireless network and a wired network. The processor 82 and the memory 81 may also communicate with each other through a system bus, and the present disclosure does not limit this.

[0119] It should be noted that Figure 8 The components of the illustrated electronic device 8 are merely exemplary and not restrictive. According to actual application needs, the electronic device 8 may also have other components. The processor 82 may control other components in the electronic device 8 to perform the desired functions.

[0120] The electronic device 8 can be implemented in software, firmware, and / or hardware, and can be integrated into a device installed with relevant application programs.

[0121] Figure 9 The block diagram of an electronic device according to other embodiments of the present disclosure is shown.

[0122] Figure 9 The illustrated electronic device 9 can be a computer system with a dedicated hardware structure, and can perform corresponding functions when installed with relevant application programs.

[0123] The electronic device includes but is not limited to mobile terminals such as smart phones, laptop computers, personal digital assistants (PDAs), tablet personal computers (Tablet PCs), PMPs (portable multimedia players), vehicle terminals (such as vehicle navigation terminals), wearable devices, etc., and fixed terminals such as digital TVs, desktop computers, etc.

[0124] As Figure 9 shown, the central processing unit (CPU) 91 executes various processes according to the programs stored in the read-only memory (ROM) 92 or the programs loaded from the storage section 98 into the random access memory (RAM) 93. In the RAM 93, data required when the CPU 91 executes various processes, etc. is stored as needed. The central processing unit is merely exemplary, and it can also be other types of processors, such as the various processors described above. The ROM 92, RAM 93, and storage section 98 can be various forms of computer-readable storage media. It should be noted that although Figure 9 the ROM 92, RAM 93, and storage section 98 are shown separately, one or more of them can be combined, or located in the same or different memories or storage modules.

[0125] The CPU 91, ROM 92, and RAM 93 are connected to each other via the bus 94. The input / output interface 95 is also connected to the bus 94.

[0126] The following components are connected to the input / output interface 95: the input section 96, such as a touch screen, touch pad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; the output section 97, including a display, such as a cathode ray tube (CRT), liquid crystal display (LCD), speaker, vibrator, etc.; the storage section 98, including a hard disk, magnetic tape, etc.; and the communication section 99, including a network interface card such as a LAN card, modem, etc. The communication section 99 allows communication processing to be performed via a network such as the Internet. It is easy to understand that although Figure 9The components shown in the electronic device 9 communicate via the bus 94, but they can also communicate via a network or other means. Here, the network can include a wireless network, a wired network, and / or any combination of a wireless network and a wired network.

[0127] As needed, the driver 910 is also connected to the input / output interface 95. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the driver 910 as needed, so that the computer program read therefrom is installed in the storage section 98 as needed.

[0128] In the case where the above-described series of processes are implemented by software, the programs constituting the software can be installed from a network such as the Internet or a storage medium such as the removable medium 911.

[0129] According to an embodiment of the present disclosure, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, some embodiments of the present disclosure include a computer program product that, when run on a computer, causes the computer to implement the method described in any of the foregoing embodiments. The computer program product includes computer instructions carried on a computer-readable medium, including program code for executing the method shown in the flowchart. In such an embodiment, the computer instructions can be downloaded and installed from the network via the communication section 99, or installed from the storage section 98, or installed from the ROM 92. When the computer program is executed by the CPU 91, the method of the embodiment of the present disclosure is executed.

[0130] It should be noted that, in the context of the present disclosure, a computer-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device.

[0131] A computer-readable medium can be a computer-readable storage medium, a computer-readable signal medium, or any combination of the two.

[0132] Computer-readable storage media include, but are not limited to: electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or components, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present disclosure, a computer-readable storage media can be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, device, or component. Computer instructions are stored on the computer-readable storage media, and when executed by a processor, implement the method described in any of the foregoing embodiments.

[0133] A computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or component. The program code contained on the computer-readable medium can be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0134] The above computer-readable media can be included in the above electronic devices; or can exist separately without being assembled into the electronic devices.

[0135] In some embodiments, a computer program is also provided, including: instructions that, when executed by a processor, cause the processor to execute the method described in any of the foregoing embodiments. For example, the instructions can be embodied as computer program code.

[0136] In embodiments of the present disclosure, computer program code for performing the operations of the present disclosure may be written in one or more programming languages or combinations thereof. The above-mentioned programming languages include, but are not limited to, object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network (including a local area network (LAN) or a wide area network (WAN)), or may be connected to an external computer (for example, by connecting through the Internet using an Internet service provider).

[0137] Some embodiments of the present disclosure provide an information recommendation method, including: generating one or more topics and content corresponding to each topic according to multiple answers generated by an agent during a conversation with one or more first users, the multiple answers being authorized by the one or more first users and not involving personal information of the one or more first users; determining a target topic to be recommended from the one or more topics; generating target content to be recommended based on the content corresponding to the target topic; and presenting the target content.

[0138] In some embodiments, generating one or more topics and content corresponding to each topic according to multiple answers generated by an agent during a conversation with one or more first users includes: clustering the multiple answers generated by the agent during a conversation with one or more first users to obtain one or more categories of answers; for each category, extracting a theme from the answers of the category and determining it as the topic of the category, or extracting a theme from the questions corresponding to the answers of the category and determining it as the topic of the category; for each category, determining the answers of the category as the content corresponding to the topic of the category.

[0139] In some embodiments, determining a target topic to be recommended from the one or more topics includes: for each of the one or more topics, in response to the number of answers of the agent to the topic being greater than a threshold, determining the topic as an alternative topic; and determining the target topic to be recommended from the one or more alternative topics.

[0140] In some embodiments, determining a target topic to be recommended from the one or more topics includes: determining the field to which each topic of the one or more topics belongs; determining the interaction information of the content in the same field as the field to which each topic belongs; and determining the target topic to be recommended according to the interaction information corresponding to each topic.

[0141] In some embodiments, determining a target topic to be recommended from one or more topics includes: for each of the one or more topics, determining the timeliness of the topic according to the content corresponding to the topic; and determining the topic whose timeliness meets the timeliness condition as the target topic to be recommended.

[0142] In some embodiments, determining a target topic to be recommended from one or more topics includes: when authorized by a second user, dividing the historical conversation data between the second user and the agent into multiple paragraphs, and the conversations in each paragraph belong to the same theme; for each paragraph, determining the recommendation degree of the theme according to the feedback of the second user on the message from the agent in the paragraph; and determining the target topic to be recommended to the second user according to the theme whose recommendation degree is higher than the recommendation condition and the matching result between the theme whose recommendation degree is higher than the recommendation condition and one or more topics.

[0143] In some embodiments, the feedback of the second user on the message from the agent includes the message from the second user. For each paragraph, determining the recommendation degree of the theme according to the feedback of the second user on the message from the agent in the paragraph includes: for the conversation in each paragraph, performing semantic analysis on the message from the second user in the paragraph to determine at least one of the number of times of asking questions about the theme and the evaluation information; and determining the recommendation degree of the theme according to the semantic analysis result.

[0144] In some embodiments, generating target content to be recommended based on the content corresponding to the target topic includes: determining a prompt message according to the target topic and the content corresponding to the target topic, where the prompt message is used to indicate generating content matching the target topic according to the content corresponding to the target topic; and inputting the prompt message into a machine learning model to obtain the target content to be recommended output by the machine learning model.

[0145] In some embodiments, the target content includes graphic and text content. Generating target content to be recommended based on the content corresponding to the target topic includes: generating an image corresponding to the text according to the text in the content corresponding to the target topic; and generating the graphic and text content to be recommended according to the text in the content corresponding to the target topic and the generated image.

[0146] In some embodiments, the target content includes a video. Generating target content to be recommended based on the content corresponding to the target topic includes: generating the video to be recommended according to the text and image in the content corresponding to the target topic.

[0147] In some embodiments, the information recommendation method further includes: displaying a first message sent by the second user to the agent, where the message is associated with the target content; generating a second message for replying to the first message according to the first message and the content corresponding to the target topic; and displaying the second message sent by the agent.

[0148] In some embodiments, presenting target content includes: displaying an entry for a content stream in the conversation interface of the agent; in response to a trigger operation on the entry, displaying the content stream, where the content stream includes the target content.

[0149] In some embodiments, generating target content to be recommended based on content corresponding to a target topic includes: displaying the recommended target topic; in response to a trigger operation on the target topic, generating the target content to be recommended based on the content corresponding to the target topic.

[0150] In some embodiments, presenting target content includes: in a recommendation interface, displaying an entry for one or more contents in a recommended content stream, where the one or more contents include the target content; in response to a trigger on the entry for the target content, displaying the target content.

[0151] In some embodiments, the present disclosure also provides an electronic device, including: a memory; and a processor coupled to the memory, where the processor is configured to execute the information recommendation method of any embodiment of the present disclosure based on instructions stored in the memory. The above embodiments improve the utilization rate of data, and the generated target content can also have higher quality. Moreover, the active recommendation method of the agent also improves the interactivity between the user and the agent, thus improving the utilization rate of the agent, and further improving the utilization rate of the computing resources configured for the agent.

[0152] In some embodiments, the present disclosure also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the processor is caused to implement the information recommendation method of any embodiment of the present disclosure. The above embodiments improve the utilization rate of data, and the generated target content can also have higher quality. Moreover, the active recommendation method of the agent also improves the interactivity between the user and the agent, thus improving the utilization rate of the agent, and further improving the utilization rate of the computing resources configured for the agent.

[0153] In some embodiments, the present disclosure also provides a computer program product, and when the computer program product runs on a computer, the computer is caused to implement the information recommendation method of any embodiment of the present disclosure. The above embodiments improve the utilization rate of data, and the generated target content can also have higher quality. Moreover, the active recommendation method of the agent also improves the interactivity between the user and the agent, thus improving the utilization rate of the agent, and further improving the utilization rate of the computing resources configured for the agent.

[0154] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a portion of code that contains one or more executable instructions for implementing the specified logical function. It should also be noted that, in some alternative implementations, the functions noted in the blocks may occur in a different order than that noted in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0155] The functions described above can be performed, at least in part, by one or more hardware logic components. By way of example, and not limitation, exemplary hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), and the like.

[0156] Although some specific embodiments of the present disclosure have been described in detail by way of example, those skilled in the art should understand that the above examples are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Those skilled in the art should understand that the above embodiments can be modified without departing from the scope and spirit of the present disclosure. The scope of the present disclosure is defined by the appended claims.

Claims

1. An information recommendation method, comprising: Generate one or more topics and content corresponding to each topic according to a plurality of answers generated by the agent in the process of talking with the one or more first users, wherein the plurality of answers are authorized by the one or more first users and do not involve personal information of the one or more first users; Determining a target topic to be recommended from the one or more topics; generating target content to be recommended based on the content corresponding to the target topic; Display the target content.

2. The information recommendation method according to claim 1, wherein: The generating one or more topics and the content corresponding to each topic according to the multiple answers generated by the intelligent agent in the process of talking with the one or more first users includes: Clustering multiple answers generated by the agent during the conversation with one or more first users to obtain one or more categories of answers; For each category, extracting a theme from the answers of the category and determining it as the topic of the category, or extracting a theme from the questions corresponding to the answers of the category and determining it as the topic of the category; For each of the categories, an answer for the category is determined as content corresponding to a topic of the category.

3. The information recommendation method according to claim 1, wherein: Determining a target topic to be recommended from the one or more topics includes: For each of the one or more topics, in response to the number of answers given by the agent to the topic being greater than a threshold, determining the topic as a candidate topic; The target topic to be recommended is determined from one or more candidate topics.

4. The information recommendation method according to claim 1 or 3, wherein: Determining a target topic to be recommended from the one or more topics includes: determining a domain to which each of the one or more topics belongs; Determine interactive information about content in the same field as the field to which each topic belongs; The target topic to be recommended is determined according to the interaction information corresponding to each topic.

5. The information recommendation method according to claim 1, wherein: Determining a target topic to be recommended from the one or more topics includes: For each of the one or more topics, determining the timeliness of the topic based on content corresponding to the topic; The topics whose timeliness meets the timeliness condition are determined as the target topics to be recommended.

6. The information recommendation method according to claim 1, wherein: Determining a target topic to be recommended from the one or more topics includes: With the authorization of the second user, dividing the historical conversation data between the second user and the agent into a plurality of paragraphs, wherein the conversation in each paragraph belongs to the same topic; For each paragraph, determining the recommendation degree of the topic according to the feedback of the second user to the message from the intelligent agent in the paragraph; The target topic to be recommended to the second user is determined according to the topic whose recommendation degree is higher than the recommendation condition and the matching result between the topic whose recommendation degree is higher than the recommendation condition and the one or more topics.

7. The information recommendation method according to claim 6, wherein: The feedback of the second user on the message from the agent includes the message from the second user, and for each paragraph, determining the recommendation degree of the topic according to the feedback of the second user on the message from the agent in the paragraph includes: For each paragraph of the conversation, semantic analysis is performed on the message from the second user in the paragraph to determine at least one of the number of follow-up questions and evaluation information on the topic; The recommendation degree of the topic is determined according to the semantic analysis result.

8. The information recommendation method according to claim 1, wherein: The generating target content to be recommended based on the content corresponding to the target topic includes: Determining prompt information according to the target topic and the content corresponding to the target topic, wherein the prompt information is used to instruct to generate content matching the target topic according to the content corresponding to the target topic; The prompt information is input into a machine learning model to obtain the target content to be recommended output by the machine learning model.

9. The information recommendation method according to claim 1 or 8, wherein: The target content includes graphic content, and the generating of the target content to be recommended based on the content corresponding to the target topic includes: generating an image corresponding to the text according to the text in the content corresponding to the target topic; The graphic content to be recommended is generated according to the text in the content corresponding to the target topic and the generated image.

10. The information recommendation method according to claim 1 or 8, wherein: The target content includes a video, and generating the target content to be recommended based on the content corresponding to the target topic includes: The video to be recommended is generated based on the text and the image in the content corresponding to the target topic.

11. The information recommendation method according to claim 1, further comprising: displaying a first message sent by a second user to the agent, wherein the message is associated with the target content; generating, according to the first message and content corresponding to the target topic, a second message for replying to the first message; The second message sent by the agent is displayed.

12. The information recommendation method according to claim 1, wherein: The display of the target content includes: In the dialogue interface of the agent, displaying an entrance to the content stream; In response to a triggering operation on the portal, the content stream is displayed, the content stream including the target content.

13. The information recommendation method according to claim 12, wherein: The generating target content to be recommended based on the content corresponding to the target topic includes: Displaying the recommended target topic; In response to a triggering operation on the target topic, the target content to be recommended is generated based on the content corresponding to the target topic.

14. The information recommendation method according to claim 1, wherein: The display of the target content includes: On the recommendation interface, displaying an entry of one or more contents in the recommended content stream, wherein the one or more contents include the target content; In response to triggering an entry to the target content, the target content is displayed.

15. An electronic device, comprising: Memory; as well as A processor coupled to the memory, wherein the processor is configured to execute the information recommendation method according to any one of claims 1 to 14 based on instructions stored in the memory.

16. A computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, the processor implements the information recommendation method according to any one of claims 1 to 14. 17 . A computer program product, when the computer program product is executed on a computer, enables the computer to implement the information recommendation method according to any one of claims 1 to 14.

Images