Resource recommendation method and device based on large model, intelligent agent, equipment and medium

Abstract

The invention provides a resource recommendation method and device based on a large model, an intelligent agent, equipment and a medium, and relates to the technical field of computer technologies, in particular to the fields of artificial intelligence, NLP and intelligent search. According to the specific implementation scheme, first demand information of an object is processed through a large model, basic recommendation resources and extended topic topology related to basic demand attributes of the object are obtained, and the extended topic topology comprises multiple extended topics representing potential demand attributes of the object and the incidence relation among the multiple extended topics; the association relationship represents a hierarchical relationship among the plurality of extension topics; pushing feedback information determined based on the basic recommendation resources; in response to a first target request of the object for the feedback information, performing theme intention recognition on second demand information of the first target request and the extended theme topology to obtain a first target extended theme; and pushing a first extension recommendation resource determined based on the first target extension theme to the object.

Description

Technical Field

[0001] The present disclosure relates to the technical field of computer technology, specifically to the technical fields of artificial intelligence, NLP (Natural Language Processing), intelligent search, etc., and especially to resource recommendation methods, devices, intelligent agents, equipment, media and program products based on large models. Background Art

[0002] With the increasing popularity and deepening application of artificial intelligence technology, Large Language Model (LLM) has shown great potential in the field of resource recommendation and content generation with its excellent natural language understanding and generation capabilities. LLM can identify user intentions and provide users with personalized resource recommendation services. Summary of the invention

[0003] The present disclosure provides a method, apparatus, intelligent agent, device, medium and program product for a resource recommendation method based on a large model.

[0004] According to one aspect of the present disclosure, a resource recommendation method based on a big model is provided, comprising: using a big model to process first demand information of an object, obtaining basic recommended resources and extended topic topology related to the basic demand attributes of the object, the extended topic topology comprising multiple extended topics representing potential demand attributes of the object, and association relationships between the multiple extended topics, the association relationships representing the hierarchical relationship between the multiple extended topics; pushing feedback information determined based on the basic recommendation resources; in response to a first target request of the object for the feedback information, performing topic intent recognition on the second demand information and extended topic topology of the first target request, obtaining a first target extended topic; and pushing a first extended recommended resource determined based on the first target extended topic to the object.

[0005] According to another aspect of the present disclosure, a resource recommendation device based on a big model is provided, including: a processing module, used to use the big model to process the first demand information of an object, and obtain basic recommended resources and extended topic topology related to the basic demand attributes of the object, the extended topic topology including multiple extended topics representing the potential demand attributes of the object, and the association relationship between the multiple extended topics, the association relationship representing the hierarchical relationship between the multiple extended topics; a first push module, used to push feedback information determined based on the basic recommendation resources; an identification module, used to respond to the object's first target request for the feedback information, perform topic intent identification on the second demand information and extended topic topology of the first target request, and obtain a first target extended topic; and a second push module, used to push the first extended recommended resource determined based on the first target extended topic to the object.

[0006] According to another aspect of the present disclosure, there is provided an artificial intelligence agent, comprising: an input module for receiving input information; a processing module for determining a target task based on the input information received by the input module, determining a large language model based on the target task, and executing the above method by calling the large language model; and an output module for outputting output information obtained by the processing module.

[0007] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described above.

[0008] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute the method as described above.

[0009] According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein the computer program implements the method described above when executed by a processor.

[0010] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0012] Figure 1 An exemplary system architecture to which a large model-based resource recommendation method and device can be applied according to an embodiment of the present disclosure is schematically shown;

[0013] Figure 2 The flowchart of the resource recommendation method based on the big model according to the embodiment of the present disclosure is schematically shown;

[0014] Figure 3A The effect diagram of feedback information display according to the embodiment of the present disclosure is schematically shown;

[0015] Figure 3B Schematic diagram showing the effect of operating the recommended resources according to the embodiment of the present disclosure;

[0016] Figure 3C Schematic diagram showing the effect of operating the recommended resources according to another embodiment of the present disclosure;

[0017] Figure 4 A schematic diagram of determining a first sub-topology according to an embodiment of the present disclosure is shown;

[0018] Figure 5 A schematic diagram of determining a second sub-topology according to an embodiment of the present disclosure is shown;

[0019] Figure 6 The flowchart of a resource recommendation method based on a large model according to another embodiment of the present disclosure is schematically shown;

[0020] Figure 7 A block diagram of a resource recommendation device based on a large model according to an embodiment of the present disclosure is schematically shown;

[0021] Figure 8 The structure diagram of the artificial intelligence agent according to the embodiment of the present disclosure is schematically shown.

[0022] Fig. 9 A block diagram of an electronic device suitable for implementing a resource recommendation method based on a large model according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0023] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0024] In the technical solution disclosed in the present invention, the acquisition, storage and application of user personal information involved are in compliance with the provisions of relevant laws and regulations, necessary confidentiality measures are taken, and do not violate public order and good morals.

[0025] Figure 1 An exemplary system architecture to which the large model-based resource recommendation method and apparatus can be applied according to an embodiment of the present disclosure is schematically shown.

[0026] It should be noted that Figure 1 The examples shown are only examples of system architectures to which the embodiments of the present disclosure can be applied, in order to help those skilled in the art understand the technical content of the present disclosure, but do not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios. For example, in another embodiment, an exemplary system architecture to which the large model-based resource recommendation method and apparatus can be applied may include a terminal device, but the terminal device may implement the large model-based resource recommendation method and apparatus provided by the embodiments of the present disclosure without interacting with the server.

[0027] like Figure 1 As shown, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0028] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as knowledge reading applications, web browser applications, search applications, instant messaging tools, email clients and / or social platform software, etc. (only as examples).

[0029] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0030] The server 105 may be a server that provides various services, such as a background management server (only an example) that provides support for the content browsed by the user using the terminal devices 101, 102, and 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to the user request) to the terminal device.

[0031] It should be noted that the resource recommendation method based on a large model provided in the embodiment of the present disclosure can generally be executed by the terminal device 101, 102, or 103. Accordingly, the resource recommendation apparatus based on a large model provided in the embodiment of the present disclosure can also be set in the terminal device 101, 102, or 103.

[0032] Alternatively, the resource recommendation method based on the big model provided in the embodiment of the present disclosure may also be generally executed by the server 105. Accordingly, the resource recommendation device based on the big model provided in the embodiment of the present disclosure may generally be arranged in the server 105. The resource recommendation method based on the big model provided in the embodiment of the present disclosure may also be executed by a server or a server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the resource recommendation device based on the big model provided in the embodiment of the present disclosure may also be arranged in a server or a server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105.

[0033] For example, when a user makes a query, the terminal devices 101, 102, 103 can obtain the user's demand information, and then send the demand information to the server 105, which analyzes the demand information, determines the subject information of the demand information, predicts the resources that the user is interested in based on the subject information of the demand information, and pushes the resources to the user. Alternatively, a server or server cluster that can communicate with the terminal devices 101, 102, 103 and / or the server 105 analyzes the demand information and finally recommends resources to the user.

[0034] In the related art, the resource recommendation method based on large models usually responds to user queries and provides users with relevant resource content. It lacks the means to better extend related topics of interest to users and cannot guide and stimulate users' potential interests.

[0035] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0036] Figure 2 The flowchart of the resource recommendation method based on a large model according to an embodiment of the present disclosure is schematically shown.

[0037] like Figure 2 As shown, the method includes operations S210 to S240.

[0038] In operation S210, the first demand information of the object is processed using the large model to obtain basic recommended resources and extended topic topology related to the basic demand attributes of the object.

[0039] In operation S220, feedback information determined based on the basic recommended resources is pushed.

[0040] In operation S230, in response to the first target request of the object for feedback information, topic intent recognition is performed on the second requirement information and the extended topic topology of the first target request to obtain a first target extended topic.

[0041] In operation S240, a first extended recommendation resource determined based on the first target extended theme is pushed to the object.

[0042] In the embodiments of the present disclosure, a large model may refer to a deep learning model with large-scale model parameters, and a large model generally contains hundreds of millions, tens of billions, hundreds of billions, trillions, or even more than ten trillion model parameters. The large model may include multimodal large models such as the VideoChat model and the Video-LlaMA model. Alternatively, the large model may also be a Large Language Model model, a visual large model, or the like. The large model involved in the embodiments of the present disclosure may be a general large model, or may be an expert large model obtained after fine-tuning (Fine Tune) based on demand, and the embodiments of the present disclosure do not limit this. The large model can be used to process information of any modality such as images, text, video, and audio.

[0043] The object may be a user using the big model, and the first demand information may be a demand or request initially proposed by the user for a topic or field. The first demand information may be text language, or information in the form of voice, picture or other modalities. The first demand information may also be a demand or request proposed by the user for a topic or field different from the previous topic during the interaction with the big model.

[0044] The basic demand attributes of the object represent the key features or elements extracted from the first demand information, which are used to clarify the specific needs of the user. The basic demand attributes may include key elements such as subject, category, purpose, etc. The basic recommended resources may be resources directly related to the first demand information of the object retrieved from a database or knowledge base based on the basic demand attributes.

[0045] For example, the first demand information of the object is "travel in city A", and the basic demand attributes of the object may include "city A" and "travel". The basic recommended resources may be articles, web links, videos and other related resources about "travel in city A".

[0046] According to an embodiment of the present disclosure, an extended topic topology includes a plurality of extended topics representing potential demand attributes of an object, and association relationships between the plurality of extended topics, wherein the association relationships represent hierarchical relationships between the plurality of extended topics.

[0047] Potential demand attributes may be demand attributes that are associated with basic demand attributes but have not yet been determined by the object. Based on the potential demand attributes, multiple related extended topics may be constructed, and the multiple extended topics represent the extension direction of the potential demand attributes of the object.

[0048] The multiple extended topics may be extensions of the initial topic determined in the first demand information in different dimensions or directions. The multiple extended topics may include multiple subtopics of the initial topic and topics at the same level of the initial topic. For example, the initial topic may be expanded horizontally to obtain extended topics in the same category or at the same level as the initial topic; or the initial topic may be expanded vertically to obtain extended subtopics of the initial topic that further refines the initial topic.

[0049] For example, the initial topic is "Tourism in City A". The multiple extended topics in the extended topic topology may include multiple sub-topics of the initial topic, such as "Famous Attractions in City A", "Tourist Routes in City A", etc.; the multiple extended topics may also include other topics related to the initial topic, such as "Special Food in City A", "Cultural Customs in City A", "Geography and Climate in City A", etc.

[0050] In some embodiments, the feedback information may be text information obtained by refining and summarizing the retrieved basic resources using a large model, and matched with the first demand information of the object. The present disclosure is not limited thereto, and one or more matching resources selected from the basic resources may be directly pushed to the object as feedback information. Alternatively, a combination of text information and one or more matching resources may be pushed to the object as feedback information.

[0051] In an embodiment of the present disclosure, the first target request may be a further request made by the object based on the feedback information, for example, it may be a further question or clarification of the feedback information, or a request for an in-depth discussion of a certain topic.

[0052] Based on the second requirement information and the extended topic topology in the first target request, the subject intent of the object can be identified. The first target extended topic may include one or more extended topics in the extended topic topology, but the present disclosure is not limited thereto. The first target extended topic may also be an extended topic newly extended on the extended topic topology based on the identified subject intent.

[0053] According to the first target extension theme, a first extension recommendation resource matching the first target extension theme can be retrieved from a database or knowledge base. The first extension recommendation resource can be an article, video, picture, product, etc. The first extension recommendation resource is pushed to the user.

[0054] According to the embodiments of the present disclosure, by utilizing the relatively powerful semantic understanding ability of the large model to process the first demand information, basic recommended resources and extended topic topology are obtained, so that the basic demand attributes of the object can be identified, and basic recommended resources that meet the basic demand attributes of the user can be recommended. The extended topic topology can characterize the direction of the object's demand intention, facilitate in-depth understanding of the user's potential needs, guide the user's potential needs, and form a logical link of demand intention. Therefore, when the object initiates the first target request for feedback information, it can more accurately understand the user's demand attributes for the first target request by combining the second demand information determined by the user's first target request with the obtained extended topic topology, thereby more accurately and quickly recommending resources that meet the potential demand attributes to the user by extending the first target topic to the first extended recommended resource determined by the first target, thereby improving the user's resource acquisition efficiency.

[0055] According to an embodiment of the present disclosure, using a big model to process the first demand information to obtain basic recommended resources and extended subject topology related to the basic demand attributes of the first demand information includes: using a big model to process the first demand information and a preset prompt template to obtain basic recommended resources, extended subject topology and basic description text, the basic description text is used to describe the matching relationship between the resource content of the basic recommended resource and the basic demand attributes, and the feedback information includes the basic description text.

[0056] In an embodiment of the present disclosure, a preset prompt template includes instructions for prompting the large model to perform multiple tasks. For example, the instructions may include input parsing instructions, recommended resource generation instructions, extended topic topology construction instructions, description text generation instructions, and feedback information integration instructions. The input parsing instructions can be used to guide the large model to parse and understand the first demand information. The recommended resource generation instructions can be used to guide the large model to search, select and generate recommended resources. The extended topic topology construction instructions can be used to guide the large model to construct an extended topic topology based on the basic demand attributes of the object. The description text generation instructions can be used to extract and generate content for the recommended resources. The feedback information integration instructions can be used to integrate the obtained recommended resources and description texts and feed them back to the user.

[0057] In some embodiments, the basic recommendation resources may be resources in various forms such as text, web links, videos, audios, pictures, etc., and the basic description text may include a summary, generalization or explanation of the content of the text, web links, videos, audios, pictures, etc.

[0058] The description text also includes the matching relationship between the basic demand attributes. For example, the basic description text may also include relevant information explaining why the basic recommended resource is relevant to the user's demand and how the basic recommended resource satisfies the user's demand.

[0059] In some embodiments, the feedback information may be displayed in text form, or a combination of text and pictures, text and audio, text and video, and text and pictures. The present disclosure is not limited thereto, and may also be a combination of two or more other modes.

[0060] In some embodiments, the feedback information may also include instruction information for guiding the user to obtain the extended recommended resources determined based on the extended theme in the extended theme topology. For example, a continue recommendation button may be set to recommend the extended recommended resources to the user in response to the user's operation on the continue recommendation button.

[0061] According to the embodiments of the present disclosure, since the basic description text is generated by the large model based on the basic recommended resources and demand attributes, it can clearly show how the basic recommended resources meet the needs of users. Users can quickly understand the value of the basic recommended resources and the degree of matching with their own needs, thereby enhancing the readability and attractiveness of the recommendations.

[0062] Figure 3A The effect diagram of feedback information display according to the embodiment of the present disclosure is shown schematically. Figure 3A As shown, the feedback information includes text information, video combination, and instruction information for guiding the user to learn more recommended content.

[0063] In some embodiments, in response to the user's operation on the recommended resource, the complete content of the recommended resource is displayed to the user. For example, for a combination of a description text and a video resource, the user can play the video by clicking a video play button.

[0064] Figure 3B The figure shows the effect of operating the recommended resources according to the embodiment of the present disclosure.

[0065] like Figure 3B As shown, the first extended recommended resource may be a first video. After clicking on the first video, the first video may be played on one side of the interactive page, and feedback information may be displayed on the other side of the interactive page, and the user may further interact with the feedback information.

[0066] Figure 3C FIG. 2 is a diagram showing the effect of operating recommended resources according to another embodiment of the present disclosure.

[0067] like Figure 3C As shown, in another example, after clicking on the second video as the first extended recommended resource, it can be enlarged and played on the interactive page.

[0068] According to an embodiment of the present disclosure, the second demand information includes at least one of the following: operation input information of the object, and resource content information of the basic recommended resource.

[0069] In the embodiment of the present disclosure, the second requirement information may be new text information, audio information, image information, etc. input in response to the feedback information.

[0070] The second demand information is obtained by further questioning the video content, text content, and audio content in the feedback information. For example, a question button for further explanation or expansion of the resource content information can be set, and the user can click the question button to obtain more content related to the basic recommended resource.

[0071] According to the embodiments of the present disclosure, by providing users with a variety of ways to input demand information, the needs of users in different scenarios can be flexibly met, thereby improving user experience.

[0072] According to an embodiment of the present disclosure, performing topic intent recognition on the second requirement information and the extended topic topology of the interaction request to obtain the first target extended topic may include: using a large model to process the second requirement information, the extended topic topology and the feedback information to obtain a first sub-topology. The first sub-topology includes multiple first intermediate extended topics and association relationships between the multiple first intermediate extended topics; and determining the first target extended topic from the first sub-topology.

[0073] In the embodiments of the present disclosure, a large model is used to process user needs, expand the topic topology and feedback information, and the first sub-topology closely related to the user's current intention can be directly obtained.

[0074] In an embodiment of the present disclosure, a plurality of first intermediate extended topics in the first sub-topology may overlap with some extended topics in the extended topic topology. The present disclosure is not limited thereto, and the first intermediate extended topic in the first sub-topology may also be a newly extended topic.

[0075] The first intermediate extended topics in the first sub-topology may include refinements and extensions of the initial topic. For example, the topic of "famous attractions in city A" may be further refined into specific attractions, such as the travel guides and introductions of each sub-topic of scenic spots B, C, and D.

[0076] The first intermediate extended topics in the first sub-topology may also include newly extended topics based on the second demand information. For example, if the second demand information is "self-driving tour in city A", the first intermediate extended topics may include new topics such as "vehicle preparation", "self-driving equipment", and "self-driving route".

[0077] The association relationship between multiple first intermediate extended topics can be a hierarchical relationship, a parallel relationship, a similarity relationship, a causal relationship, etc.

[0078] In some embodiments, the large model may preferentially select the first intermediate extension topic that is most directly related to the second demand information of the object as the first target extension topic. The present disclosure is not limited thereto, and the large model may select the first target extension topic based on the weight or importance of different first intermediate extension topics in the first sub-topology. The large model may also select the first intermediate extension topic that is most closely related to the current context of the object as the first target extension topic.

[0079] According to an embodiment of the present disclosure, the first sub-topology is obtained by utilizing a large model to process the second demand information, the extended topic topology and the feedback information. Since the first sub-topology includes multiple first intermediate extended topics with associated relationships, it is possible to provide users with more potential topic content according to the user's intention context, thereby helping users obtain efficient and rich extended satisfaction and enhancing users' sense of gain.

[0080] According to an embodiment of the present disclosure, after obtaining the first sub-topology, the method may further include: using the large model to process the first sub-topology to obtain a first extended topic description text. The first extended topic description text is used to characterize a first intermediate extended topic that has an association relationship with the first target extended topic. The feedback information includes the first intermediate extended topic description text.

[0081] It can be understood that when the first extended recommended resource determined by the first target extended theme is recommended to the object, that is, when the first extended recommended resource is recommended to the object as new feedback information, the first intermediate extended theme associated with the first target extended theme will also be pushed to the object in the form of a description text as a part of the feedback information.

[0082] When processing the first sub-topology using the large model, the large model can be used to process multiple first intermediate extension topics respectively to obtain description texts of the multiple intermediate extension topics. The description texts may include a content summary of the first intermediate extension topic, a recommendation reason, and a relationship with the first target extension topic.

[0083] In some embodiments, when there are multiple first intermediate extended themes, the priority order of the multiple first intermediate extended themes can be determined according to the association relationship between the multiple first intermediate extended themes, and according to the priority order, a predetermined number of description texts of the first intermediate extended themes are selected as part of the feedback information and pushed to the user.

[0084] According to an embodiment of the present disclosure, the user can be prompted with relevant knowledge of the first intermediate extended topic through the first extended topic description text, so that the user can ask questions, make selections, and other operations on the first intermediate extended topic to generate a first target request, thereby improving the knowledge learning of the first intermediate extended topic and improving the user's information acquisition efficiency.

[0085] Figure 4A schematic diagram of determining a first sub-topology according to an embodiment of the present disclosure is shown.

[0086] like Figure 4 As shown, the extended topic topology 410, the feedback information 420 and the second demand information 430 are input into the large model for processing, and the first sub-topology 440 can be obtained. Among them, the first extended topic topology 411 is an example of the extended topic topology 410, and the first example sub-topology 441 and the second example sub-topology 442 are two examples of the first sub-topology. In some cases, the first sub-topology 440 can be a subset of the extended topic topology 410. For example, the first example sub-topology 441 is composed of extended topic 12, extended topic 121, extended topic 122 and extended topic 123, and the first example sub-topology 441 is a part of the first extended topic topology 411. In some cases, the first sub-topology 440 can also be obtained by extending the extended topic topology 410. For example, the second example sub-topology 442 is composed of extended topic 2, extended topic 21, extended topic 22 and extended topic 23, wherein extended topic 22 and extended topic 23 are obtained by further extending the first extended topic topology 411. After obtaining the first sub-topology 440 , the first target extended topic 450 that best matches the user's topic intention can be determined from the first sub-topology. For example, the first target extended topic 450 can be the “B City Dining Guide Topic” represented by the extended topic 22 .

[0087] It should be noted that the extended topics in the first extended topic topology 411, the first example sub-topology 441 and the second example sub-topology 442 are based on Figure 4 The nodes shown in the figure are used to represent each node, and each node may have an actual extended theme meaning, and the arrows between the nodes may represent the association relationship between different extended themes. For example, extended theme 1 represents "City A", extended theme 2 represents "City B", and extended theme 3 represents "City C". Extended theme 11 and extended theme 12 represent "catering" and "attractions" respectively, and extended theme 121, extended theme 122 and extended theme 123 represent the three attractions of "A Mountain Attractions", "City Park" and "Museum" respectively. Extended theme 111 and extended theme 112 represent "A City Barbecue" and "A City Hot Pot" respectively. It should be understood that extended theme 1, extended theme 2 and extended theme 3 have the same level, extended theme 11 and extended theme 12 have the same level, extended theme 111 and extended theme 112 have the same level, and extended theme 121, extended theme 122 and extended theme 123 have the same level.

[0088] The first extended resource determined based on the first target extended theme 450 may be, for example, a travel guide video or travel guide text related to the theme of dining guide for city B, so that the object can more conveniently extend the direction of the subject of interest, improve the flexibility of resource recommendation for the object, and improve the resource acquisition efficiency of the object.

[0089] According to an embodiment of the present disclosure, performing topic intent identification on the second requirement information and the extended topic topology of the interaction request to obtain the first target extended topic may also include: performing topic intent identification on the second requirement information and the feedback information to obtain a second intermediate extended topic. According to the second intermediate extended topic, a second sub-topology is determined from the extended topic topology. The second sub-topology includes a second intermediate extended topic and a second associated extended topic that has an associated relationship with the second intermediate extended topic. Based on the associated relationship of the second sub-topology, the first target extended topic is determined from the second sub-topology.

[0090] In an embodiment of the present disclosure, it is possible to first determine whether the requirement attribute of the second requirement information is a potential requirement attribute of the first requirement information of the object based on the first requirement information and the second requirement information. When it is determined that the requirement attribute of the second requirement information is a potential requirement attribute of the first requirement information of the object, the second intermediate extended topic can be quickly determined from the extended topic topology in combination with the upper text information provided by the feedback information.

[0091] According to the second intermediate extended topic, the second sub-topology is determined from the extended topic topology. This can be by starting from the position of the second intermediate extended topic in the extended topic topology, searching for extended topics that are directly or indirectly associated with the second intermediate extended topic, thereby obtaining multiple second associated extended topics, and obtaining the second sub-topology according to the second intermediate extended topic and the multiple second associated extended topics.

[0092] After obtaining the second sub-topology, multiple second related extended topics in the second sub-topology can be screened based on the hierarchical relationship, parallel relationship, similarity relationship, causal relationship, etc. between the second intermediate extended topics and the second related extended topics to determine the first target extended topic that matches the user's theme picture.

[0093] In some embodiments, the first target extended theme may be the second intermediate extended theme itself, or may be a second associated extended theme closely related thereto.

[0094] According to an embodiment of the present disclosure, by identifying the intent of the second demand information and the feedback information and determining the second intermediate extended topic and the second associated extended topic in the constructed extended topic topology, the search scope can be quickly narrowed and the efficiency of locating the first target extended topic can be improved.

[0095] According to an embodiment of the present disclosure, after obtaining the second sub-topology, the method may further include: using the large model to process the second sub-topology to obtain a second extended topic description text. The second extended topic description text is used to characterize a second associated extended topic that has an associated relationship with the first target extended topic. The feedback information includes the second extended topic description text.

[0096] When processing the second sub-topology using the large model, the large model can be used to process multiple second related extension topics respectively to obtain description texts of the multiple second related extension topics. The description texts may include content summary, recommendation reasons, and relationship with the first target extension topic of the second related extension topic.

[0097] When recommending the first extended recommended resource determined by the first target extended theme to the object, that is, when recommending the first extended recommended resource to the object as new feedback information, the second associated extended theme associated with the first target extended theme can also be pushed to the object in the form of a description text as part of the feedback information.

[0098] In some embodiments, when there are multiple second associated extended topics, the priority order of the multiple second associated extended topics can be determined according to the association relationship between the multiple second associated extended topics. According to the priority order, a predetermined number of description texts of the first intermediate extended topics are selected as part of the feedback information and pushed to the user.

[0099] According to an embodiment of the present disclosure, the second extended topic description text can prompt the user with relevant knowledge of the second associated extended topic, so that the user can ask questions, make selections, and other operations on the second associated extended topic to generate a first target request, thereby improving the learning of the second associated extended topic and improving the user's information acquisition efficiency.

[0100] Figure 5 A schematic diagram of determining a second sub-topology according to an embodiment of the present disclosure is shown.

[0101] like Figure 5As shown, based on the feedback information 510 and the second demand information 520, the second intermediate extended theme 530 is identified. According to the second extended theme 30, a query is performed from the extended theme topology 540 to obtain a second sub-topology 550. Among them, the extended theme 531 is an example of the second intermediate extended theme, the second extended theme topology 541 is an example of the extended theme topology 540, and the third example sub-topology 551 is an example of the second sub-topology 550. According to the extended theme 531, the extended theme 12, the extended theme 121, the extended theme 122 and the extended theme 123 that are associated with the extended theme 531 can be determined from the second extended theme topology 541, and the third example sub-topology 551 composed of the extended theme 121 and the extended theme 122 can be determined. After obtaining the second sub-topology 560, the first target extended theme 560 that best matches the user's theme intention can be determined from the second sub-topology. For example, the first target extended theme 560 can be "City Park" represented by the extended theme 122. The first extended resource determined based on the first target extended theme 560 may be, for example, a travel guide related to a city park, so that the object can more conveniently extend the direction of the subject of interest, improve the flexibility of recommending resources for the object, and improve the resource acquisition efficiency of the object.

[0102] According to an embodiment of the present disclosure, there is a candidate association relationship between the first target extension topic and multiple second candidate extension topics. After obtaining the first target extension topic, it can also include: determining structured feedback information based on the candidate association relationship between the first target extension topic and multiple second candidate extension topics. The structured feedback information includes multiple second candidate extension topics in a structured format, and the structured format characterizes the candidate association relationship. Pushing structured feedback information to the object. In response to the second target request, determining the second target extension topic from the multiple second candidate extension topics. The second target interaction request is determined by the object performing an interactive operation on the second candidate extension topic in the structured demand feedback information. Pushing the second extension recommendation resource determined based on the second target extension topic to the object.

[0103] In some embodiments, the plurality of second candidate extended topics may be a plurality of extended topics determined from an extended topic topology and having an association relationship with the first target extended topic.

[0104] In some embodiments, when the large model is used to process the second demand information, the extended topic topology and the feedback information to obtain the first sub-topology, the plurality of second candidate extended topics may be first intermediate extended topics determined from the first sub-topology and associated with the first target extended topic.

[0105] In some embodiments, when subject intent identification is performed on the second demand information and the feedback information to obtain a second intermediate extended topic and a second sub-topology is determined from the extended topic topology, the plurality of second candidate extended topics may be second associated extended topics determined from the second sub-topology and having an associated relationship with the first target extended topic.

[0106] In some embodiments, the structured feedback information may include a first extension recommendation resource determined based on the first target extension theme and a description text determined based on multiple second candidate extension themes. The first extension recommendation resource serves as the first feedback information in response to the user's intention, and the description texts determined by the multiple second candidate extension themes serve as multiple second feedback information of the user's potential theme intention. The first feedback information and the multiple second feedback information are combined into structured feedback information and fed back to the user.

[0107] In some embodiments, the first feedback information has a structured format, for example, displayed in the interactive interface in the format of serial number-title-content. The plurality of second feedback information can be arranged in sequence after the first feedback information in a preferred order. The present disclosure is not limited thereto, as long as the feedback information is output in a structured form.

[0108] In some embodiments, the second feedback information is structured in a form that can interact with the user. For example, in response to the user clicking the second feedback information by a mouse, the second extension recommended resource determined by the corresponding second candidate extension theme can be recommended to the user. The present disclosure is not limited to this, and the second extension recommended resource determined by the corresponding second candidate extension theme can also be recommended to the user in response to the second target request input by the user's second feedback information.

[0109] According to the disclosed embodiment, structured feedback information is determined based on the first target extension topic and multiple second candidate extension topics. Since the structured feedback information not only includes the first extension recommendation resource that matches the user's topic intention, but also includes the second candidate extension topic that matches the user's potential intention, while meeting the user's intention needs, it also guides the user to explore new topics that may be of interest, promotes user exploration and discovery, and enhances the user's interaction with the big model, providing users with a more personalized and intelligent information service experience.

[0110] Figure 6 The flowchart of a resource method based on a large model according to another embodiment of the present disclosure is schematically shown.

[0111] like Figure 6As shown, the first demand information 610 is input into the large model M620 for processing, and the basic recommended resources 630 and the extended topic topology 630 can be obtained. Feedback information 650 can be recommended to the user based on the basic recommended resources 630. The topic intention is identified based on the extended topic topology 640, the feedback information 650 and the second demand information 660 to obtain multiple candidate extended topics 670. A target extended topic 680 is determined from the multiple candidate extended topics 670, and structured feedback information 690 is obtained based on the target extended topic 680 and the multiple candidate extended topics 670. In response to the second target request of the object for the multiple candidate extended topics in the structured feedback information, a new target extended topic is determined from the multiple candidate extended topics, and the new structured feedback information is recommended to the user.

[0112] Figure 7 The block diagram of a resource recommendation device based on a large model according to an embodiment of the present disclosure is schematically shown.

[0113] like Figure 7 As shown, the large model-based resource recommendation device 700 includes a processing module 710 , a first pushing module 720 , an identification module 730 and a second pushing module 740 .

[0114] Processing module 710 is used to use the large model to process the first demand information of the object, and obtain basic recommended resources and extended topic topology related to the basic demand attributes of the object. The extended topic topology includes multiple extended topics that represent the potential demand attributes of the object, and the association relationship between the multiple extended topics. The association relationship represents the hierarchical relationship between the multiple extended topics.

[0115] The first push module 720 is used to push feedback information determined based on the basic recommendation resources.

[0116] The identification module 730 is used to respond to the first target request of the object for feedback information, perform topic intent identification on the second requirement information and the extended topic topology of the first target request, and obtain the first target extended topic.

[0117] The second push module 740 is configured to push the first extended recommended resource determined based on the first target extended theme to the object.

[0118] According to an embodiment of the present disclosure, the identification module 730 includes a first processing submodule and a first determination submodule.

[0119] The first processing submodule is used to use the large model to process the second demand information, the extended topic topology and the feedback information to obtain the first sub-topology, which includes multiple first intermediate extended topics and the association relationship between the multiple first intermediate extended topics.

[0120] The first determination submodule is used to determine a first target extended topic from a first sub-topology.

[0121] According to an embodiment of the present disclosure, the device also includes a second processing submodule.

[0122] The second processing submodule is used to process the first sub-topology using the large model to obtain a first extended topic description text, wherein the first extended topic description text is used to characterize the first extended topic associated with the first target extended topic; wherein the feedback information includes the first extended topic description text.

[0123] According to an embodiment of the present disclosure, the identification module 730 includes a first identification submodule, a second determination submodule, and a third determination submodule.

[0124] The first identification submodule is used to perform topic intention identification on the second demand information and the feedback information to obtain a second intermediate extended topic.

[0125] The second determining submodule is used to determine a second sub-topology from the extended topic topology according to the second intermediate extended topic, wherein the second sub-topology includes the second intermediate extended topic and a second associated extended topic associated with the second intermediate extended topic.

[0126] The third determination submodule is used to determine the first target extended topic from the second sub-topology based on the association relationship of the second sub-topology.

[0127] According to an embodiment of the present disclosure, the device also includes a third processing submodule.

[0128] The third processing submodule is used to process the second sub-topology using the large model to obtain a second extended topic description text, wherein the second extended topic description text is used to characterize a second associated extended topic that has an associated relationship with the first target extended topic. The feedback information includes the second extended topic description text.

[0129] According to an embodiment of the present disclosure, a candidate association relationship exists between the first target extended theme and a plurality of second candidate extended themes; the device further includes a first determining module, a third pushing module, a second determining module and a fourth pushing module.

[0130] The first determination module is used to determine structured feedback information according to the candidate association relationship between the first target extended topic and multiple second candidate extended topics, wherein the structured feedback information includes multiple second candidate extended topics in a structured format, and the structured format represents the candidate association relationship.

[0131] The third push module is used to push structured feedback information to the object.

[0132] The second determination module is used to determine a second target extension topic from a plurality of second candidate extension topics in response to a second target request, wherein the second target interaction request is determined by the object performing an interaction operation on the second candidate extension topic in the structured demand feedback information.

[0133] The fourth pushing module is used to push the second extended recommended resource determined based on the second target extended theme to the object.

[0134] According to an embodiment of the present disclosure, the second demand information includes at least one of the following: operation input information of the object, and resource content information of the basic recommended resource.

[0135] According to an embodiment of the present disclosure, the processing module 610 includes a fourth processing submodule.

[0136] The fourth processing submodule is used to use the large model to process the first demand information and the preset prompt template to obtain basic recommended resources, extended topic topology and basic description text. The basic description text is used to describe the matching relationship between the resource content of the basic recommended resources and the basic demand attributes. The feedback information includes the basic description text.

[0137] Figure 8 The structural block diagram of an artificial intelligence agent according to an embodiment of the present disclosure is schematically shown.

[0138] In the embodiments of the present disclosure, Figure 8 As shown, the AI ​​agent 800 may include an input module 810 , a processing module 820 , and an output module 830 .

[0139] An input module 810, for receiving input information;

[0140] The processing module 820 is used to determine the target task based on the input information received by the input module, determine the big model based on the target task, and execute the resource recommendation method based on the big model provided according to the embodiment of the present disclosure by calling the big model.

[0141] The output module 830 is used to output the output information obtained by the processing module.

[0142] According to an embodiment of the present disclosure, the input module 810 is responsible for receiving or sensing information such as queries, requests, instructions, signals or data from the outside world (e.g., users or external environment), and converting them into a format that can be understood and processed by the AI ​​agent 800. The input module 810 is the primary link for the AI ​​agent 800 to interact with the outside world, which enables the AI ​​agent 800 to efficiently and accurately obtain the necessary "sensory" information from the outside world and respond to this information.

[0143] In an example, the input module 810 may input the first requirement information, the second requirement information, etc. described above.

[0144] In the example, the processing module 820 is the core support for the AI ​​agent 800 to process complex tasks. The processing module 820 can execute the resource recommendation method based on the large model described above.

[0145] In the example, the performance of the processing module 820 may be closely related to the large model on which the AI ​​agent 800 is based. In order to give full play to the capabilities of the large model, the internal structure of the processing module 820 may be designed to be highly configurable and scalable to cope with various types of tasks and requirements in real scenarios.

[0146] In the example, after the AI ​​agent 800 obtains the first demand information, the processing module 820 can use the large model to process the first demand information of the object, obtain feedback information, and pass it to the output module 830.

[0147] It is understandable that although the large language model has excellent language understanding and generation capabilities, it is just like a human being, without the help of any tools, and the tasks it can solve are very limited. When AI agent 800 is given the ability to call tools, it can perform tasks such as using a calculator to complete mathematical operations, using Python to complete data analysis, and using a search engine to complete weather forecasts.

[0148] In an example, the output module 830 may output the feedback information described above.

[0149] The AI ​​agent 800 according to the embodiment of the present disclosure can simply and effectively improve the level of intelligence, and improve flexibility and versatility.

[0150] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0151] According to an embodiment of the present disclosure, an electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above method.

[0152] According to an embodiment of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable a computer to execute the above method.

[0153] According to an embodiment of the present disclosure, a computer program product includes a computer program, and the computer program implements the above method when executed by a processor.

[0154] Fig. 9 A schematic block diagram of an example electronic device 900 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0155] like Fig. 9 As shown, the device 900 includes a computing unit 901, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 902 or a computer program loaded from a storage unit 908 into a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the device 900 can also be stored. The computing unit 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0156] A number of components in the device 900 are connected to the I / O interface 905, including: an input unit 906, such as a keyboard, a mouse, etc.; an output unit 907, such as various types of displays, speakers, etc.; a storage unit 908, such as a disk, an optical disk, etc.; and a communication unit 909, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 909 allows the device 900 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0157] The computing unit 901 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 901 performs the various methods and processes described above, such as a resource recommendation method based on a large model. For example, in some embodiments, the resource recommendation method based on a large model may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 908. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 900 via the ROM 902 and / or the communication unit 909. When the computer program is loaded into the RAM 903 and executed by the computing unit 901, one or more steps of the resource recommendation method based on the large model described above may be performed. Alternatively, in other embodiments, the computing unit 901 may be configured to execute the large model-based resource recommendation method in any other appropriate manner (for example, by means of firmware).

[0158] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0159] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0160] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0161] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0162] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0163] A computer system may include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises through computer programs running on respective computers and having a client-server relationship to each other. The server may be a cloud server, a server in a distributed system, or a server combined with a blockchain.

[0164] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0165] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A resource recommendation method based on a large model, comprising: Processing the first demand information of the object using the large model to obtain basic recommended resources and an extended topic topology related to the basic demand attributes of the object, wherein the extended topic topology includes a plurality of extended topics representing potential demand attributes of the object, and association relationships between the plurality of extended topics, wherein the association relationships represent hierarchical relationships between the plurality of extended topics; Pushing feedback information determined based on the basic recommended resources; In response to the first target request of the object for the feedback information, performing topic intent recognition on the second requirement information of the first target request and the extended topic topology to obtain a first target extended topic; as well as A first extended recommendation resource determined based on the first target extended theme is pushed to the object.

2. The method according to claim 1, wherein: The performing topic intention identification on the second requirement information of the interaction request and the extended topic topology to obtain the first target extended topic includes: Processing the second demand information, the extended topic topology and the feedback information by using the large model to obtain a first sub-topology, wherein the first sub-topology includes a plurality of first intermediate extended topics and association relationships between the plurality of first intermediate extended topics; and The first target extended topic is determined from the first sub-topology.

3. The method according to claim 2, further comprising: Processing the first sub-topology by using the large model to obtain a first extended topic description text, where the first extended topic description text is used to represent a first intermediate extended topic associated with the first target extended topic; The feedback information includes the first intermediate extended topic description text.

4. The method according to claim 1, wherein: The performing topic intention identification on the second requirement information of the interaction request and the extended topic topology to obtain the first target extended topic includes: Performing topic intention recognition on the second demand information and the feedback information to obtain a second intermediate extended topic; determining a second sub-topology from the extended topic topology according to the second intermediate extended topic, the second sub-topology including the second intermediate extended topic and a second associated extended topic having an associated relationship with the second intermediate extended topic; and Based on the association relationship of the second sub-topology, the first target extended topic is determined from the second sub-topology.

5. The method according to claim 4, wherein: Also includes: Processing the second sub-topology by using the large model to obtain a second extended topic description text, where the second extended topic description text is used to represent a second associated extended topic that has an associated relationship with the first target extended topic; The feedback information includes the second extended topic description text.

6. The method according to claim 1, wherein: There is a candidate association relationship between the first target extended topic and a plurality of second candidate extended topics; Wherein, the method further comprises: Determining structured feedback information according to the candidate association relationships between the first target extended topic and a plurality of second candidate extended topics, wherein the structured feedback information includes a plurality of the second candidate extended topics in a structured format, and the structured format represents the candidate association relationships; Pushing the structured feedback information to the object; In response to a second target request, determining a second target extension topic from a plurality of second candidate extension topics, wherein the second target interaction request is determined by the object performing an interaction operation on the second candidate extension topic in the structured requirement feedback information; and A second extended recommendation resource determined based on the second target extended theme is pushed to the object.

7. The method according to any one of claims 1 to 5, wherein: The second requirement information includes at least one of the following: The operation input information of the object and the resource content information of the basic recommended resource.

8. The method according to claim 1, wherein: The method of using the large model to process the first demand information to obtain basic recommended resources and extended topic topology related to the basic demand attributes of the first demand information includes: The first demand information and the preset prompt template are processed using a large model to obtain the basic recommended resources, the extended topic topology and the basic description text. The basic description text is used to describe the matching relationship between the resource content of the basic recommended resources and the basic demand attributes. The feedback information includes the basic description text.

9. A resource recommendation device based on a large model, comprising: A processing module, configured to process the first demand information of the object using the large model, and obtain basic recommended resources and an extended topic topology related to the basic demand attributes of the object, wherein the extended topic topology includes a plurality of extended topics representing potential demand attributes of the object, and an association relationship between the plurality of extended topics, wherein the association relationship represents a hierarchical relationship between the plurality of extended topics; A first push module, used for pushing feedback information determined based on the basic recommendation resources; an identification module, configured to, in response to a first target request of the object for the feedback information, perform topic intent identification on the second requirement information of the first target request and the extended topic topology to obtain a first target extended topic; as well as The second pushing module is configured to push a first extended recommended resource determined based on the first target extended theme to the object.

10. The device according to claim 9, wherein: The first identification module comprises: a first processing submodule, configured to use the large model to process the second demand information, the extended topic topology, and the feedback information to obtain a first sub-topology, wherein the first sub-topology includes a plurality of first intermediate extended topics and association relationships between the plurality of first intermediate extended topics; and The first determining submodule is configured to determine the first target extended topic from the first sub-topology.

11. The method according to claim 10, further comprising: A second processing submodule is used to process the first sub-topology using the large model to obtain a first extended topic description text, where the first extended topic description text is used to represent a first intermediate extended topic associated with the first target extended topic; The feedback information includes the first intermediate extended topic description text.

12. The method according to claim 9, wherein: The first identification module comprises: A first identification submodule is used to perform topic intention identification on the second demand information and the feedback information to obtain a second intermediate extended topic; a second determining submodule, configured to determine a second sub-topology from the extended topic topology according to the second intermediate extended topic, wherein the second sub-topology includes the second intermediate extended topic and a second associated extended topic having an associated relationship with the second intermediate extended topic; and The third determining submodule is configured to determine the first target extended topic from the second sub-topology based on the association relationship of the second sub-topology.

13. The device according to claim 12, wherein: Also includes: A third processing submodule is used to process the second sub-topology using the large model to obtain a second extended topic description text, where the second extended topic description text is used to represent a second associated extended topic that has an associated relationship with the first target extended topic; The feedback information includes the second extended topic description text.

14. The device according to claim 9, wherein: There is a candidate association relationship between the first target extended topic and a plurality of second candidate extended topics; Wherein, the device further comprises: a first determining module, configured to determine structured feedback information according to candidate association relationships between the first target extended topic and a plurality of second candidate extended topics, wherein the structured feedback information includes a plurality of second candidate extended topics in a structured format, wherein the structured format represents the candidate association relationships; A third pushing module, configured to push the structured feedback information to the object; a second determination module, configured to determine a second target extension topic from a plurality of second candidate extension topics in response to a second target request, wherein the second target interaction request is determined by the object performing an interaction operation on the second candidate extension topic in the structured requirement feedback information; and The fourth pushing module is configured to push a second extended recommended resource determined based on the second target extended theme to the object.

15. The device according to any one of claims 9 to 14, wherein: The second requirement information includes at least one of the following: The operation input information of the object and the resource content information of the basic recommended resource.

16. The device according to claim 9, wherein: The processing module comprises: The fourth processing submodule is used to use the large model to process the first demand information and the preset prompt template to obtain the basic recommended resources, the extended topic topology and the basic description text. The basic description text is used to describe the matching relationship between the resource content of the basic recommended resources and the basic demand attributes. The feedback information includes the basic description text.

17. An artificial intelligence agent, comprising: An input module, used for receiving input information; a processing module, configured to determine a target task based on the input information received by the input module, determine a large language model based on the target task, and execute the method according to any one of claims 1 to 8 by calling the large language model; An output module is used to output the output information obtained by the processing module.

18. An electronic device, comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 8.

19. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 8.

20. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 8.

Images