Intelligent agent-based recommendation method and device, electronic equipment and intelligent agent

Abstract

The invention provides an agent-based recommendation method, and relates to the technical field of artificial intelligence, in particular to the fields of large language models, AI agents, human-computer interaction and the like. According to the specific implementation scheme, to-be-processed dialogue data from a target object is input into a first large language model of an intelligent agent, and a basic requirement of the target object is obtained; inputting the historical dialogue data and the basic demand of the target object in the preset time period into a second large language model of the intelligent agent to obtain a target demand of the target object; and matching the target demand with multiple pieces of material information in a material library to obtain a target material matched with the target object, and displaying the target material to the target object. The invention further provides an agent-based recommendation device, electronic equipment, a computer readable storage medium, a computer program product and an artificial intelligence agent.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence (AI) technology, and in particular to the fields of large language model (LLM), AI agent, human-computer interaction, etc. More specifically, the present disclosure provides an agent-based recommendation method, device, electronic device, computer-readable storage medium, and computer program product, as well as an artificial intelligence agent. Background Art

[0002] In recent years, recommendation systems have achieved great success in the Internet industry. Recommendation systems learn user preferences based on user personal information and click-through behavior, and then recommend suitable objects to users based on their preferences. Summary of the invention

[0003] The present disclosure provides an agent-based recommendation method, device, electronic device, computer-readable storage medium, and computer program product, as well as an artificial intelligence agent.

[0004] According to one aspect of the present disclosure, there is provided an agent-based recommendation method, comprising: inputting to-be-processed conversation data from a target object into a first language model of the agent to obtain basic demands of the target object; inputting historical conversation data and basic demands of the target object in a preset time period into a second language model of the agent to obtain target demands of the target object; matching the target demands with multiple material information in a material library to obtain target materials that match the target object, and displaying the target materials to the target object.

[0005] According to another aspect of the present disclosure, there is provided an agent-based recommendation device, comprising: a basic demand determination module, used to input the to-be-processed conversation data from the target object into the first largest language model of the agent, to obtain the basic demand of the target object; a target demand determination module, used to input the historical conversation data and basic demand of the target object in a preset time period into the second largest language model of the agent, to obtain the target demand of the target object; a material matching module, used to match the target demand with multiple material information in a material library, to obtain the target material matching the target object, and to display the target material to the target object.

[0006] According to another aspect of the present disclosure, an artificial intelligence agent is provided, which is configured to execute the method provided according to the present disclosure.

[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 provided according to the present disclosure.

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

[0009] According to another aspect of the present disclosure, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the method provided according to the present disclosure is implemented.

[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 is a schematic diagram of an exemplary system architecture to which an agent-based recommendation method and apparatus can be applied according to an embodiment of the present disclosure;

[0013] Figure 2 is a flow chart of an agent-based recommendation method according to an embodiment of the present disclosure;

[0014] Figure 3 is a schematic diagram of an agent-based recommendation method according to an embodiment of the present disclosure;

[0015] Figure 4 is a schematic diagram of an interaction interface between a target object and an agent according to an embodiment of the present disclosure;

[0016] Figure 5 is a schematic diagram of the structure of a demand tree according to an embodiment of the present disclosure;

[0017] Figure 6 is a flowchart of location information acquisition according to an embodiment of the present disclosure;

[0018] Figure 7 is a flow chart of material library formation according to one embodiment of the present disclosure;

[0019] Figure 8 is a block diagram of an agent-based recommendation device according to an embodiment of the present disclosure;

[0020] Fig. 9 is a block diagram of an artificial intelligence agent according to an embodiment of the present disclosure; and

[0021] Fig.10 is a block diagram of an electronic device to which an agent-based recommendation method can be applied according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0022] 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.

[0023] At present, personalized product recommendation methods usually include collaborative filtering, content-based recommendation and hybrid recommendation, etc. These methods mainly rely on information flow recommendation and require in-depth mining based on users' historical behavior data to provide personalized recommendations for users. However, these methods have the following disadvantages: first, they rely on a large amount of historical data and are not effective for new users or users with less data; second, they cannot adapt to users' immediate needs and preference changes in real time.

[0024] In view of this, the present disclosure provides an agent-based recommendation method. When a user has a real-time conversation with the agent, the agent can interpret and utilize the current interaction context in real time, identify the user's deep needs, and then accurately match the user's deep needs with the materials in the material library.

[0025] It can be understood that an agent can refer to an agent that can perceive the environment and take actions to achieve specific goals. It can be software, hardware or a system with autonomy, adaptability and interaction capabilities. The core of the agent is the ability to learn autonomously and continuously evolve to better complete tasks and adapt to complex environments. The working principle of the agent can include three processes: perception, decision-making and execution. First, the agent collects information in the environment through the perception module, which can be static or dynamic. Then, the agent can make decisions and reasoning based on the collected information, combined with its own algorithms and learning mechanisms, to determine the appropriate behavior strategy. Then, the agent can convert the decision results into actual actions through the action module to change the state of the environment, and adjust its behavior strategy in real time to adapt to environmental changes.

[0026] The agent and the large language model can be used together. The large language model can be used as a part of the agent to process the data perceived by the agent and help the agent make more accurate decisions. For example, in an autonomous driving scenario, the agent can perceive the road environment, identify traffic signals and obstacle information, and the large language model can use this information for deep learning and detection, thereby providing the agent with more accurate and reliable decision support.

[0027] The interaction between agents and large language models in the field of artificial intelligence can promote the development and progress of intelligent systems. The autonomy and interactivity of agents enable them to flexibly respond to various challenges in various complex environments. The information processing and knowledge output capabilities of large language models provide agents with richer and more accurate data support. The combination of agents and large language models makes artificial intelligence systems more intelligent and autonomous.

[0028] Compared with the current personalized product recommendation method, the agent-based recommendation method provided by the present invention utilizes the agent to make recommendations during the interaction process, which can not only adapt to the changes in user needs in real time, but also provide more personalized and accurate product recommendations during the conversation process, thereby improving the recommendation effect and user experience.

[0029] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision, disclosure and application of user personal information involved comply with the provisions of relevant laws and regulations, take necessary confidentiality measures, and do not violate public order and good morals.

[0030] In the technical solution of the present disclosure, the user's authorization or consent is obtained before obtaining or collecting the user's personal information.

[0031] Figure 1 is a schematic diagram of an exemplary system architecture to which an agent-based recommendation method and apparatus can be applied according to an embodiment of the present disclosure.

[0032] 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 agent-based recommendation method and apparatus can be applied may include a terminal device, but the terminal device may implement the agent-based recommendation method and apparatus provided by the embodiments of the present disclosure without interacting with the server.

[0033] like Figure 1As shown, the system architecture 100 according to this embodiment may include clients 101, 102 and a server 103. The clients 101, 102 and the server 103 may communicate so as to call the large language model of the agent based on the user's operation on the interactive interface in the clients 101, 102.

[0034] Users can perform operations such as selection and editing on the interactive interface of the client 101 and 102 to generate dialogue data to be processed for interaction with the server 103, so as to call the intelligent agent of the server 103 to generate corresponding target materials and display them on the interactive interface of the client 101 and 102.

[0035] Various communication client applications may be installed on the clients 101 and 102, such as smart assistant applications, knowledge reading applications, web browser applications, search applications, instant messaging tools, email clients and / or social platform software (for example only).

[0036] The clients 101 and 102 may be configured with various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0037] The server 103 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 interactive interface of the client 101, 102. The background management server may call the agent to recommend materials for the dialogue data to be processed received from the user, and feed back the recommended target materials to the client 101, 102, and display them through the interactive interface.

[0038] The storage module in the server 103 can be used to store the user's historical conversation data, such as historical interaction questions and corresponding historical answer information.

[0039] It should be noted that the agent-based recommendation method provided in the embodiment of the present disclosure can generally be executed by the server 103. Accordingly, the agent-based recommendation device provided in the embodiment of the present disclosure can also be set in the server 103. The agent-based recommendation method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 103 and can communicate with the client 101, 102 and / or the server 103. Accordingly, the agent-based recommendation device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 103 and can communicate with the client 101, 102 and / or the server 103.

[0040] Alternatively, the agent-based recommendation method provided in the embodiment of the present disclosure may also be generally executed by the client 101, 102. Accordingly, the agent-based recommendation device provided in the embodiment of the present disclosure may generally be set in the client 101, 102.

[0041] For example, the intelligent agent can be loaded on the client first, and the user can input instructions to interact with the intelligent agent based on the interactive interface, so as to call the intelligent agent to recommend materials, and feed back the recommended target materials to the client and display them through the interactive interface.

[0042] It should be understood that Figure 1 The number of devices configured on the client and server is only illustrative. Any number of devices can be configured on the client and server as required.

[0043] Figure 2 is a flowchart of an agent-based recommendation method according to an embodiment of the present disclosure.

[0044] like Figure 2 As shown, the agent-based recommendation method 200 may include operations S210 to S230, and the method may be executed by the server.

[0045] In operation S210, the dialogue data to be processed from the target object is input into the first language model of the agent to obtain the basic needs of the target object.

[0046] For example, the target object can be a user of the client, and the agent can be set on the server. When the user has a real-time conversation with the agent on the server, the user can select, edit, and perform other operations on the client's interactive interface to generate pending conversation data for interaction with the server. The pending conversation data is input into the first language model of the agent to identify the user's basic needs.

[0047] For example, the dialogue data to be processed may be interactive questions issued to the agent, or reply information given to the agent.

[0048] For example, an agent can be a merchant agent. A merchant agent is a "24-hour professional and trustworthy" digital user operation employee generated based on the personalized demands of advertisers. This digital employee can achieve precise marketing recommendations for services and products, effectively reduce lead costs, and help merchants reduce costs and increase efficiency.

[0049] In operation S220, historical conversation data and basic demands of the target object in a preset time period are input into the second language model of the agent to obtain the target demands of the target object.

[0050] For example, the historical conversation data of the user in a preset time period can be obtained, and the historical conversation data and the basic needs obtained above can be input into the second largest language model of the intelligent agent to identify the user's target needs.

[0051] For example, the target demand can be the same as or different from the basic demand. The target demand can be the same as the basic demand, or it can be a deeper demand with finer granularity than the basic demand, or it can be other demand different from the basic demand to correct the basic demand.

[0052] In operation S230, the target demand is matched with a plurality of material information in the material library to obtain a target material matching the target object, and the target material is displayed to the target object.

[0053] For example, the materials in the material library can be physical products, or virtual products, organizations, etc. Specifically, the materials in the material library can be different types of commodities such as clothing, food, cars, digital products, etc., or different types of multimedia content such as books, music, videos, etc. This embodiment can be used to recommend only one type of object, such as only clothing, or only music, etc., and can also be used to recommend multiple types of objects at the same time, such as clothing and cars, music and food, or cars, clothing and videos, etc.

[0054] For example, the display method of the target material includes but is not limited to visual display, and can also be displayed to the user through voice broadcast. The display method of the target material can be flexibly configured according to the content of the target material.

[0055] Through the embodiments of the present disclosure, when a user and an agent have a real-time conversation, the agent can interpret and use the current interaction context in real time to identify the user's deep needs, and then accurately match the user's deep needs with the materials in the material library. The present disclosure uses the agent to make recommendations during the interaction process, which can not only adapt to the changes in user needs in real time, but also provide more personalized and accurate product recommendations during the conversation process, improving the recommendation effect and user experience.

[0056] The various large language models in this embodiment are all deep learning models trained based on massive text data. They can not only generate natural language text, but also deeply understand the meaning of the text, so as to be used to process various natural language tasks, such as text summary generation, question answering, translation, etc.

[0057] According to an embodiment of the present disclosure, the basic requirements include business scope and conversation intention.

[0058] According to an embodiment of the present disclosure, the material library is obtained by processing a plurality of materials to be recommended using an intelligent agent. The material library can be set within the intelligent agent.

[0059] Figure 3 is a schematic diagram of an agent-based recommendation method according to an embodiment of the present disclosure.

[0060] like Figure 3 As shown, the agent-based recommendation method 300 of this embodiment works as follows: first, the target object inputs the conversation data 301 to be processed into the first language model 302a of the agent 302, and outputs the basic demand 303 of the target object. Secondly, the historical conversation data 304 of the target object in a preset period is obtained, and the historical conversation data 304 and the basic demand 303 outputted above are input into the second language model 302b of the agent 302, and the target demand 305 of the target object is output. Then, the target demand 305 is matched with the multiple material information in the material library 302c of the agent 302, and the target material 306 matching the target object is obtained and displayed to the target object.

[0061] According to an embodiment of the present disclosure, the agent-based recommendation method further includes: the agent's generated object conducts at least one round of dialogue with the target object to obtain dialogue data; and extracts historical dialogue data of the target object in a preset time period from the dialogue data.

[0062] The historical conversation data, as the context information of the user's current conversation data to be processed, may include the user's language, emotions, behaviors, etc. Through the embodiments of the present disclosure, the intelligent agent has a strong context perception capability, and can fine-tune the user's basic needs and identify the user's deep needs based on the historical conversation data between the intelligent agent and the user.

[0063] For example, through the first language model, it is identified that the business scope of the user's basic needs is "study abroad business". Through the second language model, based on the historical conversation data between the merchant agent and the user, it is identified that the business scope of the user's target needs is "study abroad business in the UK", which deeply depicts the user's consulting business scope. It can be seen that the second language model is supervised fine-tuning (SFT) of the first language model based on the historical conversation data between the merchant agent and the user, and identifies the user's deep needs.

[0064] Figure 4 It is a schematic diagram of an interaction interface between a target object and an intelligent agent according to an embodiment of the present disclosure.

[0065] like Figure 4As shown, the interactive objects of the interactive interface 400 include the target object and the generated object of the agent. When the target object and the generated object of the agent have a round of dialogue, the target object can input the dialogue data 401 to be processed on the interactive interface 400; the generated object of the agent can process the dialogue data 401 to be processed, obtain the processing result, that is, the recommended target material 402, and feed the target material back to the target object, and display it through the interactive interface 400. After a round of dialogue, the dialogue data obtained includes the dialogue data 401 to be processed and the recommended target material 402, and the dialogue data can be stored in the agent.

[0066] Based on this, at the next moment when the generated object of the intelligent agent has at least one round of dialogue with the target object, in response to the to-be-processed dialogue data 403 input by the target object, the historical dialogue data of the target object in a preset time period can be extracted from the dialogue data stored by the intelligent agent.

[0067] According to an embodiment of the present disclosure, there are multiple basic requirements; the agent-based recommendation method also includes: for each basic requirement, according to the entity attributes of the agent, configuring a requirement tree corresponding to the basic requirement in the first language model; wherein the requirement tree includes multiple nodes and at least one edge between the nodes, each node represents a sub-requirement of the basic requirement, and each edge represents the relationship between the sub-requirements.

[0068] Since the agent may be a merchant agent, the entity attributes of the agent may be merchant attributes such as the merchant industry, sales product type, etc. The demand tree may be configured in the first language model according to these merchant attributes.

[0069] According to an embodiment of the present disclosure, the dialogue data to be processed from the target object is input into the first largest language model of the intelligent agent to obtain the basic requirements of the target object, including performing the following operations for each basic requirement: matching the dialogue data to be processed with multiple nodes of the requirement tree to obtain at least one matching node; and determining the sub-requirement represented by at least one node as the basic requirement of the target object.

[0070] Figure 5 is a schematic diagram of the structure of a requirement tree according to an embodiment of the present disclosure.

[0071] like Figure 5 As shown, taking the business scope as an example, a business scope hierarchy system can be constructed based on the merchant traffic of the merchant intelligent body by industry, covering multiple industries such as education, medical and aesthetic medicine, real estate and home furnishing, and business services. Based on the business scope hierarchy system, a demand tree 500 corresponding to the business scope can be established in the first language model. The demand tree 500 includes multiple nodes 501 and at least one edge 502 between the nodes.

[0072] It can be seen that the demand tree 500 is a tree structure, which is a hierarchical structure composed of nodes. Each node 501 of the demand tree 500 can contain zero or more child nodes, but can only have one parent node. According to the business scope hierarchy system, the hierarchy of the demand tree can be set to multiple layers, and can contain a total of multiple leaf nodes (nodes with an out-degree of 0), ensuring that the business needs of various industries are fully covered.

[0073] Similarly, when the basic demand is the conversation intention, the intent hierarchy system can be constructed by industry based on the business advertising and marketing business scenarios of the merchant intelligent body, covering multiple industries such as education, medical and aesthetics, real estate and home furnishing, and business services. Based on this intent hierarchy system, a demand tree corresponding to the business scope can be established in the first language model. The hierarchy of the demand tree can be set to multiple layers, and can contain multiple leaf nodes in total to refine the user's conversation intention.

[0074] Through the embodiments of the present disclosure, the merchant intelligent agent uses the reasoning and generation capabilities of the large language model to understand the unprocessed dialogue data input by the user, and identifies the key business points and the user's deep intentions contained in the unprocessed dialogue data.

[0075] Figure 6 is a flowchart of obtaining location information according to an embodiment of the present disclosure.

[0076] like Figure 6 As shown, according to an embodiment of the present disclosure, the agent-based recommendation method also includes operations S601 to S606.

[0077] In operation S601, the third language model of the agent is used to identify the position information of the dialogue data to be processed.

[0078] For example, the third largest language model may be a Natural Language Processing (NLP) model.

[0079] In operation S602, it is determined whether the location information is recognized, if so, proceed to operation S603; otherwise, proceed to operation S604.

[0080] In operation S603, in response to the identified location information, the location information and the target demand are matched with a plurality of material information in the material library to obtain a target material that matches the target object.

[0081] In operation S604, in response to the unrecognized location information, the target object is requested to authorize the location information.

[0082] In operation S605, after obtaining authorization from the target object for the location information, the location information is acquired from the geographical location where the target object is located.

[0083] For example, the geographical location of the target object may be latitude and longitude data.

[0084] In operation S606, the location information and the target demand are matched with the multiple material information in the material library to obtain the target material matching the target object.

[0085] Through the embodiments of the present disclosure, the third language model is used to identify the location of the conversation data to be processed from the user. When the location information is not identified, the latitude and longitude data authorized by the user are used to determine the location information. The user's location information is obtained, which facilitates the subsequent recommendation of localized materials (such as products or institutions) to ensure that the recommended target materials meet the user's regional needs.

[0086] At this point, the merchant intelligent agent can identify the user's key business points and deep intentions, collect the user's location information, and comprehensively portray the user's portrait in the conversation interaction in order to make accurate product recommendations and improve user experience and recommendation effects.

[0087] According to an embodiment of the present disclosure, the material library is obtained by processing a plurality of materials to be recommended using an intelligent agent.

[0088] Figure 7 It is a flow chart of forming a material library according to one embodiment of the present disclosure.

[0089] like Figure 7 As shown, according to an embodiment of the present disclosure, the material library is obtained through the following operations S731 to S735.

[0090] In operation S731, a plurality of materials to be recommended are obtained, wherein each material to be recommended includes a material image and description information.

[0091] For example, the materials to be recommended may be materials that a merchant wishes to recommend, including products and organizations. The description information may be in the form of text, voice, video, etc.

[0092] In operation S732, a plurality of materials to be recommended are classified according to the entity attributes of the agent to obtain a plurality of category material sets.

[0093] For example, the entity attributes of the agent can be the merchant attributes of the merchant agent, such as the industry in which the merchant is located, the type of products sold, etc. Products can be classified according to these merchant attributes to form product cards, case cards, institution cards, etc. for general industries, as well as disease cards and teacher cards with industry characteristics, etc., to ensure the diversity and coverage of product types.

[0094] In operation S733, for each material in each category material set, a corresponding category label is added to the material.

[0095] The category label can correspond to the sub-demand represented by each node in the demand tree, thereby marking each material in each category material set for subsequent accurate matching with the basic demand.

[0096] In operation S734, the fourth language model of the intelligent agent is used to summarize the description information corresponding to each material in each category material set to obtain summary information.

[0097] In operation S735 , the material image of each material in the plurality of materials to be recommended and the corresponding category label and summary information are sorted to obtain a material library.

[0098] Through the embodiments of the present disclosure, each material in the material library is classified and labeled, and the classified material description information is summarized using a large language model to ensure the accuracy and relevance of the recommended materials, so as to achieve accurate material recommendations.

[0099] In some embodiments, the material image of each material among multiple materials to be recommended and the corresponding category label and summary information are sorted to obtain a material library, which also includes: performing interface (User Interface, UI) visual design on the material image of each material among multiple materials to be recommended to obtain a designed material image; the designed material image of each material among multiple materials to be recommended and the corresponding category label and summary information are sorted to obtain a material library.

[0100] Through the embodiments of the present disclosure, a reasonable UI design is performed on the material image to improve the visual effect of the material display and the user experience, so that the recommended materials are more attractive to users and meet the needs of users.

[0101] In some embodiments, the fourth largest language model is used to summarize the description information corresponding to each material in each category of material set to obtain summary information, and also includes: according to the dialogue data to be processed and the historical dialogue data, the fifth largest language model of the intelligent agent is used to adjust the description information of each material to be recommended to obtain adjusted description information; the fourth largest language model is used to summarize the adjusted description information corresponding to each material in each category of material set to obtain summary information.

[0102] Through the embodiments of the present disclosure, the description information of the recommended products is polished using the fifth language model according to the user's current dialogue data to be processed and context information, so that the recommended materials can be more attractive to users and meet user needs.

[0103] In some embodiments, the target demand is matched with multiple materials in a material library to obtain a target material that matches the target object, including: based on a predetermined matching strategy, determining the matching degree between the target demand and each material in the material library, wherein the predetermined matching strategy includes one of a precise matching strategy, a fuzzy matching strategy, and a partial matching strategy; at least one material having a matching degree greater than a preset matching threshold is determined as the target material.

[0104] For example, the range of the matching degree may be [0, 1], and the preset matching threshold is 0.8. Since each material in the material library can determine the matching degree with the target demand, multiple matching degrees are obtained. If the matching degree of any material in the material library with the target demand is greater than 0.8, the material can be determined as the target material. The target material can be one or more.

[0105] For example, the exact matching strategy can be that the material content in the material library is completely consistent with the target demand; the fuzzy matching strategy can be that the material content in the material library is similar to the target demand; the partial matching strategy can be that the material content in the material library is consistent with part of the target demand.

[0106] Through the embodiments of the present disclosure, multiple matching strategies are supported to ensure that the recommended products meet the needs of users to the greatest extent, thereby improving user experience and satisfaction.

[0107] In some embodiments, the agent-based recommendation method also includes: setting a priority for each material in the material library according to the entity attributes of the agent; when it is determined that there are multiple materials with a matching degree greater than a preset matching threshold, sorting the M materials with a matching degree greater than the preset matching threshold according to priority, and determining the m materials with the highest priority as target materials, where M is an integer greater than 1, 1≤m≤M and m is an integer.

[0108] Through the embodiments of the present disclosure, priorities are set for different materials. When there are multiple materials with a matching degree greater than a preset matching threshold, they can be screened according to the preset priority to provide the most appropriate product recommendation, thereby ensuring the accuracy and practicality of the recommendation.

[0109] Through the embodiments of the present disclosure, based on the user portrait and the classification results of each material in the material library, the merchant intelligent agent can recommend different types of products according to the different business scopes or conversation intentions of the user during the conversation, thereby achieving accurate matching of the user's deep needs with the advertiser's products.

[0110] For example, when the user's conversation intention is to inquire about test materials, the information card is recommended to the user; when the user's conversation intention is to inquire about prices, the appointment card is recommended to the user; when the user's conversation intention is to inquire about institutions, the institution card is recommended to the user.

[0111] In some embodiments, displaying the target material to the target object includes: generating a visualization chart based on the category label and summary information corresponding to the target material; embedding the visualization chart into the material image of the target material, and displaying it to the target object.

[0112] For example, a visualization chart can be a bar chart, line chart, column chart, pie chart, scatter chart, etc. Visualization charts use graphics as carriers to transform abstract data into visually concrete charts to convey information. They are beautiful and intuitive, thereby improving the display effect of the target material.

[0113] In some embodiments, the agent-based recommendation method further includes: determining reply information for the conversation data to be processed; feeding back the reply information to the target object, and displaying the target material.

[0114] Through the embodiments of the present disclosure, while recommending materials, the conversation initiated by the user is taken over by the script to ensure that the recommendation process is natural and smooth, and to avoid recommending products that are too stiff or abrupt.

[0115] The agent-based recommendation method provided by the present invention can quickly respond to the user's instant input, provide personalized and accurate product recommendations based on the user portrait and the classification results of each material in the material library, and improve the recommendation effect and user experience.

[0116] Based on the agent-based recommendation method provided by the present disclosure, the present disclosure also provides an agent-based recommendation device. Figure 8 The device is described in detail.

[0117] Figure 8 is a block diagram of an agent-based recommendation apparatus according to an embodiment of the present disclosure.

[0118] like Figure 8 As shown, the agent-based recommendation device 800 may include a basic demand determination module 810 , a target demand determination module 820 and a material matching module 830 .

[0119] The basic demand determination module 810 is used to input the dialogue data to be processed from the target object into the first language model of the intelligent agent to obtain the basic demand of the target object.

[0120] The target demand determination module 820 is used to input the historical dialogue data and basic demands of the target object in a preset time period into the second language model of the intelligent agent to obtain the target demand of the target object.

[0121] The material matching module 830 is used to match the target demand with the multiple material information in the material library, obtain the target material that matches the target object, and display the target material to the target object.

[0122] It should be noted that the implementation method of the device part is similar to the implementation method of the method part, and the technical effects achieved are also similar. For specific details, please refer to the above-mentioned method implementation method part, which will not be repeated here.

[0123] Fig. 9 is a block diagram of an artificial intelligence agent according to an embodiment of the present disclosure.

[0124] In the embodiments of the present disclosure, inspired by the von Neumann structure in modern computer theory, such as Fig. 9 As shown, an artificial intelligence (AI) agent (hereinafter referred to as AI agent) 900 may include five core modules: an input module 910 , a control module 920 , a storage module 930 , a calculation module 940 and an output module 950 .

[0125] The input module 910 is responsible for receiving or sensing information such as queries, requests, instructions, signals or data from the outside world (such as users or external environment), and converting them into a format that can be understood and processed by the AI ​​agent 900. The input module 910 is the primary link for the AI ​​agent 900 to interact with the outside world. It enables the AI ​​agent 900 to efficiently and accurately obtain the necessary "sensory" information from the outside world and respond to this information.

[0126] In an example, the input module 910 may input the conversation data to be processed described above.

[0127] In the example, the control module 920 is the core support for the AI ​​agent 900 to handle complex tasks. The control module 920 can execute the agent-based recommendation method described above.

[0128] In the example, the control module 920 will continuously interact with the storage module 930, the computing module 940 and / or the output module 950 during operation. However, it should be noted that in the embodiment of the present disclosure, the control module 920 acts as a single initiator to initiate communication with the storage module 930, the computing module 940 and / or the output module 950, and there is no communication coupling between the storage module 930, the computing module 940 and the output module 950.

[0129] In the example, the performance of the control module 920 may be closely related to the large language model based on the AI ​​agent 900. In order to give full play to the capabilities of the large language model, the internal structure of the control module 920 may be designed to be highly configurable and extensible so as to cope with various types of tasks and requirements in real scenarios.

[0130] The storage module 930 may be responsible for memorizing information such as historical conversations, event streams, etc. The historical conversation data as described above may be included in the storage module 930 .

[0131] In the example, after the AI ​​agent 900 obtains the dialogue data to be processed, the AI ​​agent 900 can use the first language model to determine the basic needs of the target object from the dialogue data to be processed. The basic needs can be stored in the storage module 930. The AI ​​agent 900 can retrieve the historical dialogue data and basic needs of the relevant target object in a preset period from the storage module 930, and feed it back to the control module 920. Then, the control module 920 can use the fed-back historical dialogue data and basic needs to obtain the target needs of the target object, and the target needs can be stored in the storage module 930.

[0132] The operation module 940 can be regarded as a predefined tool library. The predetermined matching strategy and the like as mentioned above can be included in the operation module 940 .

[0133] In the example, when the AI ​​agent 900 needs to match materials for the target demand, the relevant predetermined matching strategy can be called from the operation module 940 and fed back to the control module 920. Then, the control module 920 can use the predetermined matching strategy fed back to determine the matching degree between the target demand and each material in the material library, determine at least one material with a matching degree greater than the preset matching threshold as the target material, and pass the target material to the output module 950. It can be understood that although the large language model has excellent language understanding and generation capabilities, it is like a human being, without the help of any tools, and the tasks that can be solved are very limited. When the AI ​​agent 900 is given the ability to call tools, it can achieve tasks such as completing mathematical operations with the help of a calculator, completing data analysis with the help of python, and completing weather forecasts with the help of search engines.

[0134] In an example, the output module 950 may output the target material described above.

[0135] The AI ​​agent 900 according to the embodiment of the present disclosure can simply and effectively improve the level of intelligence, and improve flexibility and versatility.

[0136] According to an embodiment of the present disclosure, an electronic device, a readable storage medium and a computer program product are also provided.

[0137] According to an embodiment 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.

[0138] 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 cause a computer to execute the method as described above.

[0139] According to an embodiment of the present disclosure, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the computer program implements the method as described above.

[0140] Fig.10 is a block diagram of an electronic device to which an agent-based recommendation method can be applied according to an embodiment of the present disclosure. 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 may 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 claimed herein.

[0141] like Fig.10 As shown, the electronic device 1000 includes a computing unit 1001, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1002 or a computer program loaded from a storage unit 1008 to a random access memory (RAM) 1003. In the RAM 1003, various programs and data required for the operation of the electronic device 1000 can also be stored. The computing unit 1001, the ROM 1002, and the RAM 1003 are connected to each other via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0142] Multiple components in the electronic device 1000 are connected to the I / O interface 1005, including: an input unit 1006, such as a keyboard, a mouse, etc.; an output unit 1007, such as various types of displays, speakers, etc.; a storage unit 1008, such as a disk, an optical disk, etc.; and a communication unit 1009, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 1009 allows the electronic device 1000 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0143] The computing unit 1001 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 1001 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 (DSP), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 1001 performs the various methods and processes described above, such as a text processing method and / or a deployment method of a deep learning framework. For example, in some embodiments, the text processing method and / or the deployment method of a deep learning framework may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 1008. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 1000 via the ROM 1002 and / or the communication unit 1009. When the computer program is loaded into RAM 1003 and executed by computing unit 1001, one or more steps of the text processing method and / or the method for deploying a deep learning framework described above may be performed. Alternatively, in other embodiments, computing unit 1001 may be configured to execute the text processing method and / or the method for deploying a deep learning framework in any other appropriate manner (e.g., by means of firmware).

[0144] Various embodiments of the systems and techniques described above herein may be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard parts (ASSPs), system on chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may 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 may be a dedicated 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.

[0145] 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.

[0146] 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. More specific examples of machine-readable storage media may include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories, read-only memories, erasable programmable read-only memories (EPROM) or flash memories, optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0147] 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 cathode ray tube (CRT) display or a liquid crystal display (LCD)) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a 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).

[0148] The systems and techniques described herein may be implemented in a computing system that includes backend 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 frontend 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 backend components, middleware components, or frontend 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.

[0149] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0150] 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.

[0151] 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. An agent-based recommendation method, comprising: Inputting the dialogue data to be processed from the target object into the first language model of the agent to obtain the basic needs of the target object; Inputting the historical conversation data of the target object in a preset time period and the basic demand into the second language model of the agent to obtain the target demand of the target object; The target demand is matched with a plurality of material information in a material library to obtain a target material matching the target object, and the target material is displayed to the target object.

2. The method according to claim 1, further comprising: The generated object of the agent conducts at least one round of dialogue with the target object to obtain dialogue data; The historical conversation data of the target object in a preset time period is extracted from the conversation data.

3. The method according to claim 1, wherein: The basic requirement is multiple; the method further includes: For each basic requirement, according to the entity attributes of the agent, a requirement tree corresponding to the basic requirement is configured in the first language model; The demand tree includes multiple nodes and at least one edge between the nodes, each node represents a sub-demand of the basic demand, and each edge represents the relationship between the sub-demands.

4. The method according to claim 3, wherein: Input the to-be-processed dialogue data from the target object into the first language model of the agent to obtain the basic requirements of the target object, including performing the following operations for each basic requirement: Matching the to-be-processed conversation data with a plurality of nodes of the demand tree to obtain at least one matching node; The sub-demand represented by the at least one node is determined as the basic demand of the target object.

5. The method according to claim 1, wherein: The basic requirements include business scope and conversation intention.

6. The method according to claim 1, further comprising: Using the third language model of the agent, identifying position information of the dialogue data to be processed; In response to the identified location information, the location information and the target demand are matched with a plurality of material information in a material library to obtain a target material that matches the target object.

7. The method according to claim 6, further comprising: In response to the unrecognized location information, requesting the target object to authorize the location information; Acquire the location information from the geographical location of the target object when the target object authorizes the location information; The location information and the target demand are matched together with a plurality of material information in a material library to obtain a target material that matches the target object.

8. The method according to claim 1, wherein: The material library is obtained by processing a plurality of materials to be recommended using the intelligent agent.

9. The method according to claim 8, wherein: The material library is obtained in the following way: Obtain multiple materials to be recommended, wherein each material to be recommended includes a material image and description information; Classifying the plurality of materials to be recommended according to the entity attributes of the agent to obtain a plurality of category material sets; For each material in each category material set, add a corresponding category label to the material; Using the fourth language model of the agent, the description information corresponding to each material in each category material set is summarized to obtain summary information; The material image of each material in the plurality of materials to be recommended and the corresponding category label and summary information are sorted to obtain the material library.

10. The method according to claim 9, wherein: The material image and the corresponding category label and summary information of each material in the plurality of materials to be recommended are sorted to obtain the material library, further comprising: Performing interface visual design on the material image of each of the multiple materials to be recommended to obtain a designed material image; The designed material image of each material among the multiple materials to be recommended and the corresponding category label and summary information are sorted to obtain the material library.

11. The method according to claim 9, wherein: The fourth language model is used to summarize the description information corresponding to each material in each category material set to obtain summary information, which also includes: According to the to-be-processed conversation data and the historical conversation data, using the fifth language model of the agent, adjusting the description information of each to-be-recommended material to obtain adjusted description information; The fourth language model is used to summarize the adjusted description information corresponding to each material in each category material set to obtain summary information.

12. The method according to claim 9, wherein: Displaying the target material to the target object includes: Generate a visualization chart based on the category label and summary information corresponding to the target material; The visualization chart is embedded in the material image of the target material and displayed to the target object.

13. The method according to claim 1, wherein: Matching the target demand with multiple materials in the material library to obtain a target material that matches the target object includes: Based on a predetermined matching strategy, determining the matching degree between the target demand and each material in the material library, wherein the predetermined matching strategy includes one of an exact matching strategy, a fuzzy matching strategy and a partial matching strategy; At least one material having a matching degree greater than a preset matching threshold is determined as the target material.

14. The method according to claim 13, further comprising: According to the entity attributes of the intelligent agent, a priority is set for each material in the material library; When it is determined that there are multiple materials whose matching degree is greater than the preset matching threshold, the M materials whose matching degree is greater than the preset matching threshold are sorted according to the priority, and the m materials with the highest priority are determined as the target materials, where M is an integer greater than 1, 1≤m≤M and m is an integer.

15. The method according to claim 1, further comprising: Determining reply information for the conversation data to be processed; The reply information is fed back to the target object, and the target material is displayed.

16. An agent-based recommendation device, comprising: A basic demand determination module, used for inputting the dialogue data to be processed from the target object into the first language model of the intelligent agent to obtain the basic demand of the target object; A target demand determination module, used for inputting the historical conversation data of the target object in a preset time period and the basic demand into the second language model of the intelligent agent to obtain the target demand of the target object; The material matching module is used to match the target demand with multiple material information in the material library to obtain the target material that matches the target object, and display the target material to the target object.

17. An artificial intelligence agent configured to perform the method according to any one of claims 1 to 15.

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 executable 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 15.

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 15.

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 15.

Images