Object recommendation method and device based on large model, equipment, medium, program product and intelligent agent

Abstract

The invention provides an object recommendation method and device based on a large model, equipment, a medium, a program product and an intelligent agent, relates to the technical field of artificial intelligence, in particular to the technical fields of large models, generative models and the like, and can be applied to scenes of AIGC, product recommendation and the like. The object recommendation method comprises the following steps: in response to received demand description, determining at least one candidate object in an object library according to an object label determined based on the demand description; according to the demand description and the at least one candidate object, an object recommendation result is determined, the object recommendation result comprises at least one target object in the candidate objects and at least one explanatory description, and the explanatory description represents the attribute of the target object and the mapping relation between the demand description and the target object.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, in particular to technical fields such as large models and generative models, and can be applied to scenarios such as AIGC and product recommendation. More specifically, it relates to an object recommendation method, device, equipment, medium, program product and intelligent agent based on large models. Background Art

[0002] With the development of artificial intelligence technology, artificial intelligence models can be applied to product recommendation scenarios. As an example, a user can input requirements into a product recommendation model, and the product recommendation model can generate recommended target products based on the user's requirements. Summary of the invention

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

[0004] According to one aspect of the present disclosure, a large model-based object recommendation method is provided, comprising: in response to receiving a requirement description, determining at least one candidate object in an object library according to an object label determined based on the requirement description; and determining an object recommendation result based on the requirement description and the at least one candidate object, wherein the object recommendation result includes at least one target object among the candidate objects, and at least one explanatory description, the explanatory description characterizing attributes of the target object, and a mapping relationship between the requirement description and the target object.

[0005] According to another aspect of the present disclosure, a large model-based object recommendation device is provided, including: a first determination module, used to determine at least one candidate object in an object library in response to receiving a requirement description and according to an object label determined based on the requirement description; and a second determination module, used to determine an object recommendation result based on the requirement description and the at least one candidate object, wherein the object recommendation result includes at least one target object among the candidate objects, and at least one explanatory description, the explanatory description characterizing the attributes of the target object, and a mapping relationship between the requirement description and 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 embodiment of the present disclosure.

[0007] According to another aspect of the present disclosure, an electronic device is provided, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0008] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program or instruction is stored. When the computer program or instruction is executed by a processor, the steps of the above method are implemented.

[0009] According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program or instructions, wherein the computer program or instructions implement the steps of the above method 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 above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0012] Figure 1 Schematically illustrates a system architecture to which a large model-based object recommendation method, apparatus, device, medium, program product, and intelligent agent can be applied according to an embodiment of the present disclosure;

[0013] Figure 2 A flowchart of a method for recommending an object based on a large model according to an embodiment of the present disclosure is schematically shown;

[0014] Figure 3 The object recommendation process based on a large model according to an embodiment of the present disclosure is schematically shown;

[0015] Figure 4 Schematically illustrates a large model-based object recommendation process according to another embodiment of the present disclosure;

[0016] Figure 5 A schematic diagram schematically shows a method for determining an object recommendation result according to an embodiment of the present disclosure;

[0017] Figure 6 A schematic diagram of determining a requirement description according to an embodiment of the present disclosure is schematically shown;

[0018] Figure 7 A block diagram of a large model-based object recommendation device according to an embodiment of the present disclosure is schematically shown;

[0019] Figure 8 A block diagram schematically shows a structure of an artificial intelligence agent according to an embodiment of the present disclosure; and

[0020] Fig. 9A block diagram of an electronic device suitable for implementing a large model-based object recommendation method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0021] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0022] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0023] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0024] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0025] With the development of artificial intelligence technology, artificial intelligence models can be applied to product recommendation scenarios. As an example, a user can input requirements into a product recommendation model, and the product recommendation model can generate recommended target products based on the user's requirements.

[0026] The product recommendation function is one of the core capabilities of enterprises to improve market competitiveness and customer satisfaction. Accurate product recommendations can not only improve customer conversion rates, but also help customers quickly find products that meet their needs, thereby enhancing customers' trust and loyalty to the company. In the process of product recommendation, understanding user intentions is particularly important. Due to the diversity and complexity of user demand expressions, the recommendation system must have the ability to deeply analyze user input in order to accurately capture the user's core needs.

[0027] In one example, the manner of implementing product recommendation may include at least one of the following: rule-based recommendation, collaborative filtering-based recommendation, content filtering-based recommendation, and the like.

[0028] Rule-based recommendation refers to classifying and recommending users according to predefined rules. However, since this method relies on expert knowledge, it is difficult to cope with complex and diverse demand scenarios. It has poor flexibility and scalability.

[0029] Recommendation based on collaborative filtering refers to recommending products based on the behaviors of similar users. However, due to the data sparsity problem of this method, the recommendation effect will be poor when there is insufficient interaction data between users and products. For example, the cold start problem (new users or new products cannot be recommended). At the same time, recommendations based on the similarities between users or products cannot truly understand the multi-level needs of users.

[0030] Content filtering-based recommendation refers to recommending similar products based on product features and user historical preferences, behavioral data, etc. However, since this method only matches explicit product attributes (such as insurance coverage and premium), it is difficult to capture users' potential intentions or implicit needs, and has limited ability to understand vague expressions.

[0031] In summary, current product recommendation methods are not able to understand complex user intentions and cannot extract deep intentions from complex user descriptions. At the same time, they are not robust to data sparsity and cannot generate meaningful recommendations based on a small amount of context. Moreover, in the final recommendation stage, traditional models cannot generate natural language explanations to intuitively present the recommendation process and reasons to users, and cannot gain users' trust.

[0032] To this end, the disclosed embodiment proposes an object recommendation scheme based on a large model. For example, in response to receiving a requirement description, at least one candidate object is determined in an object library according to an object label determined based on the requirement description; and an object recommendation result is determined according to the requirement description and the at least one candidate object, wherein the object recommendation result includes at least one target object among the candidate objects, and at least one explanatory description, wherein the explanatory description characterizes the attributes of the target object, and a mapping relationship between the requirement description and the target object.

[0033] According to the embodiments of the present disclosure, by determining the object label based on the demand description, the user intention represented by the demand description can be understood more deeply and accurately based on the object label, and the ability to understand complex user intentions can be improved. By using the object label to more accurately characterize the object that matches the demand description, the candidate objects that match the demand description can be quickly identified and recalled in the object library, and the target object among the candidate objects can be determined to ensure the matching degree between the target object and the user intention. The object recommendation result can be determined based on the target object and the explanatory description for the target object. The corresponding relationship between the attributes of the target object and the object label determined based on the demand description can be characterized through the explanatory description to reflect the thinking, analysis and processing process of selecting and recommending the target object, as the recommendation process and recommendation reason of the target object, so that the object recommendation result can be more in line with the user's intention, with higher interpretability and credibility. By displaying the object recommendation result to the user, it can help the user intuitively understand the recommendation process and recommendation reason of the target object, which can effectively improve the accuracy and credibility of the object recommendation, and help improve the user experience and service quality.

[0034] It should be noted that the method provided according to the embodiment of the present disclosure can be applied to multiple application scenarios such as intelligent customer service, e-commerce platforms, and financial product recommendations. The object recommendation method based on a large model provided by the embodiment of the present disclosure does not limit the application scenarios.

[0035] In the technical solution of the present invention, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0036] In the technical solution of the present invention, the user's authorization or consent is obtained before obtaining or collecting the user's personal information.

[0037] Figure 1 The system architecture to which the large model-based object recommendation method can be applied according to an embodiment of the present disclosure is schematically shown. It should be noted that: Figure 1 What is shown is merely an example of a system architecture 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 it does not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios.

[0038] like Figure 1As shown, the system architecture 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104 and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0039] The user may use at least one of the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0040] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0041] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 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 user requests) to the terminal device.

[0042] The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server can also be a server of a distributed system, or a server combined with blockchain.

[0043] It should be noted that the object recommendation method based on the big model provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the object recommendation device based on the big model provided in the embodiment of the present disclosure can generally be set in the server 105. The object recommendation method based on the big model provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Correspondingly, the object recommendation device based on the big model provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0044] Alternatively, the large model-based object recommendation method provided in the embodiment of the present disclosure may also be executed by the first terminal device 101, the second terminal device 102, or the third terminal device 103, or may also be executed by other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103. Accordingly, the large model-based object recommendation device provided in the embodiment of the present disclosure may also be provided in the first terminal device 101, the second terminal device 102, or the third terminal device 103, or may be provided in other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103.

[0045] 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.

[0046] It should be noted that the sequence numbers of the operations in the following method are only used as representations of the operations for the purpose of description, and should not be regarded as representing the execution order of the operations. Unless explicitly stated, the method does not need to be executed completely in the order shown.

[0047] The above describes the system architecture of the object recommendation method based on a large model provided by the present disclosure. Figure 2 For example, the large model-based object recommendation process of the present disclosure is further described.

[0048] Figure 2 The flowchart of the large model-based object recommendation method according to an embodiment of the present disclosure is schematically shown.

[0049] like Figure 2 As shown, the large model-based object recommendation method 200 includes operations S210 to S220.

[0050] In operation S210 , in response to receiving the requirement description, at least one candidate object is determined in an object library according to an object tag determined based on the requirement description.

[0051] In operation S220, an object recommendation result is determined based on the requirement description and at least one candidate object, wherein the object recommendation result includes at least one target object among the candidate objects, and at least one explanatory description, the explanatory description characterizing the attributes of the target object, and a mapping relationship between the requirement description and the target object.

[0052] The object can be any type of recommended object such as merchandise, media, entertainment services, educational services, real estate, financial products, etc., without limitation. In response to receiving the requirement description, an object tag can be determined based on the requirement description, and at least one candidate object can be determined in the object library based on the object tag.

[0053] The demand description may include a natural language description input by the user, and the demand description may be used to characterize the user's product demand intention. For example, the demand description may include the applicable objects of the product, the user's requirements for the product, the user's expectations for the product, etc. For example, the demand description may include "need medical insurance suitable for single young people" or "an insurance plan with high coverage but limited budget".

[0054] Object labels can be used to characterize objects that match the demand description, and object labels can be used to quickly identify and recall candidate objects that match the demand description in the preset object library. For example, based on "need medical insurance suitable for single young people", it can be determined that object labels may include "medical insurance", "age 20-35 years old", "good health", "low premium", etc.

[0055] The object library may include at least one candidate object, and the type of the candidate object corresponds to the scenario of the object recommendation. As an example, in an e-commerce recommendation scenario, the object library may include at least one commodity, and the object library may include, for example, digital commodities, beauty commodities, household commodities, home appliances, clothing commodities, food, etc.

[0056] In one example, an object recommendation result may be determined based on the requirement description and at least one candidate object, and the object recommendation result may include at least one target object among the candidate objects. For example, at least one target object may be determined from at least one candidate object. For example, an explanatory description for the target object may be determined based on the attributes of the target object and the mapping relationship between the requirement description and the target object. For example, an object recommendation result may be determined based on at least one target object and at least one explanatory description of each of the target objects.

[0057] The explanatory description may characterize the properties of the target object and the mapping relationship between the requirement description object and the target object. For example, the properties of the target object may include the price, brand, applicable object, category, specification parameters, function / effect, etc. of the target object. For example, the mapping relationship between the requirement description and the target object may characterize the corresponding relationship between the requirement description and the target object, for example, the corresponding relationship between the properties of the target object and the object label determined based on the requirement description may be characterized.

[0058] It is understandable that the correspondence between the attributes of the target object and the object label can serve as an important reason for recommending the target object, helping users understand why the target object is recommended. Based on this, the explanatory description can characterize the correspondence between the attributes of the target object and the object label determined based on the demand description, to reflect the thinking, analysis and processing process of selecting and recommending the target object, as the recommendation process and reason for the target object, so that the object recommendation results can be more in line with user intentions, have higher interpretability and credibility. In one example, the explanatory description can be a natural language description to enable users to better understand the recommendation process and reason for the target object, and increase users' trust in the object recommendation results.

[0059] The object recommendation results can be displayed to the user to intuitively present the target object and the recommendation process and reasons for the target object to the user, thereby making the object recommendation results more explainable and credible, which is conducive to enhancing the user's sense of trust.

[0060] Exemplarily, the object recommendation information may be presented in at least one of the following forms: text, voice, image, structured information (eg, table or form), and the like.

[0061] According to the embodiments of the present disclosure, by determining the object label based on the demand description, the user intention represented by the demand description can be understood more deeply and accurately based on the object label, and the ability to understand complex user intentions can be improved. By using the object label to more accurately characterize the object that matches the demand description, the candidate objects that match the demand description can be quickly identified and recalled in the object library, and the target object among the candidate objects can be determined to ensure the matching degree between the target object and the user intention. The object recommendation result can be determined based on the target object and the explanatory description for the target object. The corresponding relationship between the attributes of the target object and the object label determined based on the demand description can be characterized through the explanatory description to reflect the thinking, analysis and processing process of selecting and recommending the target object, as the recommendation process and recommendation reason of the target object, so that the object recommendation result can be more in line with the user's intention, with higher interpretability and credibility. By displaying the object recommendation result to the user, it can help the user intuitively understand the recommendation process and recommendation reason of the target object, which can effectively improve the accuracy and credibility of the object recommendation, and help improve the user experience and service quality.

[0062] The object recommendation method based on a big model provided in the embodiments of the present disclosure can be applied to insurance product recommendation scenarios. For example, the object recommendation method based on a big model can be applied to insurance product recommendation scenarios such as insurance product recommendation on online insurance platforms, insurance product recommendation intentions of intelligent customer service systems, offline auxiliary insurance product sales, and internal data analysis and marketing of insurance companies.

[0063] As an example, in the insurance product recommendation scenario, the object library may include medical insurance, accident insurance, life insurance, disease insurance, pension insurance, financial insurance, etc. The attributes of the target object may include, for example, the insured amount, premium, coverage, applicable object, insurance period, insurance type, insurance features, etc.

[0064] In one example, taking the insurance product recommendation scenario as an example, the object recommendation result may be as shown in Table 1.

[0065]

[0066] Table 1

[0067] According to an embodiment of the present disclosure, determining at least one candidate object in an object library according to an object label determined based on a requirement description includes: calling a first large model based on first prompt information, analyzing the requirement description to generate at least one first object label, wherein the first prompt information is used to instruct the first large model to generate a first object label corresponding to the requirement description; and determining at least one candidate object in the object library according to the first object label.

[0068] A large model can be a model with a large number of parameters. The parameter magnitude is usually in the tens of millions or hundreds of millions, and may reach billions or tens of billions. In terms of the network structure of a large model, for example, a network structure such as UFO (Unified Feature Optimization) can be used.

[0069] In one example, the requirement description and the first prompt information may be input into the first large model to generate at least one first object tag. The first prompt information may be used to instruct the large model to generate a first object tag corresponding to the requirement description.

[0070] By calling the first large model to analyze the demand description based on the first prompt information, the first large model can be controlled based on the first prompt information to understand the user intention represented by the demand description more deeply and accurately, thereby improving the first large model's ability to understand complex user intentions, thereby generating a more accurate first object label.

[0071] According to the embodiments of the present disclosure, by utilizing the powerful semantic understanding ability of the first large model to better capture the user's potential intentions or implicit needs, the first object label and the demand description can be more overlapped in the semantic space, thereby improving the matching degree between the first object label and the user's intention represented by the demand description. In addition, by generating the first object label through the first large model, the first object label can also avoid being restricted by the inherent label library, so that it can more flexibly adapt to the diverse needs of users, and does not require a large amount of annotation or complex preprocessing of the data, which helps to improve the generation quality of the first object label in data sparse scenarios.

[0072] According to an embodiment of the present disclosure, the first prompt information may be in the form of structured text.

[0073] In one example, taking the insurance product recommendation scenario as an example, the first prompt information may be as shown in Table 2.

[0074]

[0075] Table 2

[0076] It should be noted that in Table 2, the content of the ${} brackets is dynamically changing and can be filled in according to the application scenario in actual applications, and the subsequent prompt word information is the same.

[0077] In one example, taking the pension insurance product recommendation scenario as an example, the generated first object tag may be as shown in Table 3:

[0078]

[0079] Table 3

[0080] According to an embodiment of the present disclosure, the object library includes at least two candidate object identifiers and an object description and a description feature of each candidate object identifier; according to the first object label, determining at least one candidate object in the object library includes at least one of the following: according to the first object label, text matching is performed on at least two object descriptions to obtain at least one candidate object; according to the first label feature of the first object label, vector matching is performed on at least two description features to obtain at least one candidate object.

[0081] The object library may include at least two candidate object identifiers, and an object description and description features of each candidate object identifier. For example, the candidate object identifier may be used to characterize the candidate object. For example, the object description may include a natural language description text corresponding to the candidate object. For example, the description feature may include a semantic feature vector of the object description.

[0082] In one example, taking the insurance product recommendation scenario as an example, the storage design of the object library can be as shown in Table 4.

[0083]

[0084] Table 4

[0085] In one example, at least two object descriptions may be text matched according to the first object tag to obtain at least one candidate object. Exemplarily, the first object tag and the object description may be text matched based on text similarity, for example, based on a text segmentation algorithm.

[0086] In another example, vector matching can be performed on at least two descriptive features according to the first label feature of the first object label to obtain at least one candidate object. For example, the first label feature can include a semantic feature vector of the first object label. Exemplarily, vector matching can be performed on the first label feature and the descriptive feature based on vector similarity, for example, vector matching can be performed based on a KNN search algorithm (K-Nearest Neighbors Search).

[0087] In yet another example, text matching may be performed on at least two object descriptions according to a first object tag, and vector matching may be performed on at least two description features according to a first tag feature of the first object tag to obtain at least one candidate object.

[0088] According to an embodiment of the present disclosure, at least one of text matching and vector matching can be selected to accurately identify and recall candidate objects that match the object label from the object library based on the first object label and at least one of the first label features, so as to improve the matching degree between the candidate objects and the first object label, and further improve the matching degree between the candidate objects and the user intention represented by the requirement description.

[0089] In one example, at least one candidate object can be determined in the object library using a search engine according to the object tag. Exemplarily, the search engine can be any type of search engine such as Elastic8.x, Nutch, Apache Lucene, etc., which is not limited here.

[0090] For example, the search engine may be Elastic8.x, and the search method may be at least one of the following: text matching using ik word segmentation and vector matching using knn algorithm.

[0091] According to an embodiment of the present disclosure, the object library includes at least two candidate object identifiers and respective descriptive features of each candidate object identifier; determining at least one candidate object in the object library according to the object label determined based on the requirement description includes: calling a second large model based on second prompt information, screening at least one candidate label in the label library according to the requirement description, and obtaining at least one second object label, wherein the second prompt information is used to instruct the second large model to select a second object label corresponding to the requirement description, and the label library also includes respective second label features of each candidate label; and performing vector matching on at least two descriptive features according to the second label features to obtain at least one candidate object.

[0092] The tag library may include at least one candidate tag that matches the candidate object in the object library. For example, based on the candidate tags in the tag library, a target object that matches the candidate tag can be determined from at least one candidate object in the object library.

[0093] As an example, in the insurance product recommendation scenario, the tag library may include the insurance product's insured amount, premium, coverage, applicable object, insurance period, insurance type, insurance features, etc. Alternatively, the tag library also includes a second tag feature for each candidate tag. For example, the second tag feature may include a semantic feature vector of the candidate tag.

[0094] In one example, taking the insurance product recommendation scenario as an example, the storage design of the tag library can be as shown in Table 5.

[0095]

[0096] Table 5

[0097] According to an embodiment of the present disclosure, each candidate tag in the tag library may also be annotated with a candidate tag description, and the candidate tag description may be used to increase the semantic richness of the candidate tag to improve the precision and accuracy of tag screening.

[0098] For example, the candidate tag description may include an explanation of the candidate tag.

[0099] As an example, taking the insurance product recommendation scenario as an example, the candidate tags and candidate tag descriptions can be shown in Table 6.

[0100]

[0101] Table 6

[0102] In one example, when there are a large number of candidate tags in the tag library, the entire tag library may exceed the input limit of the second prompt word, and the second largest model can also pre-circle the screening range of candidate tags according to the requirement description to improve the screening accuracy of the second object tag while meeting the input limit. Exemplarily, the second largest model can pre-screen a set of candidate tags that better meet the requirement description from the tag library based on the candidate tag description and the requirement description.

[0103] For example, the Elastic8.x search engine can be used to pre-screen a set of candidate tags from the tag library using the knn vector matching algorithm.

[0104] In another example, the requirement description and the second prompt information may be input into the second large model to obtain at least one second object tag. The second prompt information may be used to instruct the second large model to select a second object tag corresponding to the requirement description.

[0105] According to an embodiment of the present disclosure, by calling the second largest model to filter at least one candidate tag in the tag library according to the requirement description based on the second prompt information, the filtering range and filtering conditions of the candidate tags by the second largest model can be controlled based on the second prompt information, so that the second object tag that matches the user intention represented by the requirement description can be filtered out from the determined tag library, avoiding the second object tag from exceeding expectations and ensuring that the second object tag is controllable.

[0106] According to the embodiments of the present disclosure, by controlling the screening range and screening conditions of the second largest model for candidate tags based on the determined tag library and the second prompt information, the determined second object tags can be made more controllable, avoiding the second object tags from exceeding expectations and ensuring the quality of the second object tags. In addition, by using the second largest model to screen the second object tags from the tag library, only the second largest model is required to perform classification judgment, the reasoning process is usually lightweight, the model deployment is simple, and the computational cost is low.

[0107] In one example, taking the insurance product recommendation scenario as an example, the second prompt information may be as shown in Table 7.

[0108]

[0109] Table 7

[0110] According to an embodiment of the present disclosure, determining an object recommendation result based on a requirement description and at least one candidate object includes: determining the first n candidate objects among the at least one sorted candidate objects as target objects, where n is a positive integer; and determining an object recommendation result based on the requirement description and the object description of each of the at least one target object.

[0111] In one example, the similarity value of each candidate object can be obtained by calculating the similarity between the object label determined based on the requirement description and at least one candidate object. The similarity calculation method can include vector similarity calculation, for example.

[0112] In another example, at least one candidate object can be ranked based on the size of the similarity value, and the top n candidate objects with the highest similarity values ​​can be determined as target objects. For example, taking 10 candidate objects as an example, when n=3, the top 3 candidate objects with the highest similarity values ​​can be determined as target objects. Alternatively, when n=5, the top 5 candidate objects with the highest similarity values ​​can be determined as target objects. It should be noted that the specific number of n can be configured according to actual business needs, and is not limited here. The specific number of n can be less than or equal to the number of candidate objects. On this basis, the object recommendation result can be determined based on the demand description and the object description of at least one target object.

[0113] In the embodiments of the present disclosure, object tags can be generated based on the demand description using a large model. Figure 3 Taking the example, the process of implementing object recommendation based on a large model by generating object tags based on demand description is explained.

[0114] Figure 3 The object recommendation process based on a large model according to an embodiment of the present disclosure is schematically illustrated.

[0115] like Figure 3 As shown, in response to receiving the requirement description 301, the first large model can be called based on the first prompt information 302 to analyze the requirement description 301 to generate at least one first object tag 303. The first prompt information is used to instruct the first large model to generate a first object tag corresponding to the requirement description.

[0116] The object library may include at least two candidate object identifications 304 and an object description 305 and description features 306 for each candidate object identification 304 .

[0117] Exemplarily, at least one candidate object may be determined in the object library based on at least one of the following methods.

[0118] As an example, text matching may be performed on at least two object descriptions 305 according to the first object tag 303 to obtain at least one candidate object 308 .

[0119] As another example, vector matching may be performed on at least two description features 306 according to the first label feature 307 of the first object label 303 to obtain at least one candidate object 308 .

[0120] The candidate object that is in the first preset position among the at least one sorted candidate object 308 may be determined as the target object 309. On this basis, the object recommendation result 311 may be determined according to the requirement description 301 and the object description 310 of the at least one target object 309.

[0121] In the embodiments of the present disclosure, object tags can also be selected from a tag library based on the requirement description using a large model. Figure 4 Taking the example, the process of implementing object recommendation based on a large model by selecting object labels based on demand description is explained.

[0122] Figure 4 The object recommendation process based on a large model according to another embodiment of the present disclosure is schematically illustrated.

[0123] like Figure 4 As shown, in response to receiving the requirement description 401, the second large model can be called based on the second prompt information 402, and at least one candidate tag 403 in the tag library can be screened according to the requirement description 401 to obtain at least one second object tag 404. The second prompt information is used to instruct the second large model to select the second object tag corresponding to the requirement description.

[0124] The tag library may include at least one candidate tag 403 and a second tag feature 405 of each candidate tag 403. The object library may include at least two candidate object identifiers 406 and a description feature 407 of each candidate object identifier 406.

[0125] In one example, vector matching can be performed on at least two description features 407 according to the second label feature 405 to obtain at least one candidate object 408. The candidate object that is in the first preset position among the at least one candidate object 408 after sorting is determined as the target object 409. On this basis, the object recommendation result 411 can be determined according to the object description 410 of each of the requirement description 401 and the at least one target object 409.

[0126] According to an embodiment of the present disclosure, determining an object recommendation result based on a requirement description and an object description of at least one target object includes: calling a third large model based on third prompt information, generating an explanatory description of each target object based on text features of the requirement description and descriptive features of the object description, wherein the third prompt information is used to indicate that the third large model captures the properties of the target object and the mapping relationship between the requirement description and the target object; and determining the object recommendation result based on the mapping relationship between the target object and the explanatory description.

[0127] In one example, the requirement description, the object description of at least one target object, and the third prompt information can be input into the third large model to generate an explanatory description of each target object. The third prompt information can be used to indicate that the third large model captures the properties of the target object and the mapping relationship between the requirement description and the target object.

[0128] In another example, the object recommendation result may be determined based on the mapping relationship between the target object and the explanatory description.

[0129] According to an embodiment of the present disclosure, by calling the third largest model based on the third prompt information and processing the text features of the requirement description and the descriptive features of the object description, the third largest model can be controlled based on the third prompt information to deeply understand the text features characterizing the requirement description and the descriptive features characterizing the object description, so that the third largest model can better capture the properties of the target object and the mapping relationship between the requirement description and the target object, thereby generating an explanatory description of each target object.

[0130] According to an embodiment of the present disclosure, the third large model can be controlled by the third prompt word to capture the attributes of the target object and the mapping relationship between the requirement description and the target object, so as to convert the matching rules and processes of the target object into easy-to-understand explanatory descriptions according to the attributes of the target object and the corresponding relationship between the object description and the requirement description.

[0131] In one example, taking the insurance product recommendation scenario as an example, the third prompt information may be as shown in Table 8.

[0132]

[0133] Table 8

[0134] In the embodiment of the present disclosure, after determining at least one target object, an object recommendation result can be determined according to the requirement description and the object description of each of the at least one target object. Figure 5 Take as an example to illustrate the process of determining the object recommendation result.

[0135] Figure 5 A schematic diagram of determining an object recommendation result according to an embodiment of the present disclosure is schematically shown.

[0136] like Figure 5 As shown, the third model can be called based on the third prompt information 502, and an explanatory description 504 of each target object can be generated according to the text features 504 of the requirement description and the descriptive features 503 of the object description of the target object. The third prompt information is used to indicate that the third model captures the properties of the target object and the mapping relationship between the requirement description and the target object.

[0137] In one example, the object recommendation result 506 may be determined according to the mapping relationship between the target object 505 and the explanatory description 504 .

[0138] According to an embodiment of the present disclosure, the large model-based object recommendation method also includes, in response to receiving a requirement description, before determining at least one candidate object in the object library according to an object tag determined based on the requirement description: in response to receiving an initial description, using the fourth large model to perform a requirement analysis on the initial description to obtain requirement information; when the requirement information does not meet the object recommendation condition, using the fourth large model to conduct multiple rounds of interactive dialogues with the user to guide the user to supplement the initial description based on preset rules, wherein the object recommendation condition is determined according to the attributes of each candidate object in the object library; and determining the requirement description according to the initial description and the dialogue information obtained after multiple rounds of interactive dialogues.

[0139] In another example, the initial description may be input into the fourth largest model, and the fourth largest model may be used to perform a demand analysis on the initial description to determine the demand information.

[0140] The object recommendation condition can be understood as a necessary condition for determining at least one candidate object in the object library based on the demand information. Exemplarily, the object recommendation condition can be determined according to the attributes of each candidate object in the object library.

[0141] In one example, taking the insurance product recommendation scenario as an example, the object recommendation conditions may include age, gender, job, etc.

[0142] For example, the fourth model may be called to analyze the demand information based on the demand confirmation prompt information to determine whether the demand information meets the object recommendation condition. The demand confirmation prompt information is used to instruct the fourth model to determine the demand information based on the object recommendation condition.

[0143] In one example, taking the insurance product recommendation scenario as an example, the demand confirmation prompt information may be as shown in Table 9:

[0144]

[0145] Table 9

[0146] It is understandable that when user needs are unclear or vague, it is usually difficult for object recommendation models to accurately capture the user's true intentions, and they also face greater challenges when dealing with cold start problems, which may lead to a reduction in the quality of the model's recommendation results and affect the robustness of the object recommendation model.

[0147] In one example, when the demand information does not meet the object recommendation conditions, the fourth model can be used to conduct multiple rounds of interactive dialogues with the user to guide the user to supplement the initial description based on preset rules.

[0148] As an example, the preset rules may include rules related to the attributes of the object, such as hot object rules, main push object rules, etc.

[0149] Exemplarily, the fourth model can ask questions or provide options to the user based on preset rules, so as to guide the user to further clarify, supplement, and improve the initial description through multiple rounds of interactive dialogue with the user until the demand information meets the object recommendation conditions.

[0150] For example, the fourth model can be called to guide prompt information based on demand, and multiple rounds of interactive dialogues can be conducted with the user based on the demand information to guide the user to supplement the initial description. The demand guidance prompt information is used to instruct the fourth model to guide the user to supplement the initial description based on preset rules.

[0151] In one example, taking the insurance product recommendation scenario as an example, the demand guidance prompt information may be as shown in Table 10:

[0152]

[0153] Table 10

[0154] According to the embodiments of the present disclosure, by utilizing the fourth large model to conduct multiple rounds of interactive dialogues with users to guide users to supplement the initial description based on preset rules, users can be helped to explain their needs more comprehensively, thereby enhancing the large model's understanding of user needs. This can more accurately capture the user's true intentions to support subsequent model reasoning processes, which is beneficial to improving the accuracy of object recommendation results and their relevance to user intentions.

[0155] In the embodiment of the present disclosure, before determining the object tag based on the requirement description, the requirement information may also be confirmed. Figure 6 Take the example to explain the process of determining the requirement description.

[0156] Figure 6 A schematic diagram of determining a requirement description according to an embodiment of the present disclosure is schematically shown.

[0157] like Figure 6 As shown, in response to receiving the initial description 601 , the fourth model can be used to perform demand analysis on the initial description 601 to obtain demand information 602 .

[0158] In one example, when the requirement information 602 does not meet the object recommendation condition, the fourth model can be used to conduct multiple rounds of interactive dialogues with the user to guide the user to supplement the initial description based on preset rules. The object recommendation condition can be determined based on the attributes of each candidate object in the object library.

[0159] In one example, when the requirement information 602 after multiple rounds of dialogues meets the object recommendation condition, the requirement description 604 may be determined according to the initial description 601 and the dialogue information 603 obtained through multiple rounds of interactive dialogues.

[0160] It should be noted that the first large model, the second large model, the third large model and the fourth large model mentioned above can be selected from different large models. Exemplarily, the first large model is used to analyze the demand description based on the first prompt information to generate at least one first object label. The second large model is used to screen at least one candidate label in the label library according to the demand description based on the second prompt information to obtain at least one second object label. The third large model is used to generate an explanatory description of each target object based on the third prompt information, according to the text features of the demand description and the descriptive features of the object description. The fourth large model is used to perform demand analysis on the initial description to obtain demand information.

[0161] Alternatively, the first large model, the second large model, the third large model and the fourth large model mentioned above can also be selected from the same large model, which can perform the first subtask, the second subtask, the third subtask and the fourth subtask. Exemplarily, the first subtask is used to analyze the requirement description based on the first prompt information to generate at least one first object label, the second subtask is used to screen at least one candidate label in the label library based on the second prompt information and the requirement description to obtain at least one second object label, the third subtask is used to generate an explanatory description of each target object based on the third prompt information and the text features of the requirement description and the descriptive features of the object description, and the fourth subtask is used to perform a requirement analysis on the initial description to obtain the requirement information.

[0162] Based on the above-mentioned object recommendation method based on a large model, the present invention also provides an object recommendation device based on a large model. Figure 7 The device is described in detail.

[0163] Figure 7 The block diagram of a large model-based object recommendation device according to an embodiment of the present disclosure is schematically shown.

[0164] like Figure 7 As shown, the large model-based object recommendation apparatus 700 may include a first determination module 710 and a second determination module 720 .

[0165] The first determination module 710 is configured to, in response to receiving a requirement description, determine at least one candidate object in an object library according to an object tag determined based on the requirement description.

[0166] The second determination module 720 is used to determine an object recommendation result based on the requirement description and at least one candidate object, wherein the object recommendation result includes at least one target object among the candidate objects, and at least one explanatory description, which characterizes the attributes of the target object and the mapping relationship between the requirement description and the target object.

[0167] According to an embodiment of the present disclosure, the first determination module may include a label generation submodule and a first determination submodule.

[0168] The label generation submodule is used to call the first large model based on the first prompt information, analyze the requirement description, and generate at least one first object label, wherein the first prompt information is used to instruct the first large model to generate a first object label corresponding to the requirement description.

[0169] The first determination submodule is used to determine at least one candidate object in the object library according to the first object tag.

[0170] According to an embodiment of the present disclosure, the object library includes at least two candidate object identifiers and an object description and description features of each candidate object identifier; the first determination submodule may include at least one of the following: a first matching unit, a second matching unit.

[0171] The first matching unit is used to perform text matching on at least two object descriptions according to the first object tag to obtain at least one candidate object.

[0172] The second matching unit is used to perform vector matching on at least two description features according to the first label feature of the first object label to obtain at least one candidate object.

[0173] According to an embodiment of the present disclosure, the object library includes at least two candidate object identifiers and respective descriptive features of each candidate object identifier; the first determination module may include a label screening submodule and a second determination submodule.

[0174] The label filtering submodule is used to call the second large model based on the second prompt information, and filter at least one candidate label in the label library according to the requirement description to obtain at least one second object label, wherein the second prompt information is used to instruct the second large model to select the second object label corresponding to the requirement description, and the label library also includes the second label features of each candidate label.

[0175] The second determination submodule is used to perform vector matching on at least two description features according to the second label feature to obtain at least one candidate object.

[0176] According to an embodiment of the present disclosure, the second determining module may include a third determining submodule and a fourth determining submodule.

[0177] The third determination submodule is used to determine the first n candidate objects among the at least one sorted candidate object as the target object, where n is a positive integer.

[0178] The fourth determination submodule is used to determine an object recommendation result according to the requirement description and the object description of at least one target object.

[0179] According to an embodiment of the present disclosure, the fourth determination submodule may include an explanatory description generating unit and an object recommendation result determining unit.

[0180] An explanatory description generating unit is used to call the third largest model based on the third prompt information, and generate an explanatory description for each target object according to the text features of the requirement description and the descriptive features of the object description, wherein the third prompt information is used to indicate that the third largest model captures the properties of the target object and the mapping relationship between the requirement description and the target object.

[0181] The object recommendation result determination unit is used to determine the object recommendation result according to the mapping relationship between the target object and the explanatory description.

[0182] According to an embodiment of the present disclosure, the object recommendation apparatus may further include an analyzing module, a guiding module and a third determining module.

[0183] The analysis module is used for performing demand analysis on the initial description using the fourth model in response to receiving the initial description to obtain demand information.

[0184] The guidance module is used to use the fourth model to conduct multiple rounds of interactive dialogues with users when the demand information does not meet the object recommendation conditions, so as to guide the users to supplement the initial description based on preset rules, wherein the object recommendation conditions are determined according to the attributes of each candidate object in the object library.

[0185] The third determination module is used to determine the requirement description according to the initial description and the dialogue information obtained through multiple rounds of interactive dialogues.

[0186] It should be noted that the object recommendation device part based on the big model in the embodiment of the present disclosure corresponds to the object recommendation method part based on the big model in the embodiment of the present disclosure. The description of the object recommendation device part based on the big model specifically refers to the object recommendation method part based on the big model, which will not be repeated here.

[0187] An embodiment of the present disclosure also provides an AI (Artificial Intelligence) agent configured to execute the large model-based object recommendation method as described above.

[0188] Figure 8 The structural block diagram of an artificial intelligence agent according to an embodiment of the present disclosure is schematically shown.

[0189] In the embodiments of the present disclosure, inspired by the von Neumann structure in modern computer theory, such as Figure 8 As shown, the AI ​​agent 800 may include five core modules: an input module 810 , a control module 820 , a storage module 830 , a calculation module 840 and an output module 850 .

[0190] The input module 810 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 800. The input module 810 is the primary link for the AI ​​agent 800 to interact with the outside world. It enables the AI ​​agent 800 to efficiently and accurately obtain the necessary "sensory" information from the outside world and respond to this information.

[0191] In an example, the input module 810 may input the requirement description or the initial description described above.

[0192] In the example, the control module 820 is the core support for the AI ​​agent 800 to handle complex tasks. The control module 820 can execute the large model-based object recommendation method described above.

[0193] In the example, the control module 820 will continuously interact with the storage module 830, the computing module 840 and / or the output module 850 during operation. However, it should be noted that in the embodiment of the present disclosure, the control module 820 acts as a single initiator to initiate communication with the storage module 830, the computing module 840 and / or the output module 850, and there is no communication coupling between the storage module 830, the computing module 840 and the output module 850.

[0194] In the example, the performance of the control 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 control module 820 may be designed to be highly configurable and extensible so as to cope with various types of tasks and requirements in real scenarios.

[0195] The storage module 830 may be responsible for memorizing information such as historical conversations, event flows, etc. The object library or tag library as described above may be included in the storage module 830 .

[0196] In the example, after receiving the requirement description, the AI ​​agent 800 may determine at least one candidate object in the object library based on the object label determined by the requirement description. The object library may be stored in the storage module 830. The AI ​​agent 800 may retrieve at least one candidate object matching the object label from the storage module 830 and feed it back to the control module 820. Then, the control module 820 may determine the object recommendation result using the at least one candidate object fed back and the requirement description, and pass the object recommendation result to the output module 850.

[0197] The operation module 840 may be regarded as a predefined tool library. For example, an operator for determining that the candidate object that is a preset number of the at least one candidate object after sorting is determined as the target object as described above may be included in the operation module 840 .

[0198] In the example, when the AI ​​agent 800 needs to determine at least one target object, the relevant operator can be called from the operation module 840 and fed back to the control module 820. Then, the control module 820 can use the fed-back operator to obtain the target object. The control module 820 can determine the object recommendation result based on the demand description and the object description of at least one target object, and pass the object recommendation result to the output module 850. It can be understood that although the large language model has excellent language understanding and generation capabilities, it is the same as a human being. Without the help of any tools, the tasks that can be solved are very limited. When the AI ​​agent 800 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.

[0199] In an example, the output module 850 may output the object recommendation result described above.

[0200] 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.

[0201] 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.

[0202] 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 method as described above.

[0203] 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.

[0204] According to an embodiment of the present disclosure, a computer program product includes a computer program, and when the computer program is executed by a processor, the computer program implements the method as described above.

[0205] Fig. 9A block diagram of an electronic device suitable for implementing a large model-based object recommendation method according to an embodiment of the present disclosure is schematically shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, 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.

[0206] 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.

[0207] 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.

[0208] 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 large model-based object recommendation method. For example, in some embodiments, the large model-based object recommendation method 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 large model-based object recommendation method described above may be performed. Alternatively, in other embodiments, the computing unit 901 may be configured to perform a large model-based object recommendation method in any other appropriate manner (e.g., by means of firmware).

[0209] 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), complex 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.

[0210] 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.

[0211] 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.

[0212] 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).

[0213] 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.

[0214] 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.

[0215] 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.

[0216] 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 method for object recommendation based on a large model, comprising: In response to receiving the requirement description, determining at least one candidate object in an object library according to an object tag determined based on the requirement description; as well as An object recommendation result is determined based on the requirement description and the at least one candidate object, wherein the object recommendation result includes at least one target object among the candidate objects, and at least one explanatory description, wherein the explanatory description characterizes the attributes of the target object and a mapping relationship between the requirement description and the target object.

2. The method according to claim 1, wherein: Determining at least one candidate object in the object library according to the object label determined based on the requirement description includes: Based on the first prompt information, the first large model is called to analyze the requirement description to generate at least one first object label, wherein the first prompt information is used to instruct the first large model to generate a first object label corresponding to the requirement description; and according to the first object label, the at least one candidate object is determined in the object library.

3. The method according to claim 2, wherein: The object library includes at least two candidate object identifiers and an object description and description features of each candidate object identifier; Determining the at least one candidate object in the object library according to the first object tag comprises at least one of the following: Performing text matching on the at least two object descriptions according to the first object tag to obtain the at least one candidate object; According to the first label feature of the first object label, vector matching is performed on the at least two description features to obtain the at least one candidate object.

4. The method according to claim 1, wherein: The object library includes at least two candidate object identifiers and respective descriptive features of each of the candidate object identifiers; Determining at least one candidate object in the object library according to the object label determined based on the requirement description includes: Based on the second prompt information, the second large model is called, and according to the requirement description, at least one candidate tag in the tag library is screened to obtain at least one second object tag, wherein the second prompt information is used to instruct the second large model to select a second object tag corresponding to the requirement description, and the tag library further includes a second tag feature of each candidate tag; and According to the second label feature, vector matching is performed on the at least two description features to obtain the at least one candidate object.

5. The method according to any one of claims 1 to 4, wherein: The determining of an object recommendation result according to the requirement description and the at least one candidate object comprises: Determine the first n candidate objects among the at least one sorted candidate object as the target object, where n is a positive integer; and The object recommendation result is determined according to the requirement description and the object description of each of the at least one target object.

6. The method according to claim 5, wherein: The determining the object recommendation result according to the requirement description and the object description of each of the at least one target object comprises: Invoking a third model based on the third prompt information, and generating an explanatory description of each target object according to the text features of the requirement description and the descriptive features of the object description, wherein the third prompt information is used to indicate that the third model captures the attributes of the target object and the mapping relationship between the requirement description and the target object; and The object recommendation result is determined according to the mapping relationship between the target object and the explanatory description.

7. The method according to any one of claims 1 to 4, further comprising, before determining at least one candidate object in an object library in response to receiving the requirement description and according to an object tag determined based on the requirement description: In response to receiving the initial description, using the fourth model to perform demand analysis on the initial description to obtain demand information; When the demand information does not meet the object recommendation conditions, the fourth model is used to conduct multiple rounds of interactive dialogues with the user to guide the user to supplement the initial description based on preset rules, wherein: The object recommendation condition is determined according to the attribute of each candidate object in the object library; as well as The requirement description is determined according to the initial description and dialogue information obtained through multiple rounds of interactive dialogues.

8. An object recommendation device based on a large model, comprising: A first determination module, configured to, in response to receiving a requirement description, determine at least one candidate object in an object library according to an object label determined based on the requirement description; as well as The second determination module is used to determine an object recommendation result based on the requirement description and the at least one candidate object, wherein the object recommendation result includes at least one target object among the candidate objects, and at least one explanatory description, wherein the explanatory description characterizes the attributes of the target object and the mapping relationship between the requirement description and the target object.

9. 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 at least one of the first large model, the second large model, the third large model, and the fourth large model based on the target task, and obtain output information by executing the method of any one of claims 1 to 7 by calling at least one of the first large model, the second large model, the third large model, and the fourth large model; as well as An output module is used to output the output information obtained by the processing module.

10. An electronic device, comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 7.

11. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

12. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images