Community operation method and device based on large model, equipment and medium

Abstract

One or more embodiments of the invention provide a community operation method and device based on a large model, equipment and a medium. According to the community operation method, inquiry information proposed by users in an offline tenant community can be obtained, semantic recognition is performed on the corresponding inquiry information based on a pre-trained main model, and the inquiry dimensions from which the corresponding inquiry information is proposed by the users are determined. And determining inquiry result information of the inquiry information under the corresponding inquiry dimension based on a pre-trained sub-model under each inquiry dimension. On the basis, the main model can take the inquiry result information determined by all the sub-models from different inquiry dimensions as reply information of the inquiry information and return the reply information to the corresponding community for the user to check.

Description

Technical Field

[0001] One or more embodiments of the present specification relate to the field of data processing technology, and in particular, to a community operation method, device, equipment and medium based on a large model. Background Art

[0002] With the rapid development of Internet technology and the increasing diversification of user consumption behaviors, brand merchants have gradually realized the importance of member and fan operations in market competition, and have stepped into the deep waters of member and fan operations. In this field, community operations, as an efficient and direct way of interaction, have become a key means to improve user stickiness, promote brand loyalty, and achieve precision marketing. Through the community, brand merchants can reach the target user group in real time, promote products, spread brands, collect user feedback, and communicate emotions, thereby improving user activity and conversion rate.

[0003] However, as the scale of communities continues to expand, the traditional manual community operation model faces many challenges. Specifically, most brands currently rely on full-time operators to manage the daily management of communities. Under this model, with the surge in the number of communities and the number of fans within the communities, it is difficult for a single or a small number of operators to effectively cover all communities, resulting in reduced operational efficiency and insufficient community activity, which in turn affects brand influence and user retention. In the long run, the conversion rate of communities declines, the loss of fans is serious, and some communities are even disbanded due to lack of effective maintenance, causing huge losses to brands. Summary of the invention

[0004] One or more embodiments of this specification provide a community operation method, device, equipment and medium based on a big model, which is used to simulate manual operation, quickly and efficiently provide timely services to users, reduce labor costs and maintain users' trust in the brand.

[0005] In a first aspect, one or more embodiments of the present specification provide a community operation method based on a large model, including: obtaining first inquiry information input by a first user in a first community; determining at least one inquiry dimension corresponding to the first inquiry information based on a main model; the main model is obtained by model training with first training data, and the first training data is historical inquiry information under different inquiry scenarios; determining first sub-models associated with the at least one inquiry dimension and an association order between all first sub-models; each first sub-model is obtained by model training with second training data under one inquiry dimension, and the second training data is historical question-answer pair information under the corresponding inquiry dimension, and different first sub-models are used to determine the inquiry result information of the first inquiry information under the corresponding inquiry dimension; according to the association order, obtaining the inquiry result information determined by each first sub-model in turn; and returning the inquiry result information determined by all first sub-models as first reply information to the first community.

[0006] In an optional embodiment, based on the main model, determining at least one query dimension corresponding to the first query information includes: inputting the first query information into the main model; and obtaining at least one query dimension determined by the main model after performing semantic recognition on the first query information.

[0007] In an optional embodiment, based on the main model, determining at least one inquiry dimension corresponding to the first inquiry information includes: performing a compliance check on the first inquiry information; after the check passes, determining at least one inquiry dimension corresponding to the first inquiry information based on the main model.

[0008] In an optional embodiment, determining the first sub-models respectively associated with the at least one inquiry dimension and the order of association between all the first sub-models includes: determining the first inquiry scenario corresponding to the first inquiry information according to the semantic recognition result of the main model for the first inquiry information, the first inquiry scenario corresponding to the at least one inquiry dimension; obtaining mapping information pre-configured for the first inquiry scenario; the mapping information includes the second sub-models respectively associated with all the inquiry dimensions corresponding to the first inquiry scenario, and the order of association between all the second sub-models, and all the second sub-models include all the first sub-models; according to the mapping information, determining the first sub-models respectively associated with the at least one inquiry dimension and the order of association between all the first sub-models.

[0009] In an optional embodiment, according to the association order, the query result information determined by each first sub-model is obtained in turn, including: according to the association order, determining the first first sub-model in the association order; inputting the first query information into the first first sub-model, and obtaining the query result information determined by the first first sub-model; according to the association order between the first first sub-model and other first sub-models, the query result information determined by the previous first sub-model is input as an input parameter into the next first sub-model in turn, and the query result information of each first sub-model is obtained in turn.

[0010] In an optional embodiment, the first query information is input into the first first sub-model, and the query result information determined by the first first sub-model is obtained, including: formatting the first query information according to the semantic recognition result of the first query information by the main model; inputting the formatted first query information as an input parameter into the first first sub-model; and obtaining the query result information determined by the first first sub-model.

[0011] In an optional embodiment, according to the order of association between the first sub-model and other first sub-models, the query result information determined by the previous first sub-model is input as an input parameter to the next first sub-model in sequence, and the query result information of each first sub-model is obtained in sequence, including: according to the order of association between the first sub-model and other first sub-models, the formatted processing result of the query result information determined by the previous first sub-model is input as an input parameter to the next first sub-model in sequence, and the query result information of each first sub-model is determined in sequence.

[0012] In an optional embodiment, all query result information determined by the first sub-model is returned to the first community as first reply information, including: performing compliance verification on all query result information determined by the first sub-model; after the verification passes, returning all query result information determined by the first sub-model as first reply information to the first community.

[0013] In an optional embodiment, the method further includes: obtaining second inquiry information input by a second user in the first community; determining the association between the first inquiry information and the second inquiry information; and obtaining second reply information based on the association and returning the second reply information to the first community.

[0014] On the second aspect, one or more embodiments of the present specification also provide a community operation device based on a large model, including: a first acquisition module, used to acquire first inquiry information input by a first user in a first community; a first determination module, used to determine at least one inquiry dimension corresponding to the first inquiry information based on a main model; the main model is obtained by model training with first training data, and the first training data is historical inquiry information under different inquiry scenarios; the second determination module is used to determine the first sub-models respectively associated with the at least one inquiry dimension and the association order between all first sub-models; each first sub-model is obtained by model training with second training data under one inquiry dimension, and the second training data is historical question-answer pair information under the corresponding inquiry dimension, and different first sub-models are used to determine the inquiry result information of the first inquiry information under the corresponding inquiry dimension; the second acquisition module is used to obtain the inquiry result information determined by each first sub-model in turn according to the association order; the output module is used to return the inquiry result information determined by all first sub-models as the first reply information to the first community.

[0015] In an optional embodiment, the first determination module determines at least one query dimension corresponding to the first query information based on the main model, and is used to: input the first query information into the main model; and obtain at least one query dimension determined by the main model after semantic recognition of the first query information.

[0016] In an optional embodiment, the first determination module determines at least one inquiry dimension corresponding to the first inquiry information based on the main model, and is used to: perform compliance verification on the first inquiry information; after the verification passes, determine at least one inquiry dimension corresponding to the first inquiry information based on the main model.

[0017] In an optional embodiment, the second determination module determines the first sub-models respectively associated with the at least one inquiry dimension and the order of association between all the first sub-models, and is used to: determine the first inquiry scenario corresponding to the first inquiry information according to the semantic recognition result of the main model for the first inquiry information, the first inquiry scenario corresponding to the at least one inquiry dimension; obtain mapping information pre-configured for the first inquiry scenario; the mapping information includes the second sub-models respectively associated with all the inquiry dimensions corresponding to the first inquiry scenario, and the order of association between all the second sub-models, and all the second sub-models include all the first sub-models; according to the mapping information, determine the first sub-models respectively associated with the at least one inquiry dimension and the order of association between all the first sub-models.

[0018] In an optional embodiment, the second acquisition module acquires the query result information determined by each first sub-model in sequence according to the association order, and is used to: determine the first first sub-model in the association order according to the association order; input the first query information into the first first sub-model, and acquire the query result information determined by the first first sub-model; according to the association order between the first first sub-model and other first sub-models, sequentially input the query result information determined by the previous first sub-model as an input parameter into the next first sub-model, and acquire the query result information of each first sub-model in sequence.

[0019] In an optional embodiment, the second acquisition module inputs the first query information into the first first sub-model and obtains the query result information determined by the first first sub-model, and is used to: format the first query information according to the semantic recognition result of the first query information by the main model; input the formatted first query information as an input parameter into the first first sub-model; and obtain the query result information determined by the first first sub-model.

[0020] In an optional embodiment, the second acquisition module sequentially inputs the query result information determined by the previous first submodel as an input parameter into the next first submodel according to the order of association between the first first submodel and other first submodels, and sequentially acquires the query result information of each first submodel, and is used for: sequentially inputting the formatted processing result of the query result information determined by the previous first submodel as an input parameter into the next first submodel according to the order of association between the first first submodel and other first submodels, and sequentially determining the query result information of each first submodel.

[0021] In an optional embodiment, the output module returns all the inquiry result information determined by the first sub-model as the first reply information to the first community, and is used to: perform compliance verification on all the inquiry result information determined by the first sub-model; after the verification is passed, return all the inquiry result information determined by the first sub-model as the first reply information to the first community.

[0022] In an optional embodiment, the first acquisition module is further used to: obtain second inquiry information input by a second user in the first community; determine the association relationship between the first inquiry information and the second inquiry information; and obtain second reply information based on the association relationship and return the second reply information to the first community.

[0023] In a third aspect, one or more embodiments of the present specification further provide a computer device, comprising: a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, they are used to implement the big model-based community operation method of the first aspect.

[0024] In a fourth aspect, one or more embodiments of the present specification further provide a computer-readable storage medium, wherein the computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed, the community operation method based on a large model of the first aspect described above is implemented.

[0025] In summary, based on the above, the community operation method based on the large model provided in one or more embodiments of this specification can obtain the inquiry information raised by users in the offline tenant community, perform semantic recognition on the corresponding inquiry information based on the pre-trained main model, and determine from which inquiry dimensions the user raises the corresponding inquiry information. Furthermore, the inquiry result information of the above inquiry information under the corresponding inquiry dimension is determined based on the sub-model pre-trained under each inquiry dimension. Based on this, the main model can return the inquiry result information determined by all sub-models from different inquiry dimensions as the reply information of the above inquiry information to the corresponding community for the user to view. In this method, various inquiry information raised by users in the community is automatically identified based on each model and feedback is given in a timely manner. It can not only simulate the manual operation method and reduce labor costs, but also provide users with timely services quickly and efficiently, which helps to maintain users' trust in the brand, improve user retention rate, and provide a basis for user conversion of other derivative transactions. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of one or more embodiments of the present specification, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of one or more embodiments of the present specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0027] Figure 1a A system architecture diagram of a community operation system based on a large model provided for one or more embodiments of this specification;

[0028] Figure 1b A system architecture diagram of another community operation system based on a large model provided for one or more embodiments of this specification;

[0029] Figure 1c A system architecture diagram of another community operation system based on a large model provided for one or more embodiments of this specification;

[0030] Figure 2a A flowchart of a community operation method based on a large model provided for one or more embodiments of this specification;

[0031] Figure 2b A flowchart of another community operation method based on a large model provided for one or more embodiments of this specification;

[0032] Figure 3 A structural block diagram of a community operation device based on a large model provided for one or more embodiments of this specification;

[0033] Figure 4 A structural block diagram of a computer device provided for one or more embodiments of this specification. DETAILED DESCRIPTION

[0034] One or more embodiments of the present invention are further described in detail below through the accompanying drawings and examples. Through these descriptions, the features and advantages of one or more embodiments of the present invention will become clearer and more specific.

[0035] The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustration." Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments. Although various aspects of the embodiments are shown in the drawings, the drawings are not necessarily drawn to scale unless otherwise noted.

[0036] In addition, the technical features involved in different implementations of one or more embodiments of this specification described below can be combined with each other as long as they do not conflict with each other.

[0037] To facilitate understanding, the application scenarios of the technical solutions provided by one or more embodiments of this specification are first described below.

[0038] In terms of user operations, brands usually maintain merchant relationships through communities, and provide targeted services to users in a timely manner by regularly tracking user trends within the community, in order to improve brand user stickiness and user trust. However, as the number of users within the community continues to increase, the difficulty of community operations is also increasing. It is difficult for operators to provide timely feedback on various issues raised by many users, which will not only bring users a bad service experience, but in the long run, it will also be difficult to guarantee user retention.

[0039] To this end, one or more embodiments of this specification provide a community operation method based on a large model to automatically identify various questions raised by users in the community and provide timely feedback, thereby improving operational efficiency and user experience while reducing the labor cost of community operation.

[0040] In one or more embodiments of the present specification, the execution subject of the above method is not limited. The execution subject of the above method may be different depending on the application scenario. Based on this, before describing each step in the above method, the application scenario of the above method is first described in several optional ways.

[0041] Figure 1a is an optional system architecture diagram of a community operation system based on a large model. In this embodiment, the above method is applied to a centralized system architecture, such as Figure 1a As shown, the above method is executed by the cloud server, which can communicate with the local server of the brand through an interface to obtain basic information of the brand, user information of the local community, community messages and other information from the local server of the brand; then, based on the obtained information, semantic recognition, problem confirmation and feedback results are performed on questions raised by users in the community. It can be understood that the cloud server is the server corresponding to the service platform used to provide operation services to each brand, such as Figure 1a As shown, in this embodiment, the program files of the above method and multiple models relied on during the execution process are deployed in the service platform. Each model and the above method can be packaged into a service product and provided to each brand. For example, each brand can be a tenant who rents the above service product from the service platform.

[0042] Figure 1b This is another optional system architecture diagram of a community operation system based on a large model, such as Figure 1b As shown, in this embodiment, the service platform provides the program files corresponding to the above method to each brand in an open source form, and deploys each model involved in the execution of the above method on a cloud server. Based on this, each brand can deploy the corresponding program files on a local server, and the local server executes the above method. Figure 1b As shown, in an embodiment, the cloud server can communicate with the brand's local server through an interface to obtain the brand's basic information, local community user information, community messages and other information from the brand's local server; then, through each model based on the acquired information, semantic recognition, question confirmation and feedback results are performed on questions raised by users in the community.

[0043] Figure 1c This is another optional system architecture diagram of a community operation system based on a large model, such as Figure 1c As shown, in this embodiment, the service platform provides the program files corresponding to the above method and the models involved in the execution of the above method to each brand in an open source form. Based on this, each brand can deploy the corresponding program files and each model on a local server, and the local server executes the above method. Figure 1cAs shown, during the execution of the above method, basic information of the brand, user information of the local community, community messages and other information can be obtained from the local server; then, based on the obtained information, each model is used to perform semantic recognition, question confirmation and feedback on questions raised by users in the community.

[0044] exist Figure 1b and Figure 1c In the example shown, for the program files and various models provided to the brand in open source form, each brand can make appropriate modifications to the corresponding program code according to its own needs before deployment to match the operational needs of its own brand. Correspondingly, in another optional method, the service platform also provides each brand with the program files corresponding to the above method, or provides the program files and various models corresponding to the above method at the same time. However, it is required to deploy the above program files and various models in the local server of each brand in a privatized manner, that is, each brand cannot modify the program code corresponding to the above method and various models. In this embodiment, please refer to the corresponding system architecture and specific implementation method. Figure 1b and Figure 1c The description of the examples shown will not be repeated here.

[0045] It should be noted that the specific form of the program files of the above method and the deployment method of each model involved in the execution can be determined according to the specific needs of the service platform or the agreement between the service platform and the brands, and no specific limitation is made.

[0046] It should be further explained that the above-mentioned application scenarios are only exemplary and are not limited to them in actual applications. As long as the above-mentioned method is applicable to the scenario, it can be regarded as within the protection scope of one or more embodiments of this specification and will not be elaborated here.

[0047] Below, the community operation method based on the big model provided by one or more embodiments of this specification is described in conjunction with the accompanying drawings.

[0048] Figure 2a A flowchart of a community operation method based on a large model is provided for one or more embodiments of this specification, such as Figure 2a As shown, the method includes:

[0049] S102, obtaining first inquiry information input by a first user in a first community;

[0050] S104: Determine at least one inquiry dimension corresponding to the first inquiry information based on a main model, where the main model is obtained by training the first training data, and the first training data is historical inquiry information in different inquiry scenarios;

[0051] S106, determining first sub-models respectively associated with at least one query dimension and an association order between all first sub-models; wherein each first sub-model is obtained by model training using second training data under one query dimension, the second training data is historical question-answer pair information under the corresponding query dimension, and different first sub-models are used to determine query result information of the first query information under the corresponding query dimension;

[0052] S108, sequentially obtaining query result information determined by each first sub-model according to the association order;

[0053] S110: Return all inquiry result information determined by the first sub-model as first reply information to the first community.

[0054] Before describing each step of the above method, the functions of each model involved in the execution of the above method are first described.

[0055] In one or more embodiments of the present specification, a main model is provided, and the main model is obtained by model training using the first training data, wherein the first training data refers to historical inquiry information under different inquiry scenarios, and each inquiry scenario corresponds to at least one inquiry dimension. The inquiry dimensions corresponding to different inquiry scenarios may be the same or different, and are not limited here. Among them, the inquiry scenario refers to the application scenario type of each brand, for example, including but not limited to catering, clothing, furniture and appliances, cultural and creative products, etc. The inquiry dimension corresponding to any inquiry scenario can be understood as the angle of various questions raised by users in the community for the specific object in the corresponding inquiry scenario.

[0056] For example, taking the catering inquiry scenario as an example, users in the corresponding community can ask questions like "Does A have mushroom flavor?" from the inquiry dimension of taste, ask questions like "Can different types of coupons be used in combination?" from the inquiry dimension of discount activities, ask questions like "Is store B still open?" from the inquiry dimension of merchants, ask questions like "What if I find foreign objects in C and no one is responsible?" from the inquiry dimension of complaints or suggestions, etc. Correspondingly, for communities in other inquiry scenarios, users therein can also ask different questions about specific objects in the corresponding inquiry scenarios from different inquiry dimensions, which will not be elaborated here.

[0057] In one or more embodiments of the present specification, for each inquiry dimension, a corresponding sub-model is pre-trained, wherein each sub-model is obtained by model training using the second training data under one inquiry dimension, the second training data is the historical question-answer pair information under the corresponding inquiry dimension, and different sub-models are used to determine the inquiry result information corresponding to the corresponding inquiry dimension. Based on this, each sub-model can perform semantic recognition on various questions raised by the user based on the historical question-answer pair information under the corresponding inquiry dimension, and determine the inquiry result information under the corresponding inquiry dimension.

[0058] Based on the above, in order to achieve mutual scheduling between the main model and each sub-model, in one or more embodiments of the present specification, corresponding mapping information is pre-configured for each inquiry scenario, wherein the mapping information of each inquiry scenario may include the sub-models associated with all inquiry dimensions corresponding to the corresponding inquiry scenario and the association order between all sub-models. Of course, it can also include the association relationship between all sub-models and other sub-models used to perform general functions. Based on this, according to the association order, orderly scheduling between the main model and each sub-model can be achieved, so as to determine the corresponding inquiry result information from different inquiry dimensions in turn for various questions raised by the user.

[0059] Optionally, the embodiments of this specification provide a processing sub-model for processing complaint or suggestion information. The processing sub-model can be regarded as a sub-model for performing general functions. If the main model recognizes that the content of the first inquiry information is a complaint or suggestion, etc., it can be directly handed over to the processing sub-model to determine the corresponding solution and return it to the request end. For example, the first inquiry information is a complaint that a certain product has a quality problem. After the processing sub-model identifies the specific complaint information in the first inquiry information, it can determine the detailed information of the product through the product sub-model, determine the user information corresponding to the complaining user through the user model, and determine the merchant information corresponding to the product through the merchant sub-model; then, notify the corresponding merchant of the above user information and corresponding complaint information for processing, or, the queried merchant information can also be returned to the request end for the user to contact the corresponding merchant for processing. Of course, the sub-model used to perform general functions is not limited to the above-mentioned processing sub-model. Accordingly, the operations performed by the processing sub-model in the above-mentioned embodiment are only exemplary and are not limited to this.

[0060] Based on the above, the following Figure 1a Taking the centralized deployment method shown as an example, the community operation method based on the big model proposed in one or more embodiments of this specification is explained. Accordingly, the method can be executed by a cloud server.

[0061] It should be noted that in one or more embodiments of the present specification, there is no limitation on the execution subject of each sub-model scheduled in the above-mentioned association order. Optionally, after the main model determines the inquiry scenario corresponding to the question raised by the user, it can also determine the sub-models associated with the corresponding inquiry scenario and the association order between the sub-models based on the pre-configured mapping information, and then schedule the sub-models in an orderly manner according to the association order; or, after the main model determines the inquiry scenario corresponding to the question raised by the user, it can output it as a recognition result to the cloud server, and the cloud server determines the sub-models associated with the corresponding inquiry scenario and the association order between the sub-models based on the pre-configured mapping information, and then schedule the sub-models in an orderly manner according to the association order.

[0062] In one or more embodiments of the present specification, an example is taken in which a main model determines the association order between sub-models and schedules each sub-model.

[0063] For the purpose of distinction, in one or more embodiments of the present specification, any community is referred to as the first community, and different users in the first community are distinguished as first users and second users; different questions input by users in the first community are distinguished as first inquiry information and second inquiry information; the tenant corresponding to the first community is referred to as the first tenant, and the inquiry scenario corresponding to the first tenant is referred to as the first inquiry scenario; the sub-model associated with all inquiry dimensions corresponding to the first inquiry information input by the first user is referred to as the first sub-model, and the sub-model associated with all inquiry dimensions corresponding to the first inquiry scenario is referred to as the second sub-model, that is, the second sub-model includes the first sub-model; the request information sent by the local server of the first tenant to the cloud server each time is distinguished as the first group message and the second group message, and so on, without further elaboration.

[0064] Based on this, when the first user enters the first inquiry information in the first community, the local server of the corresponding first tenant can initiate an operation service request to the cloud server through the communication interface with the cloud server, and send the first group message to the cloud server. Among them, the first group message includes the first inquiry information, and of course, it can also include the basic information of the first tenant, necessary information required for communication, etc. The specific content can be determined according to actual needs and is not limited here.

[0065] Furthermore, after receiving the first group message, the cloud server can perform semantic recognition on the first inquiry information based on the main model to determine the first inquiry scenario and at least one inquiry dimension corresponding to the first inquiry information. Further, based on the mapping information corresponding to the first inquiry scenario, the first sub-model associated with each inquiry dimension and the association order between all first sub-models can be determined. Then, based on the association order, the inquiry result information determined by each first sub-model under the corresponding inquiry dimension is obtained in turn, and the inquiry result information determined by all first sub-models is returned to the first community as the first reply information.

[0066] In one or more embodiments of the present specification, the specific manner in which the cloud server determines at least one query dimension based on the main model is not limited.

[0067] In an optional embodiment, the cloud server may first input the first query information into the main model. Since the main model is trained based on historical query information in different query scenarios, these historical query information are raised by historical users from different query dimensions for specific objects in different query scenarios. Therefore, the main model may perform semantic recognition on the first query information based on the historical query information in different query scenarios, and determine that the first query information corresponds to the first query scenario, and at least one query dimension corresponding to the first query scenario.

[0068] In another optional embodiment, the cloud server may also perform a compliance check on the first query information before inputting the first query information into the main model to determine whether the request meets the requirements. Further, if the check passes, the first query information is input into the main model after the check passes to determine at least one query dimension corresponding to the first query information based on the main model; accordingly, if the check fails, a prompt message is returned to the requesting end to inform it that the request initiated does not meet the requirements. Among them, the specific content of the compliance check is not limited. Optionally, it can be checked whether the first query information involves negative information that violates laws and regulations, and it can also be checked whether the first query information includes pre-specified sensitive information, for example, information related to personal privacy such as user identity, address, contact information, ID number, etc.

[0069] In another optional embodiment, in addition to verifying the compliance of the content of the first inquiry information, the type of the first inquiry information may also be verified to determine whether the first inquiry information is invalid information that is irrelevant to the application scenario of the first community, for example, verifying whether the first inquiry information is a third-party advertisement, spam, a third-party access link, etc. Optionally, in order to maintain the community environment, if the first inquiry information is determined to be invalid information, the requesting end may be instructed to withdraw it directly. Further optionally, a corresponding input count threshold may be set for each user to determine the number of times each user has entered invalid information in the first community. Based on this, if it is determined that the number of times the first user has entered invalid information has reached a preset input count threshold, the requesting end may be instructed to perform operations such as banning, kicking out of the community, and blacklisting on the first user, and the specific method is not limited.

[0070] It should be noted that, in the above example, the operations of checking the compliance of the content of the first inquiry information and checking the invalidity of the type of the first inquiry information can be performed by the main model or by other sub-models. Further optionally, the operations of checking the compliance of the content of the first inquiry information and checking the invalidity of the type of the first inquiry information can be performed by one sub-model or by different sub-models. For example, the compliance check of the content of the first inquiry information can be performed by the checking sub-model, and the operation of checking the invalidity of the type of the first inquiry information can be performed by the shielding sub-model. Of course, the above is only an exemplary description and is not limited to this.

[0071] In one or more embodiments of the present specification, since each first sub-model is trained based on historical question-answer pair information under a query dimension, each first sub-model can perform semantic recognition on the first query information raised by the first user based on the historical question-answer pair information under the corresponding query dimension, and determine the query result information that matches the first query information under the corresponding query dimension.

[0072] Based on this, after the main model determines the first sub-models associated with at least one query dimension corresponding to the first query information based on the pre-configured mapping information, as well as the association order between all the first sub-models, the first first sub-model can be determined according to the association order. Further, the first query information is input into the first first sub-model and the query result information determined by the first sub-model is obtained. Further, according to the association order between the first first sub-model and other first sub-models in the above association order, the query result information determined by the previous first sub-model is input as an input parameter to the next first sub-model in turn, until all the first sub-models in the above association order are executed, and the query result information of each first sub-model is obtained in turn.

[0073] In one or more embodiments of the present specification, the specific manner in which the main model inputs the first query information into each first sub-model and obtains the query result information from each first sub-model is not limited. Optionally, before the main model inputs the first query information into the first first sub-model in the above-mentioned association order, it can format the first query information according to the semantic recognition result of the first query information. Furthermore, the formatted first query information is input as an input parameter into the first first sub-model in the above-mentioned association order, and then the query result information determined by the first sub-model is obtained.

[0074] Correspondingly, for the query result information determined by multiple first sub-models, the query result information determined by the previous first sub-model can be formatted in sequence according to the association order between the multiple first sub-models, and the formatted query result information can be input as an input parameter to the next first sub-model until all the first sub-models in the above association order are executed, and the query result information of each first sub-model is determined in sequence.

[0075] In one or more embodiments of the present specification, the execution entity of the formatting operation on the query result information determined by each first sub-model is not limited. Optionally, the main model may perform the formatting operation on the query results of all first sub-models, or each first sub-model may perform the formatting operation on the query result information determined by it.

[0076] Optionally, when executed by the main model, each first sub-model only needs to determine the query result information corresponding to the first query information under the corresponding query dimension and output it to the main model, and then the main model formats the query result information output by each first sub-model. Accordingly, when executed by each first sub-model, each first sub-model first determines the query result information corresponding to the first query information under the corresponding query dimension, and then formats the corresponding query result information, and outputs the formatted processing result to the main model.

[0077] It should be noted that the specific method can be selected according to actual needs. In one or more embodiments of this specification, the formatting operation of the main model on all the query result information determined by the first sub-model is taken as an example for explanation.

[0078] Below, taking the catering inquiry scenario as an example, the process of the main model performing the above operations is exemplified.

[0079] Assume that the user ID corresponding to the first user is "2088XYZW", and the first inquiry information entered in the first community in text format is: "How many flavors of French fries are there?" When the local server of the first tenant initiates an operation service request to the cloud server, it can send the first inquiry information and user ID entered by the first user as the first group message to the cloud server. Based on this, after receiving the first group message, the cloud server can first perform a compliance check on the first group message to determine whether it includes any violation information. Further, if it is determined that the first group message includes preset violation information, a prompt message is returned to inform that the request initiated does not meet the requirements. Based on this, the local server of the first tenant can resend a second group message that meets the requirements according to the prompt message to re-initiate the operation service request. Accordingly, if it is determined that the first group message is compliant, it is input into the main model for subsequent processing by the main model.

[0080] Optionally, after receiving the first group message, the main model can identify the user identifier corresponding to the first user from the first group message, and identify the first query information input by the first user. Further, semantic recognition is performed on the first query information based on historical query information in different query scenarios to determine at least one query dimension involved in the first query information. In addition, the main model can also identify other necessary information from the first query information, and use the identified information as the recognition result. For example, when the main model identifies the keyword "taste" from the first query information, it can determine that the query dimension corresponding to the first query information is "flavor type"; further, when the main model identifies the keyword "French fries" from the first query information, it can determine that the product type corresponding to the first query information is "French fries".

[0081] Based on this, the main model can format the above recognition results. Assume that after the main model formatting, the recognition results in the following format are obtained:

[0082]

[0083] Among them, "GoodsType" is used to identify the type of goods, and the corresponding "French fries" thereafter represents the content of the type of goods involved in the first inquiry information; "UserId" is used to identify the user ID, and the corresponding "2088XYZW" thereafter is the content of the user ID corresponding to the first user who inputs the first inquiry information in the first community; "InquiryType" is used to identify the inquiry dimension, and the corresponding "flavor type" thereafter represents the content of the inquiry dimension corresponding to the first inquiry information. It can be seen that after formatting, the first inquiry information is converted from a spoken text format to a string in the form of a "key-value pair", which is easier for machines to recognize and process.

[0084] It should be noted that the above recognition result adopts the string format in the form of "key-value pair" for exemplary purposes only. Optionally, the above string format may be a lightweight data exchange format (JavaScript Object Notation, JSON), but is not limited thereto in practical applications. According to specific processing requirements, the recognition result after formatting may also be in other formats. For example, the format of the recognition result of the first query information after formatting may also be Extensible Markup Language (XML), a binary file, or an ordinary text file, etc. The specific format is not limited here. In the following embodiments, the JSON format is used as an example for exemplary description.

[0085] Assume that, according to the mapping information pre-configured for the catering inquiry scenario, the first sub-model directly associated with the inquiry dimension of "flavor type" is determined to be the product sub-model, and the first sub-models sequentially associated therewith are the mapping sub-model and the promotion sub-model respectively; wherein, the product sub-model can query the product information corresponding to the inquiry dimension of "flavor type"; the mapping sub-model can query the corresponding user information in combination with the user identifier recognized by the main model, and determine the mapping information between the corresponding user information and the above-mentioned product information based on the inquiry dimension of "flavor type" recognized by the main model and the product information queried by the product sub-model; the promotion sub-model can recommend to the user based on the mapping information the promotion activities that can be participated in, the coupons that can be used, the related products that can be purchased, etc., without limitation to the specific content.

[0086] It should be noted that one or more embodiments of this specification do not limit the specific content of product information and user information, as well as the specific content of mapping information determined by the mapping sub-model. In specific application scenarios, user information and product information may be different, and accordingly, the mapping information between the two may also be different. For example, user information may have differences in region, level, personal information, etc., and product information may also have differences in product type, flavor type, sales region, sales volume, recommendation index, inventory, etc. Therefore, the mapping information determined by the mapping sub-model based on different user information and different product information is also different.

[0087] In one or more examples of this specification, the information content determined about each first sub-model is only an illustrative description, which is intended to illustrate the implementation principle of the corresponding example and does not constitute a specific limitation.

[0088] For the sake of convenience, the content obtained after the main model formats the recognition result of the first query information is called parameter 1. Based on this, the main model can use parameter 1 as the input parameter of the product sub-model and input it into the product sub-model, so that the product sub-model can perform subsequent processing according to parameter 1.

[0089] Furthermore, after receiving parameter 1 input from the main model, the product sub-model can identify that the value of "GoodsType" is "French fries" and that the value of "InquiryType" is "flavor type". Based on this, the product sub-model can query the quantity and name of the "flavor type" corresponding to the product type of "French fries" from the product information library as the inquiry result information. Further optionally, in addition to the quantity and name directly queried, the product sub-model can also query other relevant information. For example, among the names of the queried "flavor types", the product sub-model can also query the name of the "flavor type" with the highest sales volume. Of course, it is not limited to this, and other relevant information can also be queried, which can be determined according to actual needs.

[0090] Assume that, after the above query, the product sub-model determines that there are three "flavor types" of the product type "French fries", and the corresponding names are "cheese flavor", "sea salt flavor" and "black pepper flavor", among which "cheese flavor" has the highest sales volume. Based on this, the product sub-model can output the queried information as query result information to the main model. After receiving the query result information output by the product sub-model, the main model can perform formatting processing in the same way as the first query information to convert the query result information output by the product sub-model into JSON format.

[0091] Assume that the query result information output by the product sub-model is formatted to obtain the following content:

[0092]

[0093] For the sake of convenience, the content obtained after the above formatting process is called parameter 2. Based on this, the main model can determine the input parameters of the mapping sub-model according to parameters 1 and 2, and input the input parameters into the mapping sub-model to schedule the mapping sub-model to perform subsequent operations. Optionally, when determining the input parameters of the mapping sub-model, the main model can first determine at least one indication parameter based on the query dimension of "flavor type" identified from the first query information, wherein each indication parameter is used to instruct the mapping sub-model to perform a corresponding operation.

[0094] In this embodiment, the number and specific content of the indication parameters corresponding to the inquiry dimension of "flavor type" are not limited. Assume that the main model determines that the inquiry dimension of "flavor type" corresponds to 3 indication parameters, which are respectively used to indicate the mapping sub-model to determine the "activity level", "flavor preference" and "recommendation acceptance" of the first user. Optionally, "activity level" refers to the number of goods purchased by the first user the most, "flavor preference" refers to the name of the "flavor type" with the largest number of foods purchased by the first user, and "recommendation acceptance" refers to the number of goods purchased by the first user from the goods recommended by the brand or other users. The corresponding content of this type of goods in the "purchase source" type can be marked with "recommendation".

[0095] Furthermore, this embodiment does not limit the method for determining the above three indication parameters. Optionally, based on the historical purchase record information of the first user, the main model can determine the "activity level" according to the quantity of purchases of the product type "French fries" or all foods, determine the "taste preference" according to the "flavor type" with the largest quantity purchased of the product type "French fries" or all foods, and determine the "recommendation acceptance" according to the number of "purchase sources" of "recommendations" for the product type "French fries" or all foods.

[0096] Of course, the above explanation of each indication parameter and the method for determining the content of each indication parameter are only exemplary illustrations and are not limited to this in practical applications. For each indication parameter, the specific content of each parameter can also be determined from other perspectives; accordingly, for different inquiry scenarios, the content of the corresponding indication parameter and the method for determining the corresponding content may also be different, which will not be described in detail here.

[0097] Further, after determining at least one indication parameter corresponding to the query dimension of "flavor type", the main model can determine the input parameters of the mapping sub-model based on parameter 1, parameter 2 and at least one indication parameter.

[0098] In this embodiment, the specific method for the main model to determine the input parameters of the mapping sub-model is not limited. Optionally, the main model can directly determine parameter 1, parameter 2 and at least one indication parameter as the input parameters of the mapping sub-model; or, a parameter 3 can be determined based on parameter 1, parameter 2 and at least one indication parameter, and the parameter 3 can be used as the input parameter of the mapping sub-model. The specific method can be determined according to the input form of the mapping sub-model. In this embodiment, parameter 3 is used as the input parameter of the mapping sub-model as an example. In order to simplify the processing process, the main model can directly splice parameter 1, parameter 2 and at least one indication parameter into JSON format, and determine the spliced ​​content as parameter 3 and input it into the mapping sub-model.

[0099] In this embodiment, "ActivityLevel" is used to identify "activity level", "FoodPrefer" is used to identify "taste preference", and "AcceptLevel" is used to identify "recommendation acceptance level". Assume that after splicing parameter 1, parameter 2 and the above three indication parameters, the obtained parameter 3 is as follows:

[0100]

[0101] Based on this, after the main model inputs parameter 3 into the mapping sub-model, the mapping sub-model can identify the value of "UserId" from parameter 3 as "2088XYZW", and when it is identified that the values ​​of "ActivityLevel", "FoodPrefer" and "AcceptLevel" are "?" respectively, it determines that the operation to be performed is to determine the values ​​of the three parameters. Based on this, the mapping sub-model can query the historical purchase record information of the first user according to the value "2088XYZW" of the above user identifier, and determine the values ​​of the above three parameters therefrom according to the above determination method. For example, taking the query of the historical purchase record information of the first user who purchased the commodity type "French fries" as an example, it is assumed that after the mapping sub-model queries the purchase record information related to the commodity type "French fries" in the historical purchase record information of the first user according to "2088XYZW", it determines that in the historical purchase record information of the first user, the number of purchases of the commodity type "French fries" by the first user is 7, the name of the "flavor type" with the largest purchase quantity, and the number of purchases of the commodity type "French fries" with the "purchase source" of "recommended" by the first user is 8. Based on this, the mapping sub-model can output the query result information as the query result information, so that the main model can perform subsequent processing based on the query result information.

[0102] It should be noted that when the mapping sub-model queries the historical purchase record information of the first user, it can query all the historical purchase record information, or it can query the historical purchase record information within a specified historical period, for example, the past month, three months, six months, etc., and the specific time length is not limited.

[0103] Furthermore, after receiving the query result information output by the mapping sub-model, the main model can format the query result information in the same way, and concatenate the formatted content with parameter 3 to obtain parameter 4, which is used as the input parameter of the promotion sub-model and input into the promotion sub-model to schedule the promotion sub-model to perform subsequent operations.

[0104] Assume that after the above formatting and concatenation, parameter 4 is as follows:

[0105]

[0106] Based on this, after receiving parameter 4, the promotion sub-model can identify various parameter types and corresponding parameter values. For example, based on the identification of parameter 4, the promotion sub-model can determine that the user ID of the first user is "2088XYZW", the type of product inquired by the first user is "French fries", and the inquiry dimension of the first inquiry information input by the first user is "flavor type"; there are currently three flavors of the "French fries" product type on sale, namely "cheese flavor", "sea salt flavor" and "black pepper flavor", among which the "flavor type" with the highest sales volume is "cheese flavor"; in the historical purchase record information of the first user, the "activity level" of purchasing the "French fries" product type is 7, the "flavor preference" is mostly "sweet", and the "recommendation acceptance" is 8. Furthermore, based on the identification results, the promotion sub-model can determine the recommended information such as the preferential activities that can be participated in, the coupons that can be used, and the related products that can be purchased for the first user.

[0107] In this embodiment, the specific method for determining the recommendation information by the recommendation sub-model is not limited. Optionally, based on the above identification results, it can be preferentially determined whether the first user's account includes a coupon corresponding to the "French fries" that can be purchased in the above three "flavor types"; if it is determined that the first user's account has a corresponding coupon, the corresponding coupon and the access address of the specific product of the "French fries" product type that can be purchased are used as recommendation information. Correspondingly, if the first user's account does not have a coupon, it is queried whether there is a corresponding discount activity for the "French fries" product type based on the above three "flavor types"; if there is a corresponding discount activity, it is further determined whether the first user is eligible to participate in the discount activity. If it is determined that the first user is eligible to participate, the access address of the corresponding activity is used as recommendation information.

[0108] Alternatively, when it is determined that there is a corresponding coupon in the first user's account, it can also be further determined whether there is a discount activity for the product type "French fries" and whether the first user is eligible to participate; if it is determined that the first user is eligible to participate, the corresponding coupon and the access address of the specific product type "French fries" that can be purchased, as well as the access address of the corresponding activity can be used as recommendation information. Alternatively, it is also possible to directly determine from the sales dimension whether the first user has a coupon corresponding to the purchase of "French fries" with "cheese flavor", determine whether there is a corresponding discount activity for "French fries" with "cheese flavor", and determine whether the first user is eligible to participate in the corresponding discount activity, etc.; then, when the query has one or more of the above information, the queried information is used as recommendation information. Of course, the above method is only an exemplary description, and is not limited to this in actual application, so it will not be described in detail here.

[0109] Assume that the recommendation sub-model finds out that there is a "Coupon A" in the first user's account based on the user ID "2088XYZW", which can be used to purchase "French Fries Q" with a "cheese flavor". Further, the corresponding access address is determined to be "http: / / XXXXXX" based on "French Fries Q". Based on this, the promotion sub-model can output the above information as query result information to the main model. After receiving the query result information output by the promotion sub-model, the main model can format it and concatenate it with the above parameter 4 to obtain parameter 5. Then, parameter 5 is returned to the first community as the first reply information, so that the first user can perform a purchase operation based on the first reply information.

[0110] Assume that in this embodiment, "Discount" is used to identify the recommended information, "Purchasable" is used to identify the purchasable goods, and "Address" is used to identify the access address. Based on this, after the query result information output by the promotion sub-model is formatted and concatenated with parameter 4, parameter 5 is obtained as follows:

[0111]

[0112] It should be noted that, in actual applications, when the main model inputs corresponding parameters to each first sub-model, it can only input the necessary information of each first sub-model as input parameters to the corresponding first sub-model. In the embodiment of this specification, inputting all previously acquired information into the corresponding first sub-model is only an exemplary description. The specific form can be determined according to processing requirements and is not limited here.

[0113] In one or more embodiments of the present specification, before returning the first reply information to the first community, the main model may also perform compliance verification on the first reply information based on the verification sub-model to determine whether it includes information that violates laws and regulations and personal privacy information that is prohibited from being disclosed. Optionally, if the verification passes, the verification sub-model may directly output indication information indicating that the verification has passed, so that the main model can return the original first reply information to the first community; correspondingly, if the verification fails, the non-compliant content in the first reply information may be filtered based on the screening sub-model and the filtered content may be output to the main model, so that the main model can return the output result as the new first reply information to the first community.

[0114] Further optionally, the present embodiment may also provide an output sub-model for personalizing the first reply information, so as to set the output form of the first reply information. For example, in the above example, if the first reply information is verified to include personal privacy information, the verification sub-model may not perform filtering, but instead set the corresponding information to an output form of "only visible to the first user" through the output sub-model, thereby ensuring information security while also allowing the first user to confirm the relevant information.

[0115] It should be noted that the types and numbers of the various first sub-models in the above examples, the specific operations performed by each first sub-model, and the scheduling method and information processing method between the main model and each first sub-model are only exemplary descriptions and are not limited to this in practical applications. For example, in the above example, the mapping sub-model can also be split into two sub-models. Optionally, it can be split into a user sub-model for determining user information and an order sub-model for determining historical purchase record information. For another example, the verification sub-model and the shielding sub-model can also be combined into a first sub-model, which can perform both compliance verification operations and invalid information filtering operations.

[0116] In actual applications, since different brands have their own brand characteristics, the output model can also set the output form of the first reply information to be adapted to the brand characteristics of the first tenant, so as to enhance the user experience and highlight the characteristics of the brand. For example, if the first tenant is mainly in the "two-dimensional" style, the output sub-model can set the output form of the first reply information to "output in the form of a cartoon character's voice"; for another example, if the first tenant is mainly in the "national style" style, the output sub-model can set the output form of the first reply information to "output in the form of graphics or animation with ink landscape painting as the background", etc. Based on this, after receiving the first reply information returned by the cloud server, the local server of the first tenant can perform output control, page rendering and other operations according to its corresponding output form. The specific implementation form can be determined according to actual needs and will not be described in detail here.

[0117] In actual applications, since there are multiple users in a community, each user may input query information in the community, and each user may input multiple query information. There may be an association relationship between multiple query information input by one user or multiple query information input by multiple users. Optionally, the association relationship between multiple query information can be determined based on the query scenarios corresponding to the multiple query information. If the query scenarios corresponding to the multiple query information are the same or similar, they are determined to be related to each other, otherwise they are determined to be unrelated. Of course, the specific determination method is not limited to this.

[0118] For example, a user wants to inquire about a product from multiple dimensions and inputs multiple inquiry messages respectively. In this case, the first inquiry message obtained by the main model can be a group message fragment in the first community. Based on this, the main model can perform context recognition on the group message fragment, determine the inquiry scenario corresponding to each inquiry message, and determine the correlation between the two. Then, according to the method in the above embodiment, the corresponding first reply information is uniformly determined for these multiple inquiry messages based on each first sub-model.

[0119] For another example, in response to a query message input by one user, another user inputs a reply message, or inputs a similar query message in the same query scenario. In this case, in addition to obtaining the first query message input by the first user in the first community, the main model can also obtain the query message input by the second user in the first community and use it as the second query message. Optionally, the main model can obtain a group message segment in the first community and perform context recognition on the group message segment to extract the first query message and the second query message. Furthermore, semantic recognition is performed on the first query message and the second query message, and the query scenarios corresponding to the two are determined to determine the association relationship between the two. Then, the second reply information is obtained based on the association relationship, and the second reply information is returned to the first community.

[0120] In the above embodiment, there is no limitation on the way in which the main model obtains the second reply information. Optionally, when the main model determines that the first query information and the second query information are not related, it can directly determine the reply information corresponding to the first query information based on each first sub-model in the manner of the above embodiment, and return the reply information as the second reply information to the first community. Correspondingly, when the main model determines that the first query information and the second query information are related, it can determine whether the second query information can be used as the reply information of the first query information based on each first sub-model in the manner of the above embodiment.

[0121] Optionally, when it is determined that the second inquiry information is not accurate enough as the reply information of the first inquiry information, the reply information determined based on each first sub-model can be returned to the first community as the second reply information in the manner in the above embodiment; or, when it is determined that the second inquiry information can be used as the reply information of the first inquiry information, indication information for expressing affirmation, such as indication information similar to "yes" or "correct", can be returned to the first community as the second reply information to enhance the sense of interaction within the community.

[0122] It should be noted that in the above embodiments, the method of determining whether multiple inquiry messages within the first community are related, and determining reply information based on the correlation between the multiple inquiry messages is only an exemplary description and is not limited to this in practical applications. The above embodiments are intended to illustrate that the embodiments of the present application support determining reply information based on the correlation between multiple inquiry messages, and do not constitute a specific limitation.

[0123] The overall execution process of the community operation method based on the big model provided by one or more embodiments of this specification is described below in conjunction with the accompanying drawings.

[0124] Figure 2b This is a flowchart of another large model-based community operation method provided in an embodiment of this specification. In this example, the method is still executed by a cloud server and the first user asks "How many flavors of French fries are there?" as an example for explanation.

[0125] like Figure 2b As shown, in this embodiment, the cloud server can obtain the first query information input by the first user in the first community, and then input the first query information into the verification sub-model and the shielding sub-model to verify the compliance of the first query information based on the verification sub-model, and filter invalid information from the first query information based on the shielding sub-model. Figure 2b As shown, if the verification sub-model fails to pass the verification, a prompt message is returned to the requesting end to prompt that the request does not comply with the regulations; if the screening sub-model determines that the first query information includes invalid information, the first query information can be withdrawn, a warning message can be returned to the requesting end, or the first user can be banned, kicked out of the community, blacklisted, etc. according to the number of times the first user enters invalid information. Further, if the first query information passes the verification sub-model and is confirmed by the screening sub-model to be not invalid information, as shown in FIG. Figure 2b As shown, the first query information that has passed the verification is input into the main model.

[0126] Furthermore, based on the main model, semantic recognition can be performed on the first query information to determine whether the first query information is a complaint or suggestion, etc. If it is determined to be, the processing sub-model will be used to determine the specific solution and return it to the request end; correspondingly, if it is determined not to be, the first sub-models associated with the query scenario and at least one query dimension corresponding to the first query information and the association order between all the first sub-models will be further determined. Then, according to the association order, the query result information determined by the previous first sub-model is input as an input parameter to the next first sub-model in turn, and the query result information of each first sub-model is obtained in turn.

[0127] like Figure 2b As shown, after the main model identifies, it is determined that the first sub-models sequentially associated with the first query information are the product sub-model, the mapping sub-model, and the promotion sub-model. Based on this, the main model can first input the formatted first query information into the product sub-model according to the association order, so that the product sub-model can determine the product information corresponding to the first query information, and output it as the query result. Further, the main model can obtain the product information determined by the product sub-model and format it, and input the processing result as an input parameter into the mapping sub-model. Based on this, the mapping sub-model can determine the user information corresponding to the first user who inputs the first query information, and the mapping information between the user information and the product information, and output it as the query result. Further, the main model can obtain the mapping information determined by the mapping sub-model and format it, and input the processing result as an input parameter into the promotion sub-model. Based on this, the promotion sub-model can determine the preferential activities that the first user can participate in, the coupons that can be used, the related products that can be purchased, and other recommended information based on the mapping information, and output it as the query result.

[0128] like Figure 2b As shown, after obtaining the query results output by all the above sub-models, the main model can also determine whether to set the output form of the reply information through the output sub-model. If necessary, the output sub-model sets the corresponding output form based on the brand characteristics of the first tenant corresponding to the first community. Figure 2b As shown, the query result can also be checked for compliance through a verification model to ensure that the reply information returned to the first community is compliant and does not involve sensitive information. Based on this, after the above verification is passed, the main model can return the query result information determined by all sub-models as the first reply information corresponding to the first query information to the first community.

[0129] It should be noted that Figure 2b The example shown is intended to illustrate the overall process of the community operation method provided in the embodiments of this specification. The specific execution details of each step can be found in the description of the corresponding part in the above embodiments, and will not be repeated here.

[0130] Based on the above, the community operation method based on the large model provided in one or more embodiments of this specification can obtain the inquiry information raised by users in the offline tenant community, perform semantic recognition on the corresponding inquiry information based on the pre-trained main model, and determine from which inquiry dimensions the user raises the corresponding inquiry information. Furthermore, the inquiry result information of the above inquiry information under the corresponding inquiry dimension is determined based on the sub-model pre-trained under each inquiry dimension. Based on this, the main model can return the inquiry result information determined by all sub-models from different inquiry dimensions to the corresponding community as the reply information of the above inquiry information for the user to view. In this method, based on each model, various inquiry information raised by users in the community is automatically identified and timely feedback is given, which can not only simulate the manual operation method and reduce labor costs, but also provide users with timely services quickly and efficiently, which helps to maintain users' trust in the brand, improve user retention rate, and provide a basis for user conversion of other derivative transactions.

[0131] It can be understood that the above embodiments are only examples and can be modified in actual implementation. Those skilled in the art can understand that the modification methods of the above embodiments without creative labor fall within the protection scope of one or more embodiments of this specification and will not be repeated in the embodiments.

[0132] Based on the same inventive concept, one or more embodiments of this specification further provide a community operation device based on a large model. Figure 3 This is a structural diagram of the community operation device, such as Figure 3 As shown, the community operation device 300 includes: a first acquisition module 301, a first determination module 302, a second determination module 303, a second acquisition module 304 and an output module 305, wherein:

[0133] The first acquisition module 301 is used to obtain the first query information input by the first user in the first community; the first determination module 302 is used to determine at least one query dimension corresponding to the first query information based on the main model; wherein the main model is obtained by model training with the first training data, and the first training data is historical query information under different query scenarios; the second determination module 303 is used to determine the first sub-models respectively associated with at least one query dimension and the association order between all the first sub-models; wherein each first sub-model is obtained by model training with the second training data under one query dimension, and the second training data is historical question-answer pair information under the corresponding query dimension, and different first sub-models are used to determine the query result information of the first query information under the corresponding query dimension; the second acquisition module 304 is used to obtain the query result information determined by each first sub-model in turn according to the association order; the output module 305 is used to return the query result information determined by all the first sub-models as the first reply information to the first community.

[0134] In an optional embodiment, the first determination module 302 determines at least one query dimension corresponding to the first query information based on the main model, and is used to: input the first query information into the main model; and obtain at least one query dimension determined after the main model performs semantic recognition on the first query information.

[0135] In an optional embodiment, the first determination module 302 determines at least one inquiry dimension corresponding to the first inquiry information based on the main model, and is used to: perform compliance verification on the first inquiry information; after the verification passes, determine at least one inquiry dimension corresponding to the first inquiry information based on the main model.

[0136] In an optional embodiment, the second determination module 303 determines the first sub-models respectively associated with at least one inquiry dimension and the order of association between all the first sub-models, and is used to: determine the first inquiry scenario corresponding to the first inquiry information according to the semantic recognition result of the main model for the first inquiry information, and the first inquiry scenario corresponds to at least one inquiry dimension; obtain mapping information pre-configured for the first inquiry scenario; the mapping information includes the second sub-models respectively associated with all the inquiry dimensions corresponding to the first inquiry scenario, and the order of association between all the second sub-models, and all the second sub-models include all the first sub-models; according to the mapping information, determine the first sub-models respectively associated with at least one inquiry dimension and the order of association between all the first sub-models.

[0137] In an optional embodiment, the second acquisition module 304 acquires the query result information determined by each first sub-model in sequence according to the association order, and is used to: determine the first first sub-model in the association order according to the association order; input the first query information into the first first sub-model, and acquire the query result information determined by the first first sub-model; according to the association order between the first first sub-model and other first sub-models, input the query result information determined by the previous first sub-model as an input parameter into the next first sub-model in sequence, and acquire the query result information of each first sub-model in sequence.

[0138] In an optional embodiment, the second acquisition module 304 inputs the first query information into the first first sub-model, and obtains the query result information determined by the first first sub-model, and is used to: format the first query information according to the semantic recognition result of the main model on the first query information; input the formatted first query information as an input parameter into the first first sub-model; and obtain the query result information determined by the first first sub-model.

[0139] In an optional embodiment, the second acquisition module 304 sequentially inputs the query result information determined by the previous first submodel as an input parameter into the next first submodel according to the order of association between the first first submodel and other first submodels, and sequentially acquires the query result information of each first submodel, and is used for: sequentially inputting the formatted processing result of the query result information determined by the previous first submodel as an input parameter into the next first submodel according to the order of association between the first first submodel and other first submodels, and sequentially determining the query result information of each first submodel.

[0140] In an optional embodiment, the output module 305 returns all the query result information determined by the first sub-model as the first reply information to the first community, and is used to: perform compliance verification on all the query result information determined by the first sub-model; after the verification is passed, return all the query result information determined by the first sub-model as the first reply information to the first community.

[0141] In an optional embodiment, the first acquisition module 301 is further used to: obtain second inquiry information input by a second user in the first community; determine the association between the first inquiry information and the second inquiry information; and obtain second reply information based on the association and return the second reply information to the first community.

[0142] It should be noted that, regarding the specific functions and processing principles of each module in the above-mentioned community operation device based on the big model, please refer to the description of the corresponding part in the above-mentioned method embodiment, which will not be repeated here.

[0143] Based on the same inventive concept, one or more embodiments of this specification further provide a computer device, Figure 4 is the structural block diagram of the computer device. Figure 4 As shown, the computer device 400 includes a processor 401 and a memory 402, wherein the processor 401 is used to perform the following operations:

[0144] Obtaining first query information input by a first user in the first community;

[0145] Based on the main model, determining at least one inquiry dimension corresponding to the first inquiry information; wherein the main model is obtained by model training based on the first training data, and the first training data is historical inquiry information under different inquiry scenarios;

[0146] Determine the first sub-models respectively associated with at least one query dimension and the association order between all the first sub-models; wherein each first sub-model is obtained by model training using second training data under one query dimension, the second training data is historical question-answer pair information under the corresponding query dimension, and different first sub-models are used to determine query result information of the first query information under the corresponding query dimension;

[0147] According to the association order, sequentially obtain the query result information determined by each first sub-model;

[0148] All query result information determined by the first sub-model is returned to the first community as first reply information.

[0149] It should be noted that the functional principles of each component in the above-mentioned computer device are the same as the implementation principles of the corresponding steps in the above-mentioned community operation method based on the big model. For details, please refer to the description of the corresponding parts in the above-mentioned method embodiment, which will not be repeated here.

[0150] In some optional implementations, the processor 401 is sometimes also referred to as a controller or an operation control, and may include a microprocessor or other processor device and / or logic device, which receives input and controls the operation of various components of the computer device 400.

[0151] The memory 302 may be coupled to the processor 401, and the memory 302 may be one or more of a cache, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory, or other suitable devices. The program for executing the above-mentioned community operation method based on the large model may be stored. The processor 401 may execute the program stored in the memory 302 to implement information storage or processing, etc.

[0152] The memory 302 may be a solid-state memory, such as a read-only memory (ROM), a random access memory (RAM), a SIM card, etc. It may also be a memory that saves information even when the power is off, can be selectively erased, and is provided with more data, examples of which are sometimes referred to as EPROMs, etc. The memory 302 may also be some other type of device. The memory 302 includes a buffer memory (sometimes referred to as a buffer). The memory 302 may include an application / function storage unit for storing application programs and function programs or processes for executing the operation of the computer device 400 through the processor 401.

[0153] The memory 302 may also include a data storage unit for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the device. The driver storage unit of the memory 302 may include various drivers for the communication function of the computer device and / or for executing other functions of the computer device (such as a messaging application, a contact book application, etc.).

[0154] It is worth noting that Figure 4 The components shown are exemplary, and other types of structures may be used to supplement or replace the structures to implement telecommunication functions or other functions. Of course, the computer device 400 does not necessarily have to include Figure 4 In addition, the computer device 400 may also include Figure 4 For components not shown, reference may be made to the prior art.

[0155] In some optional implementations, the computer device 400 may also include: a memory 302, a communication module, an input unit, an audio processor, a display, a power supply and other components.

[0156] The input unit can provide input to the processor 401. The input unit is, for example, a key or a touch input device. The power supply can be used to provide power to the computer device 400. The display can be used to display display objects such as images and text. The display can be, for example, an LCD display, but is not limited thereto.

[0157] The communication module is a transmitter / receiver that sends and receives signals via an antenna. The communication module (transmitter / receiver) is coupled to the processor 401 to provide input signals and receive output signals, which may be the same as in a conventional mobile communication terminal.

[0158] Based on different communication technologies, multiple communication modules may be provided in the same computer device, such as a cellular network module, a Bluetooth module and / or a wireless local area network module, etc. The communication module (transmitter / receiver) is also coupled to a speaker and a microphone via an audio processor to provide an audio output via the speaker and receive an audio input from the microphone, thereby realizing a common telecommunication function. The audio processor may include any suitable buffer, decoder, amplifier, etc. In addition, the audio processor is also coupled to the processor 401, so that recording can be performed on the machine through the microphone, and the sound stored on the machine can be played through the speaker.

[0159] One or more embodiments of the present specification also provide a computer-readable storage medium capable of implementing all steps of the community operation method based on the big model in the above embodiment. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, all steps of the community operation method based on the big model in the above embodiment are implemented. For example, when the processor executes the computer program, the following steps are implemented:

[0160] Obtaining first query information input by a first user in the first community;

[0161] Based on the main model, determining at least one inquiry dimension corresponding to the first inquiry information; wherein the main model is obtained by model training based on the first training data, and the first training data is historical inquiry information under different inquiry scenarios;

[0162] Determine the first sub-models respectively associated with at least one query dimension and the association order between all the first sub-models; wherein each first sub-model is obtained by model training using second training data under one query dimension, the second training data is historical question-answer pair information under the corresponding query dimension, and different first sub-models are used to determine query result information of the first query information under the corresponding query dimension;

[0163] According to the association order, sequentially obtain the query result information determined by each first sub-model;

[0164] All query result information determined by the first sub-model is returned to the first community as first reply information.

[0165] It should be noted that the functional principle of the above-mentioned computer-readable storage medium is the same as the implementation principle of the corresponding steps in the above-mentioned community operation method based on the big model. For details, please refer to the description of the corresponding part in the above-mentioned method embodiment, which will not be repeated here.

[0166] Although one or more embodiments of this specification provide method operation steps such as embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-creative work. The order of steps listed in the embodiments is only one way of executing the order of many steps and does not represent the only execution order. When the actual device or client product is executed, it can be executed in the order of the method shown in the embodiments or the drawings or in parallel (for example, in a parallel processor or multi-threaded processing environment).

[0167] It should be understood by those skilled in the art that the embodiments of this specification may be provided as methods, devices (systems) or computer program products. Therefore, the embodiments of this specification may take the form of complete hardware embodiments, complete software embodiments or embodiments combining software and hardware. Moreover, one or more embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0168] One or more embodiments of the present specification are described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to one or more embodiments of the present specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0169] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0170] These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0171] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0172] In this article, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. For those of ordinary skill in the art, the specific meaning of the above terms in one or more embodiments of this specification can be understood according to the specific circumstances.

[0173] It should be noted that, in the absence of conflict, one or more embodiments of this specification and the features in the embodiments may be combined with each other. One or more embodiments of this specification are not limited to any single aspect, nor to any single embodiment, nor to any combination and / or replacement of these aspects and / or embodiments. Moreover, each aspect and / or embodiment of one or more embodiments of this specification may be used alone or in combination with one or more other aspects and / or embodiments thereof.

[0174] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of one or more embodiments of this specification, rather than to limit them. Although one or more embodiments of this specification are described in detail with reference to the aforementioned embodiments, a person of ordinary skill in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some or all of the technical features therein can be replaced by equivalents. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of one or more embodiments of this specification, and they should all be included in the scope of one or more embodiments of this specification.

[0175] One or more embodiments of this specification are described above in conjunction with optional implementation methods, but these implementation methods are only exemplary and serve only as an illustration. On this basis, multiple replacements and improvements can be made to one or more embodiments of this specification, all of which fall within the scope of protection of one or more embodiments of this specification.

Claims

1. A community operation method based on a large model, characterized in that: include: Obtaining first query information input by a first user in the first community; Determining, based on the main model, at least one inquiry dimension corresponding to the first inquiry information; The main model is obtained by model training using first training data, where the first training data is historical query information in different query scenarios; Determine the first sub-models respectively associated with the at least one inquiry dimension and the order of association between all the first sub-models; each first sub-model is obtained by model training using second training data under one inquiry dimension, the second training data is historical question-answer pair information under the corresponding inquiry dimension, and different first sub-models are used to determine inquiry result information of the first inquiry information under the corresponding inquiry dimension; According to the association order, sequentially obtain the query result information determined by each first sub-model; All inquiry result information determined by the first sub-model is returned to the first community as first reply information.

2. The method according to claim 1, characterized in that: Determining, based on the main model, at least one inquiry dimension corresponding to the first inquiry information includes: inputting the first query information into a main model; At least one query dimension determined by the main model after semantic recognition of the first query information is obtained.

3. The method according to claim 1, characterized in that: Determining, based on the main model, at least one inquiry dimension corresponding to the first inquiry information includes: Performing compliance verification on the first inquiry information; After verification is passed, at least one query dimension corresponding to the first query information is determined based on the main model.

4. The method according to claim 2, characterized in that: Determining the first sub-models respectively associated with the at least one query dimension and the association order between all the first sub-models includes: Determining, according to a semantic recognition result of the first query information by the main model, a first query scenario corresponding to the first query information, where the first query scenario corresponds to the at least one query dimension; Acquire mapping information pre-configured for the first inquiry scenario; the mapping information includes second sub-models respectively associated with all inquiry dimensions corresponding to the first inquiry scenario, and an association order between all second sub-models, and all second sub-models include all first sub-models; According to the mapping information, the first sub-models respectively associated with the at least one query dimension and the association order between all the first sub-models are determined.

5. The method according to claim 4, characterized in that According to the association order, sequentially obtaining the query result information determined by each first sub-model includes: According to the association sequence, determining a first first sub-model in the association sequence; Inputting the first query information into the first first sub-model, and obtaining query result information determined by the first first sub-model; According to the association order between the first first sub-model and other first sub-models, the query result information determined by the previous first sub-model is input as an input parameter to the next first sub-model in turn, and the query result information of each first sub-model is obtained in turn.

6. The method according to claim 5, characterized in that Inputting the first query information into the first first sub-model and obtaining query result information determined by the first first sub-model includes: formatting the first query information according to a semantic recognition result of the first query information by the main model; Inputting the formatted first query information as an input parameter into the first first sub-model; Obtain query result information determined by the first first sub-model.

7. The method according to claim 6, characterized in that According to the association order between the first first sub-model and other first sub-models, the query result information determined by the previous first sub-model is sequentially input as an input parameter into the next first sub-model, and the query result information of each first sub-model is sequentially obtained, including: According to the association order between the first sub-model and other first sub-models, the formatted processing results of the query result information determined by the previous first sub-model are input as input parameters to the next first sub-model in turn, and the query result information of each first sub-model is determined in turn.

8. The method according to any one of claims 1 to 7, characterized in that: Returning all query result information determined by the first sub-model as first reply information to the first community, including: Performing compliance check on all query result information determined by the first sub-model; After the verification is passed, all inquiry result information determined by the first sub-model is returned to the first community as the first reply information.

9. The method according to any one of claims 1 to 7, characterized in that: Also includes: Obtaining second query information input by a second user in the first community; Determine an association relationship between the first query information and the second query information; According to the association relationship, second reply information is obtained and the second reply information is returned to the first community.

10. A community operation device based on a large model, characterized in that: include: A first acquisition module, used to acquire first query information input by a first user in a first community; A first determination module, configured to determine, based on a main model, at least one inquiry dimension corresponding to the first inquiry information; The main model is obtained by model training using first training data, where the first training data is historical query information in different query scenarios; A second determination module is used to determine the first sub-models respectively associated with the at least one inquiry dimension and the association order between all the first sub-models; each first sub-model is obtained by model training with second training data under one inquiry dimension, the second training data is historical question-answer pair information under the corresponding inquiry dimension, and different first sub-models are used to determine inquiry result information of the first inquiry information under the corresponding inquiry dimension; A second acquisition module, used to sequentially acquire the query result information determined by each first sub-model according to the association order; The output module is used to return all the inquiry result information determined by the first sub-model to the first community as the first reply information.

11. The device according to claim 10, characterized in that The first determination module determines, based on the main model, at least one inquiry dimension corresponding to the first inquiry information, for: inputting the first query information into a main model; At least one query dimension determined by the main model after semantic recognition of the first query information is obtained.

12. The device according to claim 10, characterized in that The first determination module determines, based on the main model, at least one inquiry dimension corresponding to the first inquiry information, for: Performing compliance verification on the first inquiry information; After verification is passed, at least one query dimension corresponding to the first query information is determined based on the main model.

13. The device according to claim 11, characterized in that The second determination module determines the first sub-models respectively associated with the at least one query dimension and the association order between all the first sub-models, for: Determining, according to a semantic recognition result of the first query information by the main model, a first query scenario corresponding to the first query information, where the first query scenario corresponds to the at least one query dimension; Acquire mapping information pre-configured for the first inquiry scenario; the mapping information includes second sub-models respectively associated with all inquiry dimensions corresponding to the first inquiry scenario, and an association order between all second sub-models, and all second sub-models include all first sub-models; According to the mapping information, the first sub-models respectively associated with the at least one query dimension and the association order between all the first sub-models are determined.

14. The device according to claim 13, characterized in that The second acquisition module sequentially acquires the query result information determined by each first sub-model according to the association order, and is used to: According to the association sequence, determining a first first sub-model in the association sequence; Inputting the first query information into the first first sub-model, and obtaining query result information determined by the first first sub-model; According to the association order between the first first sub-model and other first sub-models, the query result information determined by the previous first sub-model is input as an input parameter to the next first sub-model in turn, and the query result information of each first sub-model is obtained in turn.

15. The device according to claim 14, characterized in that The second acquisition module inputs the first query information into the first first sub-model and acquires the query result information determined by the first first sub-model, for: formatting the first query information according to a semantic recognition result of the first query information by the main model; Inputting the formatted first query information as an input parameter into the first first sub-model; Obtain query result information determined by the first first sub-model.

16. The device according to claim 15, characterized in that The second acquisition module sequentially inputs the query result information determined by the previous first sub-model as an input parameter to the next first sub-model according to the association order between the first first sub-model and other first sub-models, and sequentially acquires the query result information of each first sub-model, for: According to the association order between the first sub-model and other first sub-models, the formatted processing results of the query result information determined by the previous first sub-model are input as input parameters to the next first sub-model in turn, and the query result information of each first sub-model is determined in turn.

17. The device according to any one of claims 10 to 16, characterized in that: The output module returns all the query result information determined by the first sub-model as the first reply information to the first community, for: Performing compliance check on all query result information determined by the first sub-model; After the verification is passed, all inquiry result information determined by the first sub-model is returned to the first community as the first reply information.

18. The device according to any one of claims 10 to 16, characterized in that: The first acquisition module is also used for: Obtaining second query information input by a second user in the first community; Determine an association relationship between the first query information and the second query information; According to the association relationship, second reply information is obtained and the second reply information is returned to the first community.

19. A computer device, characterized in that: include: A memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, they are used to implement the method described in any one of claims 1 to 9.

20. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, which, when executed, implement the method described in any one of claims 1 to 9.

Images