Customer resource information distribution system, method, apparatus, and program product

Abstract

The embodiment of the invention provides a client resource information distribution system, method, device and program product, and the system comprises a client resource information collection unit which is used for obtaining target client resource information; the customer information distribution unit is used for constructing a primary distribution prompt instruction including target customer resource information and sending the target customer resource information to a store system corresponding to the target store information; the artificial intelligence unit is used for acquiring current store evaluation information corresponding to candidate store information according to the primary distribution prompt instruction, constructing a target distribution prompt instruction based on the candidate store information, the current store evaluation information and the primary distribution prompt instruction, and sending the target distribution prompt instruction to the server; and determining target store information in the candidate store information by using a generative large model according to the target distribution prompt instruction. The customer resource information distribution system can improve the accuracy of distributing customer resource information to stores.

Description

TECHNICAL FIELD

[0001] One or more embodiments of the present specification relate to the technical field of Internet, and in particular, to a customer resource information distribution system, method, device and program product. BACKGROUND

[0002] In the field of customer resource information distribution, traditional O2O consumer medical platforms usually adopt manual distribution or system distribution strategy based on fixed rules to allocate customer resource information collected online to offline stores in the hope of converting it into orders in the stores. In order to improve the accuracy and conversion rate of customer resource distribution, platforms usually rely on manual operation personnel for experience-based judgment.

[0003] However, there is a problem in the related art that the customer resource information distribution does not adapt to the actual situation of the store. SUMMARY

[0004] Therefore, one or more embodiments of the present specification provide a customer resource information distribution system, method, device and program product, which can improve the accuracy of allocating customer resource information to stores.

[0005] According to a first aspect of one or more embodiments of the present specification, a customer resource information distribution system is provided, comprising: a customer information collection unit configured to obtain target customer resource information; a customer distribution unit configured to construct a primary distribution prompt instruction comprising the target customer resource information, to call an artificial intelligence unit to determine target store information according to the primary distribution prompt instruction, and to send the target customer resource information to a store system corresponding to the target store information; and an artificial intelligence unit configured to obtain candidate store information and current store evaluation information corresponding to the candidate store information according to the target customer resource information in the primary distribution prompt instruction, to construct a target distribution prompt instruction based on the candidate store information, the current store evaluation information and the primary distribution prompt instruction, to determine target store information in the candidate store information according to the target distribution prompt instruction using a generative large model, and to feed back the target store information to the customer distribution unit; wherein the current store evaluation information is used to represent the conversion effect of the corresponding candidate store on the customer resource information that has been received.

[0006] According to a second aspect of one or more embodiments of the present specification, a method for distributing customer resource information is provided, comprising: obtaining target customer resource information; constructing a primary distribution prompt instruction including the target customer resource information, the primary distribution prompt instruction calling an artificial intelligence unit to determine target store information, and sending the target customer resource information to a store system corresponding to the target store information; wherein the artificial intelligence unit is configured to obtain candidate store information and current store evaluation information corresponding to the candidate store information according to the target customer resource information in the primary distribution prompt instruction, construct a target distribution prompt instruction based on the candidate store information, the current store evaluation information, and the primary distribution prompt instruction, and determine the target store information in the candidate store information according to the target distribution prompt instruction using a generative large model; wherein the current store evaluation information is used to represent the conversion effect of the corresponding candidate store on the customer resource information that has been received.

[0007] According to a third aspect of one or more embodiments of the present specification, a computer program product is provided, comprising computer programs / instructions that, when executed by a processor, implement the steps of the method according to the second aspect.

[0008] According to a fourth aspect of one or more embodiments of the present specification, a computer device is provided, comprising a memory and a processor, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor to implement the steps of the method according to the second aspect.

[0009] From the above embodiments, it can be seen that the embodiments of the present specification construct a customer resource information distribution system with intelligent recommendation capability by setting a customer information collection unit, a customer distribution unit, and an artificial intelligence unit. The customer information collection unit is configured to uniformly obtain target customer resource information, the customer distribution unit is configured to construct a primary distribution prompt instruction based on the target customer resource information, and call the artificial intelligence unit to determine target store information. After receiving the primary distribution prompt instruction, the artificial intelligence unit first obtains candidate store information and current store evaluation information corresponding to the candidate store information according to the target customer resource information contained in the prompt instruction; then, based on the candidate store information, the current store evaluation information, and the primary distribution prompt instruction, a target distribution prompt instruction is constructed, and a generative large model is further used to intelligently analyze the candidate store information to determine the target store information, and the target store information is fed back to the customer distribution unit. Through the above cooperative processing, the customer resource information distribution system can comprehensively consider customer characteristics and actual performance of store reception, and distribute customer resource information more accurately to stores. BRIEF DESCRIPTION OF DRAWINGS

[0010] Figure 1 This is a schematic diagram of the architecture of a customer resource information distribution system provided by an exemplary embodiment.

[0011] Figure 2 The present invention is a flowchart of a method for distributing customer resource information provided by an exemplary embodiment.

[0012] Figure 3 This is a module diagram of a computer device provided by an exemplary embodiment. DETAILED DESCRIPTION

[0013] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments.

[0014] In the description of the embodiments of the present application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0015] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this manual are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0016] In related technologies, consumer medical platforms can collect customer resource information through multiple channels and send the customer resource information to appropriate stores so that the stores can contact the users corresponding to the customer resource information, hoping to bring orders to the stores and convenience to users.

[0017] When distributing customer resource information, consumer healthcare platforms typically rely on manual operations personnel to distribute this information collected online to offline stores. Specifically, these personnel manually identify and distribute customer resource information to appropriate stores based on factors such as the customer's basic information, browsing behavior, and location, providing the customer resource information to the corresponding store systems. While this approach offers a certain degree of flexibility, the distribution process is highly dependent on manual experience, resulting in low efficiency, strong subjectivity, and prone to errors, which in turn impacts the quality of customer resource distribution and subsequent conversion rates.

[0018] To improve the distribution efficiency, the consumer medical platform introduces an automatic distribution method based on regular expression matching rules. This method identifies the key fields in the customer resource information (such as the positioning area, the intended project keyword, etc.) through pre-set regular expressions, and matches them with the store feature fields to achieve regular automatic distribution. This method improves the distribution efficiency of customer resource information to a certain extent and reduces manual participation. However, since regular expressions are essentially a pattern matching mechanism, it is difficult to understand the deep semantics and personalized needs behind customer behavior, so there is still a problem of low distribution accuracy.

[0019] With the development of artificial intelligence technology, large language model can be applied to the transaction scenario of distributing customer resource information. In related technology, after selecting a large model base, the large model base can be further trained with a large number of samples in the consumer medical field, so that the large language model can understand the relevant background knowledge in the consumer medical field. However, when directly using the trained large language model to distribute customer resource information, there is still a problem of inaccuracy. Researchers found that after training the large language model with new training samples, the large language model has good distribution performance, but after a period of use, the accuracy of the large language model in distributing customer resource information will decrease. After further research and analysis, researchers found that the actual operation of the store in the real world is dynamically changing. When the large language model is trained with samples, it learns how to distribute different customer resource information to the corresponding store. However, after the store has changed dynamically, it may no longer be suitable for the customers represented by the customer resource information, while the large language model still distributes according to the knowledge it has learned, so it leads to inaccurate distribution. Therefore, researchers have to periodically fine-tune the large language model to improve its accuracy. Each cycle is short, which greatly increases the workload of researchers, but if each cycle is too long, the accuracy of the large language model at the end of each cycle will decrease to some extent.

[0020] In summary, the existing technology for distributing customer resource information still has the problem of low accuracy.

[0021] Please refer to Figure 1 One embodiment of the present application provides an application scenario example of a customer resource information distribution system. The customer resource information distribution system obtains target customer resource information of a user through a customer information collection unit, and calls an artificial intelligence unit to determine target store information that matches the customer's needs, and then completes intelligent distribution of customer resources. In this scenario example, the medical and beauty consumer scenario is taken as an example, and the customer resource information distribution system determines the store information suitable for providing services to the user by calling the artificial intelligence unit.

[0022] For example, a user frequently browses medical and cosmetic project pages such as "skin whitening" and "hyaluronic acid injection" on the mobile App of the platform, and repeatedly consults online customer service about specific prices and service details, showing a high willingness to consume. The customer information collection unit obtains the customer resource information of the user through multiple resource paths, such as through the user's browsing behavior, consultation dialogue record, and service reservation data, and generates intention strength information representing the degree of the user's real consumption willingness based on the collected customer resource information as part of the target customer resource information.

[0023] Subsequently, the customer distribution unit constructs a primary distribution prompt instruction according to the received target customer resource information and the predefined static distribution rule, and sends the primary distribution prompt instruction to the artificial intelligence unit. After receiving the primary distribution prompt instruction, the artificial intelligence unit calls the RAG module through the scheduling module, retrieves and filters the candidate store information and the corresponding current store evaluation information from the RAG knowledge base according to the target customer resource information. The candidate store information includes, for example, the store service range, geographical location, and historical performance of accepting services related to the user's needs; the current store evaluation information specifically includes the actual store visit rate, conversion rate, and other conversion effect data after the store accepts similar customer resources in a recent period of time.

[0024] The scheduling module further constructs a target distribution prompt instruction based on the static distribution rule, candidate store information, and current store evaluation information obtained from the primary distribution prompt instruction. In addition to the static distribution rule, the target distribution prompt instruction also includes dynamic distribution rules corresponding to the target customer resource information, such as specific service discount policies dynamically generated during recent beauty market promotion activities, to guide the generative large model to make intelligent choices of stores.

[0025] After receiving the target distribution prompt instruction, the generative large model conducts comprehensive analysis combining the customer's intention strength information, candidate store information, and current store evaluation information. For example, if the customer's intention strength score is high, the generative large model will prioritize stores with high store visit conversion rates among the candidate stores to improve the actual conversion probability of customer resources. Specifically, assuming that the store visit conversion rate of candidate store A receiving similar customer resource information in the past month is 80%, the conversion rate of candidate store B is 60%, and the conversion rate of candidate store C is only 30%, the generative large model will preferentially determine store A as the target store information.

[0026] After determining the target store information, the artificial intelligence unit feeds back the target store information to the customer distribution unit, which further sends the target customer resource information to the store system corresponding to store A, so that store A can timely receive and respond to customer service needs.

[0027] After the user actually goes to the store to receive services, the store system of store A returns feedback information such as the actual store service of the customer, whether a transaction is made, the transaction amount, and the satisfaction score to the customer resource information distribution system to the customer feedback unit in real time. The customer feedback unit updates the feedback information to the corresponding store information of the RAG knowledge base in real time, so as to dynamically adjust the current store evaluation information of the store. Thus, the next time the customer resource information distribution system calls the generative large model to select a store, it can make a decision based on the updated new store evaluation information, and more accurately improve the customer resource distribution and store conversion effect.

[0028] As can be seen from the above scenario example, the customer resource information distribution system can accurately match customer demand and store capacity by collecting multi-dimensional data and analyzing intention strength through the customer information collection unit, combining dynamically updated store evaluation information, and using the artificial intelligence unit to intelligently distribute customer resource information, thereby improving the actual store and transaction conversion rate of customers.

[0029] In the multiple embodiments provided in the present application, the customer resource information distribution system can be applied to electronic devices with certain computing power and network access capability. The electronic device can be a desktop computer, a notebook computer, a tablet computer, a smart phone, or a data processing server deployed in the cloud. The electronic device can call the artificial intelligence unit through network access and interact with the store system to complete the distribution task of customer resource information. Of course, the customer resource information distribution system can also be deployed in multiple electronic devices in a distributed manner.

[0030] One embodiment of the present application provides a customer resource information distribution system. The customer resource information distribution system can include a customer information collection unit, a customer distribution unit, and an artificial intelligence unit. The customer information collection unit is configured to obtain target customer resource information. The customer distribution unit is configured to construct a primary distribution prompt instruction including the target customer resource information, call the artificial intelligence unit to determine target store information according to the primary distribution prompt instruction, and send the target customer resource information to the store system corresponding to the target store information. The artificial intelligence unit is configured to obtain candidate store information according to the target customer resource information in the primary distribution prompt instruction, construct a target distribution prompt instruction based on the candidate store information and the primary distribution prompt instruction, determine target store information in the candidate store information according to the target distribution prompt instruction by using a generative large model, and feed back the target store information to the customer distribution unit.

[0031] In some cases, the store distribution of customer resource information is implemented by using a generative large model. Although large-scale sample training has been performed in advance before the deployment of the large model, in order to improve the accuracy of the large model, it is still necessary to periodically fine-tune or retrain the model to adapt to new transaction scenarios or changing market environment. Such training and fine-tuning need to be performed once every interval, which makes the process time-consuming and costly. Furthermore, since the model fine-tuning or retraining is periodic, it is difficult to achieve real-time or quasi-real-time evaluation of the current service capacity and conversion performance of the stores, which affects the accuracy of the store selection decision and makes it difficult to quickly adjust the target store information according to the real-time or dynamic actual situation of the stores.

[0032] In the present embodiment, the customer resource information distribution system obtains target customer resource information through the customer resource information collection unit, so as to further distribute the customer resource information to the stores. The target customer resource information can include basic attributes, geographic location, browsing behavior, consumption history, and possible store-willingness of the customers, etc. Specifically, the target customer resource information can come from multiple channels, such as page records browsed by the user, contact information left by the user through platform consultation customer service, order information or reservation records of consumption of medical related products, etc. In some embodiments, the target customer resource information can be provided by the client, for example, browsing and interaction data generated by the user through the mobile phone App, or automatically collected and provided by the server side service.

[0033] In the present embodiment, the customer distribution unit can be used to receive the target customer resource information and construct a primary distribution prompt instruction based on the target customer resource information. The primary distribution prompt instruction can be a structured task instruction containing the target customer resource information and the corresponding static distribution rule, which can be used to guide the artificial intelligence unit to determine the target store information of the target store suitable for receiving the target customer resource information. Specifically, after sending the primary distribution prompt instruction to the artificial intelligence unit, the customer distribution unit can receive the target store information fed back by the artificial intelligence unit, and send the target customer resource information to the store system corresponding to the target store based on the target store information.

[0034] In the embodiment, the artificial intelligence unit can be used to intelligently analyze the target customer resource information according to the primary distribution prompt instruction and determine the target store. Specifically, the artificial intelligence unit first matches the target customer resource information in the primary distribution prompt instruction from the pre-stored store information to obtain suitable candidate store information. The candidate store information refers to the basic information of a plurality of stores that may have a service matching relationship with the target customer resource information. For example, the candidate store information can include but is not limited to service items, geographic location, service capacity, supported appointment types, etc. At the same time, the artificial intelligence unit also obtains the current store evaluation information corresponding to each candidate store, which is used to represent the actual conversion performance of the candidate store for the received customer resource information. The current store evaluation information can include but is not limited to the following contents: customer store visit rate of the store, transaction rate after visiting the store, service completion condition, customer satisfaction score, refund / complaint rate, etc. The current store evaluation information can be reported by the store system and aggregated and analyzed by the customer resource collection system. The artificial intelligence unit can construct the target distribution prompt instruction based on the candidate store information, the corresponding current store evaluation information and the primary distribution prompt instruction, which is used to guide the generative large model to perform store selection reasoning.

[0035] In the embodiment, the generative large model is an artificial intelligence model with natural language understanding and context reasoning capability, which can evaluate the target store information of the more matched store according to the target customer resource information, candidate store information and current store evaluation information provided in the target distribution prompt instruction. The conversion performance of the store can be reflected by the current store evaluation information corresponding to the store, which not only reflects the service capacity of the store, but also quantifies the actual acceptance effect of the historical customer resource. The generative large model can refer to the current store evaluation information to filter and sort the candidate store information when making intelligent decision of the store, thereby enhancing the reasoning basis of target store selection. The artificial intelligence unit feeds back the target store information determined based on the above comprehensive analysis to the customer distribution unit, thereby completing the intelligent distribution process of the customer resource information.

[0036] In some embodiments, the artificial intelligence unit comprises a scheduling module, an RAG module, an RAG knowledge base, and the generative large model; the scheduling module is configured to receive the primary distribution prompt instruction, invoke the RAG module based on the target customer resource, obtain the candidate store information derived by the RAG module and the corresponding current store evaluation information, and construct a target distribution prompt instruction comprising the candidate store information; the RAG module is configured to derive candidate store information and corresponding current store evaluation information from store information in the RAG knowledge base according to the primary distribution prompt instruction; the RAG knowledge base is configured to store store information and corresponding current store evaluation information; and the generative large model is configured to receive the target distribution prompt instruction to determine target store information from the candidate store information.

[0037] In the present embodiment, the artificial intelligence unit can specifically include a scheduling module, an RAG module, an RAG knowledge base, and a generative large model. The scheduling module is configured to receive a primary distribution prompt instruction from the customer distribution unit, and invoke the RAG module according to the target customer resource information contained in the primary distribution prompt instruction to obtain candidate store information. Specifically, the scheduling module can generate a retrieval query for the RAG module based on the target customer resource information, guide the RAG module to perform an information retrieval task, so as to ensure that the candidate store information matched with the target customer resource information and the current store evaluation information expressing the actual conversion performance are obtained.

[0038] In the present embodiment, the RAG module can be configured to retrieve and filter candidate store information and corresponding current store evaluation information matched with the target customer resource information from the RAG knowledge base according to the received primary distribution prompt instruction. Specifically, the RAG module can use the Retrieval-Augmented Generation (RAG) technology to quickly find candidate store information highly matched with the target customer resource information in dimensions such as service range, service capability, geographic location, and historical conversion performance from a large amount of store information pre-stored in the RAG knowledge base, for subsequent decision-making.

[0039] In the present embodiment, the RAG knowledge base can be configured to store detailed information of multiple stores, including but not limited to geographic location information, service categories, historical service records, service capability ratings, and the like of each store. The RAG knowledge base stores store information in a structured manner, facilitating the RAG module to quickly and efficiently retrieve and invoke, so as to assist the scheduling module to accurately select candidate store information.

[0040] In the embodiment, the scheduling module further constructs a target distribution prompt instruction based on the primary distribution prompt instruction after obtaining the candidate store information provided by the RAG module and the current store evaluation information, which is used to guide the generative large model to intelligently complete the store selection. The target distribution prompt instruction can include the candidate store information, the current store evaluation information, and elements related to the target customer resource information.

[0041] The generative large model in the embodiment is used to receive the target distribution prompt instruction constructed by the scheduling module and intelligently perform an analysis task on the candidate store information based on the target distribution prompt instruction. The generative model performs semantic understanding and reasoning judgment based on the input candidate store information, the current store evaluation information, and the target customer resource features, so as to determine a target store with better matching degree and conversion potential. Subsequently, the generative large model feeds back the determined target store information to the scheduling module, and the scheduling module transmits the target store information to the customer distribution unit to complete the distribution process of the target customer resource information.

[0042] Through the cooperation between the modules of the artificial intelligence unit, and by taking the current store evaluation information as a dynamic feedback index, the embodiment not only improves the accuracy and real-time performance of the distribution decision, but also significantly enhances the adaptability between the target customer resource and the store capability, thereby improving the store conversion rate of the customer resource.

[0043] In some embodiments, the customer resource information distribution system further includes a customer feedback unit configured to receive feedback information corresponding to the target customer resource information fed back by the store system; the RAG knowledge base stores the feedback information corresponding to the store information; wherein the feedback information is used to represent the actual store transaction situation of the target customer resource information sent to the store system; and the target distribution prompt instruction constructed by the customer distribution unit further includes current store evaluation information generated based on the feedback information.

[0044] In the embodiment, the customer feedback unit is further provided to receive feedback information corresponding to the target customer resource information fed back by the store system. The feedback information can include actual store transaction situation data of the customer after the store system of the store receives the target customer resource information. For example, the feedback information includes whether the customer actually goes to the store, whether the customer completes the consumption, the actual transaction amount, the customer satisfaction score, or the reason for non-transaction, and the like. By collecting and processing the feedback information in real time or quasi-real time through the customer feedback unit, the embodiment can grasp the new service capability of the store in real time or quasi-real time, and provide dynamic and objective data basis for store evaluation and target customer resource information distribution decision.

[0045] Specifically, after receiving the feedback information returned by the store system, the customer information feedback unit in this embodiment updates the feedback information in real time to the corresponding store information in the RAG knowledge base to ensure the real-time and accuracy of the RAG knowledge base data. As a result, the RAG knowledge base not only stores static store basic information, but also includes dynamically updated feedback information that can truly reflect the store's recent actual conversion effect on customer resources. The continuously updated feedback information mechanism provided in this embodiment enables the customer resource information distribution system to timely understand the store's current performance and provides a reliable data basis for generating current store evaluation information.

[0046] In this embodiment, when constructing the target distribution prompt instruction, the customer information distribution unit can generate the current store evaluation information based on the feedback information updated in real time in the RAG knowledge base, and include the current store evaluation information in the target distribution prompt instruction to guide the generative large model to more accurately select the target store information.

[0047] In some embodiments, the primary distribution prompt instructions constructed by the customer resource distribution unit include static distribution rules; the RAG knowledge base is also used to receive and store dynamic distribution rules; the target distribution prompt instructions constructed by the customer distribution unit include the static distribution rules, and the dynamic distribution rules corresponding to the target customer resource information.

[0048] In some cases, while RAG modules and generative big models can enable intelligent store distribution of customer resource information, relying solely on fixed, pre-defined static distribution rules to make distribution decisions may not be able to fully adapt to the differentiated service scenarios and ever-changing market conditions reflected by different customer resource information. Static distribution rules are typically general rules based on fixed transaction logic and customer characteristics, and cannot be adjusted in a timely manner to dynamic changes in stores and fluctuations in customer preferences. This can lead to insufficient distribution accuracy and affect the actual conversion of customer resources to stores.

[0049] To address this issue, in this implementation, the customer information distribution unit, in addition to including target customer resource information, also includes static distribution rules when constructing preliminary distribution prompts. These rules are a set of predefined rules, including but not limited to matching customer location with store coverage, and matching customer behavior with store service categories. These rules are predefined and serve as initial guidance for the RAG module in its initial screening of candidate stores.

[0050] In addition, the RAG knowledge base in the embodiment is further configured to receive and store dynamic distribution rules. The dynamic distribution rules refer to distribution rules that are dynamically set. The management end can dynamically update the distribution rules to the RAG knowledge base, so as to conveniently adjust the distribution rules of the generative model by using the RAG technology. Specifically, the dynamic distribution rules can be adjusted periodically or in real time according to the actual in-store transaction conditions, the changes in the phased service capability performance, the real-time passenger flow conditions and other information fed back by the stores in real time, so as to ensure that the distribution decision can timely respond to the actual performance of the stores and the changes in the customer intention.

[0051] In the embodiment, the artificial intelligence unit further constructs the target distribution prompt instruction, and in addition to the static distribution rules, the dynamic distribution rules corresponding to the target customer resource information obtained from the RAG knowledge base are also included. The dynamic distribution rules can be combined with the real-time feedback information and the store evaluation information to realize more refined and personalized distribution decisions. By embedding the dynamic distribution rules in the target distribution prompt instruction, the generative large model can be guided to make decisions based on the current market environment and the store service performance data, and the accuracy of the customer resource information distribution is further improved.

[0052] In some embodiments, the customer information collection unit generates intention strength information for the target customer resource information according to user behaviors; wherein the intention strength information is used to represent the degree of willingness of the actual in-store ordering of the customer represented by the corresponding target customer resource information; and the primary distribution prompt instruction and the target distribution prompt instruction both include the intention strength information of the target customer resource information.

[0053] In the embodiment, the customer information collection unit is further optimized, so that it can generate corresponding intention strength information for the target customer resource information according to the browsing behaviors, interaction data and other behavior characteristics of the user. Specifically, the intention strength information can be a quantitative index used to represent the degree of willingness of the actual in-store consumption or ordering of the customer. For example, the intention strength information can be comprehensively analyzed and scored based on the browsing frequency, page dwell time, consultation times or shopping cart placement of the user, so as to more accurately represent the consumption willingness level of the customer.

[0054] In the embodiment, the primary distribution prompt instruction can include the intention strength information generated by the customer information collection unit, and the target customer resource information is input as a customer characteristic. In this way, when the RAG module is preliminarily guided to perform candidate store screening, the intention strength information of the customer can be used to realize more accurate preliminary screening and improve the matching quality of the candidate store information.

[0055] Further, the artificial intelligence unit also includes the intention strength information when constructing the target distribution prompt instruction, so that the generative large model can comprehensively refer to the real store-willingness degree of the customer when analyzing and determining the target store information. Specifically, the generative large model can make a comprehensive decision based on the intention strength information and the service ability, historical conversion performance, and real-time store evaluation information of the candidate store, so as to ensure that the target customer resource information is accurately distributed to the store with better conversion possibility.

[0056] Through the intention strength information generated by the above-mentioned customer information collection unit, the embodiment can realize a more refined customer resource information distribution strategy, so that the store selection decision is more in line with the real consumer demand of the customer, thereby further improving the accuracy of the actual store conversion of the customer resource.

[0057] In some embodiments, the current store evaluation information includes a store transaction rate; and the generative large model, for target customer resource information with a high value of the intention strength information, preferentially determines the candidate store information with a high value of the store transaction rate as the corresponding target store information.

[0058] In the embodiment, the intention strength information is a quantitative index generated by the customer information collection unit based on user behavior data, which is used to represent the willingness degree of the user represented by the customer resource information to actually go to the store and place an order. The generation of the intention strength information can be realized by weighted fusion of multi-dimensional behavior characteristics, such as the frequency of browsing product detail pages, the length of stay, the frequency of clicking the reservation button, the number of historical consultation records, whether to join the collection or shopping cart, and the like. The customer resource information distribution system can weight and score these behavior characteristics based on a rule model or a machine learning model to generate a standardized intention score (such as a floating-point value between 0 and 1), which is the intention strength information. For example, for a customer who has repeatedly browsed and stayed for a long time on the teeth whitening service page on the platform, and has repeatedly inquired the customer service about the specific price and reservation method, and has repeatedly added the service to the shopping cart, the customer information collection unit of the embodiment can determine that the customer has a high intention strength for the teeth whitening service, and thus gives a high intention strength information score (such as 8 points, with a full score of 10 points).

[0059] In the embodiment, the store transaction rate can be an evaluation index generated by the store transaction data collected by the customer resource feedback unit, which is used to measure the proportion of customers who successfully visit the store and make transactions after receiving customer resource information. The store transaction rate can be updated in real time through the feedback information reported by the store system, including whether the customer visits the store as scheduled, whether the customer actually receives the service, the amount of consumption, the satisfaction score, and other information. The customer resource information distribution system can continuously count the historical recommendation-visit data of the store in the RAG knowledge base and calculate the store transaction rate. For example, in a certain store, among the 100 customer resource information received in the past week, 70 customers actually visited the store and completed the consumption, so the historical store transaction rate of the store is 70%.

[0060] In the embodiment, when performing the store selection decision task, the generative large model can prioritize the candidate store information differently based on the value of the intention strength information corresponding to the target customer resource information. Specifically, for customer resources with high intention strength information values, the generative large model will preferentially select candidate stores with high store transaction rates as target store information, thereby improving the actual transaction probability of high-willing customers; for customer resources with low intention strength, stores with relatively relaxed transaction rate requirements and more flexible service capabilities can be selected for more exploratory or nurturing recommendations. In the embodiment, the determination method of the target store information can be indicated by setting the corresponding content in the dynamic distribution rule. Of course, in some embodiments, the determination method of the target store information can also be achieved by training the generative large model.

[0061] In some embodiments, the customer information collection unit obtains customer resource information through multiple resource paths; the customer information collection unit constructs information extraction prompt instructions according to the customer resource information, and calls the artificial intelligence unit with the information extraction prompt instructions; the scheduling module inputs the information extraction prompt instructions to the generative large model to obtain target customer resource information, and feeds back the target customer resource information to the customer information collection unit.

[0062] In the embodiment, to further improve the completeness of the collection of customer resource information, the customer information collection unit is configured to support a multi-resource path data access mechanism. Specifically, the resource paths include but are not limited to user online consultation dialogues, registration form filling, historical appointment records, third-party platform authorized data, intelligent device collected data, etc. Through multi-source path access, the customer information collection unit can more comprehensively obtain structured and unstructured original information related to customers to form original customer resource information.

[0063] In the embodiment, the customer information collection unit constructs an information extraction prompt instruction based on the customer resource information. The information extraction prompt instruction is a prompt information for guiding the artificial intelligence unit to complete a specific customer resource structured extraction task, and generally includes customer resource information, structured field targets, task context limitations, and the like. The customer information collection unit can dynamically construct personalized information extraction prompt instructions according to the characteristics of the customer resource information generated by different resource paths, and call the artificial intelligence unit for processing.

[0064] In the embodiment, after receiving the information extraction prompt instruction, the scheduling module inputs the information extraction prompt instruction to the generative large model to extract the customer resource information. The generative large model can identify key fields in the customer resource information based on the information extraction prompt instruction, such as the basic attributes of the customer, the geographic location, the browsing behavior, the consumption history, and the possible willingness to visit the store.

[0065] In the embodiment, the customer information collection unit can structure and store the target customer resource information, and use the target customer resource information as information input for subsequent construction of a primary distribution prompt instruction.

[0066] Referring to Figure 2 The embodiment of the present application also provides a customer resource information distribution method. The customer resource information distribution method can include the following steps.

[0067] Step S110: Obtain target customer resource information.

[0068] Step S120: Construct a primary distribution prompt instruction including the target customer resource information, use the primary distribution prompt instruction to call an artificial intelligence unit to determine target store information, and send the target customer resource information to a store system corresponding to the target store information; wherein the artificial intelligence unit is configured to obtain candidate store information according to the target customer resource information in the primary distribution prompt instruction, construct a target distribution prompt instruction based on the candidate store information and the primary distribution prompt instruction, and use a generative large model to determine the target store information from the candidate store information according to the target distribution prompt instruction.

[0069] In the embodiment, the functions and effects of the customer resource information distribution method can be explained in combination with the foregoing embodiments, and will not be repeated here.

[0070] Referring to Figure 3 The embodiment of the present application also provides a computer device, which includes a memory and a processor. The memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor to implement the method as described above.

[0071] The memory, the processor and the communication interface in the computer device can communicate with each other through a system bus, and network communication.

[0072] In the embodiment, the functions and effects realized by the computer device can be explained by referring to the foregoing embodiment, and will not be repeated.

[0073] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to enable the processor to implement the method as described above.

[0074] In the embodiment, the functions and effects realized can be explained by referring to other embodiments, and will not be repeated.

[0075] The embodiment of the present application further provides a computer program product comprising instructions, which are executed by a processor to implement the method as described above.

[0076] In the embodiment, the functions and effects realized can be explained by referring to other embodiments, and will not be repeated.

[0077] It can be understood that the specific examples herein are only to help those skilled in the art better understand the embodiments of the present application, and do not limit the scope of the present application.

[0078] It can be understood that in various embodiments of the present application, the size of the serial number of each process does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0079] It can be understood that the various embodiments described in the present application can be implemented alone or in combination, and the embodiments of the present application do not limit this.

[0080] Unless otherwise specified, all technical and scientific terms used in the embodiments of the present application have the same meanings as those commonly understood by those skilled in the art of the present application. The terms used in the present application are only for the purpose of describing the specific embodiments and are not intended to limit the scope of the present application. The term "and / or" used in the present application includes any and all combinations of one or more related listed terms. The singular forms "a", "an" and "the" used in the embodiments of the present application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise.

[0081] It can be understood that the processor of the embodiments of the present application can be an integrated circuit chip with processing capability. In the implementation process, each step of the method embodiments described above can be completed by integrated logic circuits in hardware or instructions in software form in the processor. The processor described above can be a general processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. Each method, step and logic block diagram disclosed in the embodiments of the present application can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as a hardware code processor for execution, or a combination of hardware and software modules in the code processor for execution. The software module can be located in a random access memory, a flash memory, a read only memory, a programmable read only memory or an electrically erasable programmable memory, a register or other mature storage medium in the art. The storage medium is located in the memory, and the processor reads the information in the memory, and combines the hardware to complete the steps of the above method.

[0082] It can be understood that the memory in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read only memory (ROM), a programmable read only memory (PROM), an erasable programmable read only memory (EPROM), an electrically erasable programmable read only memory (EEPROM) or a flash memory. The volatile memory can be a random access memory (RAM). It should be noted that the memory of the system and method described herein is intended to include but not limited to these and any other suitable type of memory.

[0083] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0084] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0085] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the above-described device embodiments are merely schematic, and the division of the units is merely a logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between the units can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0086] The units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments.

[0087] In addition, each functional unit in the embodiments of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit.

[0088] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the parts that make contributions to the prior art or the parts of the technical solutions can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the embodiments of the present application. The foregoing storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, and various media that can store program codes.

[0089] The above describes only specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A customer resource information distribution system, characterized in that: include: Customer information collection unit, used to obtain target customer resource information; A customer resource distribution unit is configured to construct a primary distribution prompt instruction including target customer resource information, use the primary distribution prompt instruction to call the artificial intelligence unit to determine target store information, and send the target customer resource information to the store system corresponding to the target store information; An artificial intelligence unit is used to obtain candidate store information and current store evaluation information corresponding to the candidate store information based on the target customer resource information in the primary distribution prompt instruction, construct a target distribution prompt instruction based on the candidate store information, the current store evaluation information and the primary distribution prompt instruction, and use a generative large model to determine the target store information in the candidate store information according to the target distribution prompt instruction, and feed back the target store information to the customer information distribution unit; wherein, the current store evaluation information is used to characterize the conversion effect of the corresponding candidate store on the customer resource information that has been received.

2. The customer resource information distribution system according to claim 1, characterized in that: The artificial intelligence unit includes: a scheduling module, a RAG module, a RAG knowledge base and the generative large model; The scheduling module is configured to receive the primary distribution prompt instruction, call the RAG module based on the target customer resources, obtain the candidate store information and corresponding current store evaluation information obtained by the RAG module, and construct a target distribution prompt instruction including the candidate store information; The RAG module is configured to obtain candidate store information and corresponding current store evaluation information from the store information in the RAG knowledge base according to the primary distribution prompt instruction; The RAG knowledge base is used to store store information and corresponding current store evaluation information; The generative large model is used to receive the target distribution prompt instruction to determine the target store information from the candidate store information.

3. The customer resource information distribution system according to claim 2, characterized in that: It also includes a customer information feedback unit, which is used to receive feedback information corresponding to the target customer resource information fed back by the store system; The RAG knowledge base stores the feedback information corresponding to the store information; wherein the feedback information is used to represent the actual in-store transaction status of the target customer resource information sent to the store system; The target distribution prompt instruction constructed by the customer information distribution unit also includes current store evaluation information generated based on the feedback information.

4. The customer resource information distribution system according to claim 2, characterized in that: The primary distribution prompt instruction constructed by the customer information distribution unit includes static distribution rules; The RAG knowledge base is also used to receive and store dynamic distribution rules; The target distribution prompt instruction constructed by the artificial intelligence unit includes the static distribution rules and the dynamic distribution rules corresponding to the target customer resource information.

5. The customer resource information distribution system according to claim 3, characterized in that: The customer information collection unit generates intention strength information for the target customer resource information based on user behavior; wherein the intention strength information is used to indicate the degree of willingness of the customer represented by the corresponding target customer resource information to actually place an order in the store; The primary distribution prompt instruction and the target distribution prompt instruction both include the intention strength information of the target customer resource information.

6. The customer resource information distribution system according to claim 5, characterized in that: The current store evaluation information includes the store transaction rate; the generative large model prioritizes candidate store information with high store transaction rate values ​​as the corresponding target store information for target customer resource information with high intention strength information values.

7. The customer resource information distribution system according to claim 2, characterized in that: The customer information collection unit obtains customer resource information through multiple resource paths; The customer information collection unit constructs an information extraction prompt instruction based on the customer resource information, and calls the artificial intelligence unit with the information extraction prompt instruction; The scheduling module inputs the information extraction prompt instruction into the generative large model to obtain target customer resource information, and the scheduling module feeds back the target customer resource information to the customer information collection unit.

8. A method for distributing customer resource information, characterized in that: include: Obtain target customer resource information; Constructing a primary distribution prompt instruction including target customer resource information, using the primary distribution prompt instruction to call an artificial intelligence unit to determine target store information, and sending the target customer resource information to a store system corresponding to the target store information; Among them, the artificial intelligence unit is used to obtain candidate store information and current store evaluation information corresponding to the candidate store information based on the target customer resource information in the primary distribution prompt instruction, construct a target distribution prompt instruction based on the candidate store information, the current store evaluation information and the primary distribution prompt instruction, and use a generative large model to determine the target store information in the candidate store information according to the target distribution prompt instruction; wherein, the current store evaluation information is used to characterize the conversion effect of the corresponding candidate store on the customer resource information that has been received.

9. The method according to claim 8, characterized in that The artificial intelligence unit includes: scheduling module, RAG module, RAG knowledge base and generative large model; The scheduling module is configured to receive the primary distribution prompt instruction, call the RAG module based on the target customer resources, obtain the candidate store information and corresponding current store evaluation information obtained by the RAG module, and feed back the candidate store information to the customer information distribution unit; The RAG module is configured to obtain candidate store information and corresponding current store evaluation information from the store information in the RAG knowledge base according to the primary distribution prompt instruction; The RAG knowledge base is used to store store information and corresponding current store evaluation information.

10. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor to implement the method according to claim 8 or 9.

11. A computer program product, characterized in that Comprising computer instructions for implementing the method according to claim 8 or 9.

Images