Intelligent interaction method and device based on AI large model logistics data and storage medium

Abstract

The invention relates to an intelligent interaction method and device based on AI large model logistics data and a storage medium. The method comprises the following steps: acquiring business data in a logistics system, and converting the business data into standard business data; analyzing and identifying a system function in the logistics system to generate structured description information; in response to the received service query request, performing semantic recognition on the service query request, and generating prompt information; and obtaining a target prediction result based on at least one of the standard business data, the structured description information and the prompt information. According to the method, multi-dimensional data such as standardized business data, structured description information and prompt information are fused, logistics total elements can be covered, accurate demand prediction is realized, a user is assisted in rapid decision making, and the demands of query, calculation and / or scheduling in a logistics full scene are met.

Description

Technical Field

[0001] This application relates to the technical field of logistics management, and particularly to an intelligent interaction method, device, and storage medium based on AI large model logistics data. Background Art

[0002] In the context of the rapid development of global digitalization and the logistics field today, the research on intelligent logistics processing methods has become crucial. With the growth of e-commerce and supply chain management, the logistics industry is facing increasingly complex challenges, such as rapidly growing orders, inventory management, logistics path optimization, etc. How to effectively allocate resources in different logistics transfer yards in a coordinated manner and how to plan the path have gradually become the key to improving logistics efficiency, ensuring the smooth flow of goods, and customer satisfaction.

[0003] However, traditional logistics management and scheduling methods are often limited by human resources and time costs and cannot meet the growing demand. Therefore, how to efficiently and accurately perform processing such as querying / calculating logistics data to meet logistics scheduling requirements is an urgent problem to be solved. Summary of the Invention

[0004] Based on this, in view of the above technical problems, it is necessary to provide an intelligent interaction method, device, and storage medium based on AI large model logistics data.

[0005] In a first aspect, this application provides an intelligent interaction method based on AI large model logistics data, and the method includes: Obtain business data in the logistics system and convert the business data into standard business data; Parse and identify system functions in the logistics system to generate structured description information; In response to receiving a business query request, perform semantic recognition on the business query request to generate a prompt message; Based on at least one of the standard business data, the structured description information, and the prompt message, obtain a target prediction result.

[0006] In one embodiment, the converting the business data into standard business data includes: Perform semantic recognition on the business data to extract target entities; wherein, the target entities include at least one of order entities, transportation entities, inventory entities, and equipment entities; Create a data view that matches the entity type of the target entity according to the target entity; wherein, the data view includes at least one of a global inventory view, a risk warning view, and a real-time transport capacity view; Based on foreign key constraints, associate each data view to obtain the standard business data.

[0007] In one embodiment, parsing and identifying system functions in the logistics system to generate structured description information includes: Parsing and identifying the code corresponding to the system function to obtain function description information; wherein, the system function includes at least one of a path optimization function, an inventory prediction function, and a transportation capacity scheduling function; the function description information includes a function name, input parameters, and functional logic; Performing standardization processing on the function description information to obtain the structured description information; wherein, the structured description information includes a function name, input requirements, and processing logic.

[0008] In one embodiment, obtaining a target prediction result based on at least one of the standard business data, the structured description information, and the prompt information includes: Matching the prompt information with the structured description information to obtain a matching result; When the matching result is a successful match, inputting the prompt information, the standard business data matching the prompt information, and the structured description information into an AI large model to obtain a candidate result; Executing the system function matching the prompt information and combining it with the candidate result to obtain the target prediction result.

[0009] In one embodiment, the method further includes: When the matching result is a failed match, inputting the prompt information and the standard business data matching the prompt information into an AI large model to obtain the target prediction result.

[0010] In one embodiment, executing the system function matching the prompt information and combining it with the candidate result to obtain the target prediction result includes at least one of the following: First, when executing the path optimization function matching the prompt information, obtaining a set of predicted paths based on an improved ant colony algorithm; obtaining the target prediction result based on the set of predicted paths and the candidate result; Second, when executing the inventory prediction function matching the prompt information, obtaining an inventory prediction result based on the Prophet model; obtaining the target prediction result based on the inventory prediction result and the candidate result; Third, when executing the transportation capacity scheduling function matching the prompt information, obtaining a matching matrix between vehicles and goods based on a mixed integer programming algorithm; obtaining the target prediction result based on the matching matrix and the candidate result.

[0011] In one of the embodiments, the method further includes: Upon receiving feedback information on the target prediction result, when the evaluation score included in the feedback information is less than the qualified threshold, obtain modification information for the prompt information; Update the prompt information based on the modification information to obtain updated prompt information; Obtain an updated target prediction result based on at least one of the standard business data, the structured description information, and the updated prompt information.

[0012] In a second aspect, the present application further provides an intelligent interaction device based on AI large model logistics data, and the device includes: A conversion module, configured to obtain business data in a logistics system and convert the business data into standard business data; A first generation module, configured to parse and identify system functions in the logistics system to generate structured description information; A second generation module, configured to, in response to receiving a business query request, perform semantic recognition on the business query request to generate prompt information; A determination module, configured to obtain a target prediction result based on at least one of the standard business data, the structured description information, and the prompt information.

[0013] In a third aspect, the present application further provides a computer device, including a processor and a memory for storing a computer program of the processor; wherein, the processor is configured to: when executing the computer program, implement the steps performed by the method in any embodiment of the present application.

[0014] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, and when the computer program is executed by a processor, it implements the steps performed by the method in any embodiment of the present application.

[0015] In the above intelligent interaction method based on AI large model logistics data, standardizing business data can eliminate data differences between different systems (corresponding to different application scenarios), platforms, or devices, facilitating subsequent processing and analysis of business data; parsing system functions and generating structured description information helps to automatically understand and call relevant functional modules, improving the intelligent scheduling ability of the logistics system; fusing multi-dimensional data such as standardized business data (such as historical sales volume, supply chain delivery date, etc.), structured description information (such as promotion rules, vehicle scheduling algorithms, etc.), and prompt information can cover all elements of logistics, achieve accurate demand prediction, assist users in making quick decisions, and meet the requirements of querying, calculating, and / or scheduling in all logistics scenarios. Brief Description of the Drawings

[0016] Figure 1 is an application environment diagram of an intelligent interaction method based on AI large model logistics data shown according to an exemplary embodiment; Figure 2 is a schematic flowchart of an intelligent interaction method based on AI large model logistics data shown according to an exemplary embodiment; Figure 3 is a structural block diagram of an intelligent interaction device based on AI large model logistics data shown according to an exemplary embodiment; Figure 4 is an internal structure diagram of a computer device shown according to an exemplary embodiment. Detailed Description of the Embodiment

[0017] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0018] The terms "first", "second", and "third" in the embodiments of the present application are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first", "second", and "third" may explicitly or implicitly include at least one of such features. In the description of the present application, the meaning of "a plurality" is at least two, such as two, three, etc., unless otherwise specifically and clearly defined. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, method, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or devices.

[0019] Referring to "embodiment" herein means that a specific feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the present application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein may be combined with other embodiments.

[0020] In some embodiments, the intelligent interaction method based on AI large model logistics data provided in the embodiments of the present application can be applied as follows Figure 1In the application environment shown, the computer device 102 communicates with the server 104 via a network, and the server 104 communicates with the database 106 via a network; the computer device 102 can obtain in real time from the database 106 the business data that needs to be processed through the server 104. Among them, the computer device 102 can be any mobile terminal or fixed terminal. A terminal can be a device that provides voice and / or data connectivity to a user. Exemplarily, the terminal can be an Internet of Things terminal, such as a sensor device, a mobile phone or a so-called "cellular" phone, and a computer with an Internet of Things terminal. For example, it can be a fixed, portable, pocket-sized, handheld, or computer-built-in device.

[0021] In some embodiments, the intelligent interaction method based on AI large model logistics data provided in the embodiments of the present application can be applied to a computer device.

[0022] In some embodiments, as Figure 2 shown, an intelligent interaction method based on AI large model logistics data is provided, and the method includes the following steps: S201, obtain business data in the logistics system, and convert the business data into standard business data.

[0023] In the embodiments of the present application, the logistics system may include, but is not limited to, at least one of an order management system, a transportation management system, and a warehousing management system.

[0024] In the embodiments of the present application, the business data may include, but is not limited to, at least one of order data, inventory data, transportation data, vehicle and equipment data, personnel data, and customer service data.

[0025] In some embodiments, the computer device can standardize the original business data from different systems and in various formats into standard business data with a unified standard structure / format (such as coding standardization, consistent naming rules, etc.) to facilitate the effective storage, sharing, analysis, and application of the data.

[0026] S202, parse and identify the system functions in the logistics system to generate structured description information.

[0027] In the embodiments of the present application, the system functions in the logistics system may include, but are not limited to, at least one of a path optimization function, an inventory prediction function, and a capacity scheduling function. Different logistics systems may include one or more system functions; the system functions included in different logistics systems may be the same or different.

[0028] In some embodiments, the parsing and identifying the system functions in the logistics system to generate structured description information includes: Parse and identify the code corresponding to the system function to obtain function description information; wherein, the system function includes at least one of a path optimization function, an inventory prediction function, and a transportation capacity scheduling function; the function description information includes a function name, input parameters, and a functional logic. Perform a standardization process on the function description information to obtain the structured description information; wherein, the structured description information includes a function name, input requirements, and a processing logic.

[0029] In an embodiment of the present application, the function name is used to characterize the purpose or function of the function (such as "generate a transportation cost quote").

[0030] In an embodiment of the present application, the input requirements are used to clarify the parameter type (such as the parameter type corresponding to weight is a number) and the value range (such as the transportation mode includes truck / truck / airplane).

[0031] In an embodiment of the present application, the processing logic is used to indicate the calculation rule described in natural language (such as basic cost = weight × distance × unit price).

[0032] In some embodiments, the computer device can parse the system code through an abstract syntax tree, identify the input / output of the system function, and generate preliminary description information in combination with Javadoc / annotations; associate the system function with the corresponding business document to supplement the business logic description to obtain the function description information.

[0033] In some embodiments, the computer device automatically identifies the function name, input parameters, and functional logic by analyzing the code corresponding to the system function; for example, extract the parameters (cargo weight, distance, transportation mode / speed) of the freight calculation function and the corresponding pricing rule (such as a 30% surcharge for cold chain); perform a standardization process on the function description information to obtain the structured description information.

[0034] In one embodiment, the computer device can configure the processing logic with a complexity greater than a predetermined threshold (such as time-of-day freight, regional pricing strategy, etc.) as a settable parameter, so that the system function can flexibly respond to market / policy changes (such as a 10% increase in freight during holidays).

[0035] S203, in response to receiving a service query request, perform semantic recognition on the service query request to generate a prompt message.

[0036] In some embodiments, the computer device uses an intent classification model in the field of Natural Language Processing (NLP) to perform semantic recognition processing on the speech data or text data corresponding to the business query request, obtains the semantic recognition result, and obtains the intent classification of the business query request based on the semantic recognition result, that is, the user requirement; according to the intent classification, a prompt message including the business background, data range, and operation options is generated; for example, the prompt message can be "The user needs to query the fresh food transportation status from place A to place B."

[0037] In some embodiments, the computer device trains the intent classification model based on the training sample data with marked intent classification. For example, the intent classification can include but is not limited to computing transportation, AI cost, querying transportation status, etc.; during the training process, the cross-entropy loss function can be used to iteratively optimize the model, and the model parameters of the intent classification model are updated through the backpropagation algorithm to improve the accuracy of intent classification.

[0038] S204, obtain a target prediction result based on at least one of the standard business data, the structured description information, and the prompt message.

[0039] In some embodiments, the computer device can input one or more of the standard business data, the structured description information, and the prompt message into an Artificial Intelligence (AI) large model to obtain a target prediction result for the prompt message (i.e., the user requirement) in real time.

[0040] In the above intelligent interaction method for logistics data based on an AI large model, standardizing the business data can eliminate data differences between different systems (corresponding to different application scenarios), platforms, or devices, facilitating subsequent processing and analysis of the business data; parsing the system functions and generating structured description information helps automate the understanding and invocation of relevant functional modules, improving the intelligent scheduling ability of the logistics system; fusing multi-dimensional data such as the standardized business data (such as historical sales volume, supply chain delivery date, etc.), the structured description information (such as promotion rules, vehicle scheduling algorithms, etc.), and the prompt message can cover all logistics elements, achieve accurate demand prediction, assist users in making quick decisions, and meet the requirements of querying, computing, and / or scheduling in all logistics scenarios.

[0041] In some embodiments, the conversion of the business data into standard business data includes: Performing semantic recognition on the business data to extract target entities; where the target entities include at least one of order entities, transportation entities, inventory entities, and equipment entities; Creating a data view matching the entity type of the target entity according to the target entity; wherein the data view includes at least one of a global inventory view, a risk warning view, and a real-time transportation capacity view; Based on foreign key constraints, various data views are associated to obtain standard business data.

[0042] In the embodiment of the present application, the foreign key constraint is a mechanism used in a relational database to maintain the referential integrity between different tables / data views.

[0043] In some embodiments, the computer device may input raw business data into a naming recognition model, capture contextual information of the business data through a multi-layer Transformer encoder, and use a conditional random field (CRF) in the output layer to predict and label target entities and categories.

[0044] Optionally, the order entity may include but is not limited to at least one of the fields of order number, customer information, goods details and timeliness requirements.

[0045] Optionally, the transport entity may include but is not limited to at least one field of waybill number, carrier, vehicle information, personnel information and positioning information.

[0046] Optionally, the inventory entity may include but is not limited to at least one field of a storage location code, a minimum inventory unit (Stock Keeping Unit, SKU), a batch, an expiration date, an inventory quantity, and temperature / humidity data.

[0047] Optionally, the equipment entity may include but is not limited to at least one field of an equipment tag, equipment load, and a shelf radio frequency identification (RFID) tag.

[0048] In one embodiment, the computer device can use data modeling technology to perform entity-relationship analysis on data tables in the logistics system to extract target entities; and generate data views based on the target entities and database view creation technology.

[0049] In one embodiment, the target entity is an inventory entity, and the data view that matches the entity type of the inventory entity is a global inventory view. The global inventory view can be associated with the reservation record in the order tracking view by foreign key according to the fields such as the available quantity, the quantity in transit, and the locked quantity of the minimum stock keeping unit (SKU).

[0050] In one embodiment, if the target entity is a transportation entity, the data view matching the entity type of the transportation entity is a real-time transport capacity view. The real-time transport capacity view can, based on fields such as vehicle location, remaining load, route planning, and driver working hours, externally key-constrain the allocation status of the transport task table.

[0051] In one embodiment, when the target entity includes an inventory entity, a transportation entity, and a device entity, the data view matching the entity type of the target entity can be a risk warning view. The risk warning view can integrate fields such as inventory expiration date, transport delay records, and device failure history.

[0052] In the embodiments of the present application, by performing semantic recognition on the original business data to extract the target entities, and then converting these entities into data frame titles in a unified format, the business data from different sources can be made logically consistent, facilitating subsequent analysis and integration; by creating and displaying data views, the visualization ability of business data can be improved. After new functions, new modules, or new data sources are accessed, they can be quickly integrated with the existing parts, enhancing the scalability of the system.

[0053] In some embodiments, obtaining the target prediction result based on at least one of the standard business data, the structured description information, and the prompt information includes: Matching the prompt information with the structured description information to obtain a matching result; When the matching result is a successful match, inputting the prompt information, the standard business data matching the prompt information, and the structured description information into the AI large model to obtain a candidate result; Executing the system function matching the prompt information and combining it with the candidate result to obtain the target prediction result.

[0054] In some embodiments, the computer device matches the intent classification corresponding to the prompt information with the function name field in the structured description information to determine whether the matching result is a successful match; when the intent classification and the function name field match successfully, inputting the prompt information, the standard business data (i.e., the data view) matching the prompt information, and the structured description information matching the prompt information into the AI large model to obtain a candidate result; using the AI large model to call the system function matching the prompt information to obtain a second result; based on the candidate result and the second result, obtaining the target prediction result.

[0055] In the embodiments of the present application, by matching the prompt information with the structured description information, both can be input into the AI large model, providing more accurate input for the AI large model, making the output candidate results more in line with the actual situation, thereby improving the accuracy of the target prediction results; in the case of successful matching, the corresponding system function can be executed to make appropriate responses for different scenarios, enhancing the flexibility of the logistics system and its adaptability to diverse requirements. Moreover, by determining the target prediction result based on the candidate result and the second result (dual results), compared with single input and single output, the accuracy of the target prediction result can be further improved.

[0056] In some embodiments, the method further includes: In the case where the matching result is a failure, input the prompt information and the standard business data matching the prompt information into the AI large model to obtain the target prediction result.

[0057] In some embodiments, when the matching result is a failure, the computer device can directly input the prompt information and the data view matching the prompt information into the AI large model, and the output result of the model is the target prediction result.

[0058] In the embodiments of the present application, in the case where the matching result is a failure, even if the structured description information and the system function fail to match successfully, the prompt information and the standard business data still contain valuable information; inputting this information into the AI large model can reduce the information waste caused by the matching failure, and the AI large model can still make reasonable and accurate predictions based on the prompt information and the standard business data, providing guarantee for the user experience.

[0059] In some embodiments, the execution of the system function matching the prompt information and combining the candidate result to obtain the target prediction result includes at least one of the following: First, when executing the path optimization function matching the prompt information, based on the improved ant colony algorithm, obtain a set of predicted paths; based on the set of predicted paths and the candidate result, obtain the target prediction result; Second, when executing the inventory prediction function matching the prompt information, based on the Prophet model, obtain the inventory prediction result; based on the inventory prediction result and the candidate result, obtain the target prediction result; Third, when executing the transportation capacity scheduling function matching the prompt information, based on the mixed integer programming algorithm, obtain the matching matrix between vehicles and goods; based on the matching matrix and the candidate result, obtain the target prediction result.

[0060] In the embodiments of the present application, the path optimization function may include at least one parameter among, but not limited to, the starting point coordinates, the list of end points, the vehicle load, the traffic conditions, and the cost model (fuel consumption / hour consumption).

[0061] In some embodiments, the computer device may obtain dynamic road condition data through the traffic condition API, construct a time series database of the road network status, and update the passage time of road segments according to the interval time; use the cost of the path (including fuel consumption, road and bridge tolls, driver's salary, etc.) and the time (estimated arrival time) as the optimization objectives, and each path as a function solution (including the sequence of nodes passed and the corresponding transportation cost and timeliness); calculate the probability of selecting the next node j at node i, and retain the non-dominated solutions in each iteration of calculating the probability based on the Pareto (Pareto Dominance) dominance relationship to form a dynamically updated solution set, that is, the predicted path set.

[0062] Exemplarily, the computer device calculates the probability of selecting the next node j at node i through the ant colony algorithm as follows:

[0063] Wherein, Indicates the concentration of pheromone related to the accumulated cost on the edge (road segment) from i to j, reflecting the pros and cons of the economic cost of this road segment in historical path selection; Indicates the concentration of pheromone related to the accumulated time on the edge (road segment) from i to j, reflecting the time efficiency (such as passing speed, etc.) of this road segment in historical path selection; Indicates the heuristic factor of the edge (road segment) from i to j, determined by the current road conditions; Indicates a hyperparameter used to control the weight of cost pheromone; Indicates a hyperparameter used to control the weight of time pheromone; Indicates a hyperparameter used to control the weight of the heuristic factor.

[0064] In one embodiment, when the computer device obtains a new order, it parses the order, extracts the pick-up point, delivery point, time window, and cargo attributes (volume / weight) of the order; traverses the paths in the existing solution set to identify the vehicles that conflict with the new order (such as overlapping paths and insufficient capacity); searches for the best insertion point in the path to minimize the cost and time changes after adding the new path / stoppage point; if the target value is improved after inserting the new path (such as cost reduction or timeliness shortening), enhance the pheromone of the relevant edge, or, for the paths that are overloaded or overtime due to order insertion, volatilize their pheromones to reduce the subsequent selection probability.

[0065] In the embodiments of the present application, the inventory prediction function may include at least one parameter among, but not limited to, historical sales volume (time series), supply chain delivery period, seasonal factor, and promotion factor.

[0066] In some embodiments, the computer device may extract historical sales data from the business system, align it at daily / weekly granularity, and clean outliers (such as smoothing the abnormal peaks during the promotion period); obtain the promotion plan data within a preset time, label the promotion type, covered SKUs, expected traffic increase rate, etc. for the promotion plan data; obtain the historical on-time delivery rate of suppliers, and dynamically calculate the safety inventory buffer days in combination with the current production capacity and logistics timeliness; extract the annual periodicity (such as holidays) and monthly fluctuations (such as the impact of the rainy season on goods) by fitting the historical data with Fourier series; establish a Prophet model to simulate the trend change according to the growth trend, periodic terms, and holiday effects, and automatically identify the mutation points by combining piecewise linear or logistic growth curves; calculate the inventory forecast result according to the Prophet model.

[0067] In the embodiments of the present application, the transportation capacity scheduling function may include at least one parameter such as the volume / weight of the goods to be transported, vehicle type (cold chain transportation / ordinary transportation), and delivery time window.

[0068] In one embodiment, the computer device obtains the goods attribute data, standardizes the goods attribute data, unifies the transportation unit and marks special goods (such as cold chain goods, vibration-sensitive goods); synchronizes the vehicle position, remaining capacity, and cold chain status in real time, and verifies the driver compliance; establishes a Mixed-Integer Programming (MIP) model according to the fixed cost (basic vehicle cost), variable cost (fuel consumption, road and bridge tolls, refrigeration energy consumption), penalty cost (time window deviation, customer priority adjustment), in combination with the constraint conditions; calculates the matching matrix between the vehicle and the goods based on the MIP model; wherein, the constraint conditions may include capacity constraints, time constraints, regulatory constraints, etc. For example, 8% of the emergency space needs to be reserved for the vehicle load as the capacity buffer; or, the driver's single-day driving must be less than or equal to eight hours, etc.

[0069] In the embodiments of the present application, the improved ant colony algorithm can effectively solve complex path planning problems; the Prophet model can accurately predict future inventory requirements according to historical data, meeting the processing requirements of datasets with seasonality or trend changes; the mixed integer programming algorithm can accurately calculate the best matching plan between vehicles and goods to ensure the effective utilization of resources. Different types of prediction tasks require different algorithms / functions to support. Determine the algorithm suitable for the current scenario according to the prompt information to improve the accuracy, reliability, and flexibility of the target prediction result.

[0070] In some embodiments, the method further includes: In response to receiving the feedback information of the target prediction result, when the evaluation score included in the feedback information is less than the qualified threshold, obtain the modification information of the prompt information; Update the prompt information based on the modification information to obtain the updated prompt information; Obtain an updated target prediction result based on at least one of the standard business data, the structured description information, and the updated prompt information.

[0071] In the embodiments of the present application, the evaluation score may include, but is not limited to, at least one of an accuracy evaluation score, a timeliness evaluation score, and a completeness evaluation score.

[0072] In one embodiment, the qualified thresholds corresponding to different evaluation scores may be the same or different. The computer device may adopt a sliding window algorithm to automatically adjust the qualified threshold according to the historical feedback information of a predetermined duration (such as seven days).

[0073] In some embodiments, after the computer device outputs the target prediction result, it may obtain the feedback information of the user on the target detection result, and the feedback information includes the evaluation scores of each index; when the evaluation score of any index is less than the qualified threshold, it may obtain the modification information of the user on the prompt information or generate corresponding modification information based on the index whose evaluation score is less than the qualified threshold; the modification information may include, but is not limited to, at least one of parameter adjustment, model replacement, and data correction; update the prompt information based on the modification information to obtain the updated prompt information.

[0074] Exemplarily, the feedback information is "the predicted transport capacity is 5, the transport capacity is insufficient, and the accuracy evaluation score is 70"; the qualified threshold corresponding to the accuracy evaluation score in the current scenario is 80, then the accuracy evaluation score is less than the qualified threshold 80, and the modification information "it is recommended to increase the parameter of the number of transport vehicles, but do not modify the time window" may be generated based on the feedback information; adjust the parameter of the number of transport vehicles based on the modification information, and keep other parameters unchanged to obtain the updated prompt information.

[0075] In the embodiments of the present application, by receiving the feedback information of the target prediction result and making corrections based on the feedback information, a closed-loop feedback mechanism is formed; when the target prediction result is not ideal, the prompt information can be automatically or assisted to optimize by the user according to the feedback information, so as to guide the AI large model to output results that more meet the actual needs and increase the adaptability to complex scenarios. And, the same type of prediction task may require different processing logics in different business scenarios. Through feedback-driven prompt optimization, different business needs can be met; for example, in logistics scheduling, some pay more attention to timeliness, and some pay more attention to cost. The priority can be highlighted by adjusting the prompt information through feedback to improve the user experience.

[0076] In the embodiments of the present application, in combination with any of the above embodiments, specific examples are provided as follows: Specific Example 1: The present application provides an example process for a computer device to implement an intelligent interaction method for logistics data based on an AI large model; when a processor in the computer device executes a computer program, the following steps are implemented: Use data modeling technology to perform entity-relationship analysis on the business data tables in the logistics system, and extract target entities; generate a data view for the target entities through database view creation technology; use foreign key constraint technology to establish relationships between views to obtain standard business data.

[0077] Generate structured description information for the system functions in the logistics system, where the structured description information includes function names, input requirements, and processing logic.

[0078] In response to receiving a business query request, perform demand analysis on the voice data or text data input by the user through an intent classification model to determine the corresponding intent classification; based on the intent classification, generate a prompt word representing the user's demand.

[0079] Synchronously input the standard business data, structured description information, and prompt information into the AI large model to obtain candidate results; match the prompt information with the system functions, and in the case of successful matching, execute the successfully matched system function to obtain a function result; based on the function result and the candidate results, obtain a target prediction result; in the case of failed matching, determine the candidate result output by the AI large model as the target prediction result.

[0080] In the embodiments of the present application, standardizing business data can eliminate data differences between different systems (corresponding to different application scenarios), platforms, or devices, facilitating subsequent processing and analysis of business data; parsing system functions and generating structured description information helps automate the understanding and invocation of relevant functional modules, enhancing the intelligent scheduling ability of the logistics system; fusing multi-dimensional data such as standardized business data (such as historical sales volume, supply chain delivery date, etc.), structured description information (such as promotion rules, vehicle scheduling algorithms, etc.), and prompt information can cover all elements of logistics, achieve accurate demand prediction, assist users in making quick decisions, and meet the query, calculation, and / or scheduling requirements in all logistics scenarios.

[0081] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are sequentially shown according to the indications of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0082] Based on the same inventive concept, an embodiment of the present application further provides an intelligent interaction device for AI large model logistics data for implementing the intelligent interaction method for AI large model logistics data involved above. The solution provided by this device for solving problems is similar to the solution described in the above method. Therefore, the specific limitations in one or more of the following embodiments of the intelligent interaction device for AI large model logistics data can refer to the limitations on the intelligent interaction method for AI large model logistics data in the above text, and will not be repeated here.

[0083] In one embodiment, as Figure 3 shown, an intelligent interaction device for AI large model logistics data is provided, and the device includes: A conversion module 10, configured to obtain service data in a logistics system and convert the service data into standard service data; A first generation module 20, configured to parse and identify system functions in the logistics system and generate structured description information; A second generation module 30, configured to perform semantic recognition on the service query request in response to receiving the service query request and generate prompt information; A determination module 40, configured to obtain a target prediction result based on at least one of the standard service data, the structured description information, and the prompt information.

[0084] In one embodiment, the conversion module 10 is configured to perform the following steps: Perform semantic recognition on the service data and extract target entities; wherein the target entities include at least one of order entities, transportation entities, inventory entities, and equipment entities; Create a data view that matches the entity type of the target entity according to the target entity; wherein the data view includes at least one of a global inventory view, a risk warning view, and a real-time transport capacity view; Associate each data view based on foreign key constraints to obtain the standard business data.

[0085] In one embodiment, the first generation module 20 is configured to perform the following steps: Parse and identify the code corresponding to the system function to obtain function description information; wherein, the system function includes at least one of a path optimization function, an inventory prediction function, and a transportation capacity scheduling function; the function description information includes a function name, input parameters, and functional logic. Perform standardization processing on the function description information to obtain the structured description information; wherein, the structured description information includes a function name, input requirements, and processing logic.

[0086] In one embodiment, the determination module 40 includes: A matching unit configured to match the prompt information with the structured description information to obtain a matching result. A first processing unit configured to, when the matching result is a successful match, input the prompt information, the standard business data matching the prompt information, and the structured description information into the AI large model to obtain a candidate result. A second processing unit configured to execute the system function matching the prompt information and combine it with the candidate result to obtain the target prediction result.

[0087] In one embodiment, the determination module 40 is further configured to perform the following steps: When the matching result is a failed match, input the prompt information and the standard business data matching the prompt information into the AI large model to obtain the target prediction result.

[0088] In one embodiment, the second processing unit is configured to perform at least one of the following steps: First, when executing the path optimization function matching the prompt information, obtain a set of predicted paths based on the improved ant colony algorithm; obtain the target prediction result based on the set of predicted paths and the candidate result. Second, when executing the inventory prediction function matching the prompt information, obtain an inventory prediction result based on the Prophet model; obtain the target prediction result based on the inventory prediction result and the candidate result. Third, when executing the transportation capacity scheduling function matching the prompt information, obtain a matching matrix between vehicles and goods based on the mixed integer programming algorithm; obtain the target prediction result based on the matching matrix and the candidate result.

[0089] In one embodiment, the apparatus further comprises: an acquisition module, configured to, in response to receiving feedback information of the target prediction result, when the evaluation score included in the feedback information is less than a qualified threshold, acquire modification information for the prompt information; the second generation module 30, configured to update the prompt information based on the modification information to obtain updated prompt information; the determination module 40, configured to obtain an updated target prediction result based on at least one of the standard service data, the structured description information, and the updated prompt information.

[0090] Each module in the above intelligent interaction device based on AI large model logistics data can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above respective modules.

[0091] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as Figure 4 shown. The computer device includes a processor, a memory, a communication interface, a display unit, and an input device connected through a method bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operation method and a computer program. The internal memory provides an environment for the operation of the operation method and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements an image processing method. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device may be a touch layer covered on the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0092] Those skilled in the art can understand that Figure 4 the structure shown in

[0093] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0094] In one embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps executed by the processor of the computer device in any of the above are implemented.

[0095] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0096] Those of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memories. Non-volatile memories can include read-only memory (ROM), magnetic tapes, floppy disks, flash memories, optical memories, high-density embedded non-volatile memories, resistive random access memories (ReRAM), magnetoresistive random access memories (MRAM), ferroelectric random access memories (FRAM), phase change memories (PCM), graphene memories, etc. Volatile memories can include random access memory (RAM) or external cache memories, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logics, data processing logics based on quantum computing, etc., without limitation.

[0097] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0098] The above-described embodiments only represent several implementation manners of the present application, and the description thereof is relatively specific and detailed. However, it should not be construed as a limitation to the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. An intelligent interaction method based on AI large model logistics data, characterized in that, The method includes: Obtain business data in the logistics system and convert the business data into standard business data; Parse and identify system functions in the logistics system to generate structured description information; In response to receiving a business query request, perform semantic recognition on the business query request to generate a prompt message; Obtain a target prediction result based on at least one of the standard business data, the structured description information, and the prompt message.

2. The method according to claim 1, wherein The converting the business data into standard business data includes: Perform semantic recognition on the business data to extract target entities; wherein, the target entities include at least one of order entities, transportation entities, inventory entities, and equipment entities; Create a data view that matches the entity type of the target entity according to the target entity; wherein, the data view includes at least one of a global inventory view, a risk warning view, and a real-time transportation capacity view; Associate each data view based on foreign key constraints to obtain the standard business data.

3. The method according to claim 1, characterized in that, The parsing and identifying system functions in the logistics system to generate structured description information includes: Parse and identify the code corresponding to the system function to obtain function description information; wherein, the system functions include at least one of a path optimization function, an inventory prediction function, and a transportation capacity scheduling function; the function description information includes a function name, input parameters, and functional logic; Perform standardization processing on the function description information to obtain the structured description information; wherein, the structured description information includes a function name, input requirements, and processing logic.

4. The method according to claim 3, characterized in that, The obtaining a target prediction result based on at least one of the standard business data, the structured description information, and the prompt message includes: Match the prompt message with the structured description information to obtain a matching result; In the case where the matching result is a successful match, input the prompt message, the standard business data that matches the prompt message, and the structured description information into an AI large model to obtain a candidate result; Execute the system function that matches the prompt message and combine it with the candidate result to obtain the target prediction result.

5. The method according to claim 4, characterized in that, The method further includes: In the case where the matching result is a failed match, input the prompt message and the standard business data that matches the prompt message into an AI large model to obtain the target prediction result.

6. The method according to claim 4, characterized in that The executing the system function that matches the prompt message and combining it with the candidate result to obtain the target prediction result includes at least one of the following: First, when executing the path optimization function that matches the prompt message, obtain a set of predicted paths based on an improved ant colony algorithm; obtain the target prediction result based on the set of predicted paths and the candidate result; Second, when executing the inventory prediction function that matches the prompt message, obtain an inventory prediction result based on the Prophet model; obtain the target prediction result based on the inventory prediction result and the candidate result; Thirdly, when executing the transportation capacity scheduling function that matches the prompt information, based on the mixed integer programming model, obtain the matching matrix between vehicles and goods; Obtain the target prediction result based on the matching matrix and the candidate result.

7. The method according to claim 1, wherein The method further includes: In response to receiving the feedback information of the target prediction result, when the evaluation score included in the feedback information is less than the qualified threshold, obtain the modification information of the prompt information; Update the prompt information based on the modification information to obtain the updated prompt information; Obtain the updated target prediction result based on at least one of the standard business data, the structured description information, and the updated prompt information.

8. An intelligent interaction device based on AI large model logistics data, characterized in that, The device includes: A conversion module, configured to obtain business data in the logistics system and convert the business data into standard business data; A first generation module, configured to parse and identify system functions in the logistics system and generate structured description information; A second generation module, configured to, in response to receiving a business query request, perform semantic recognition on the business query request and generate prompt information; A determination module, configured to obtain a target prediction result based on at least one of the standard business data, the structured description information, and the prompt information.

9. A computer device, characterized in that, It includes a processor and a memory for storing the computer program of the processor; wherein, the processor is configured to: when executing the computer program, implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps executed by the method according to any one of claims 1 to 7.

Images