Clothing quick response supplier delivery overload handling system

Abstract

The present application relates to the technical field of clothing supply management, and specifically discloses a clothing quick response supplier delivery overload processing system, which first synchronizes the latest purchase order information from the enterprise resource planning system, then extracts the product information by scanning the QR code of the arrived goods, and calculates the actual arrival quantity based on the arrived product information, and at the same time determines whether there is overload in combination with the purchase order information. When overload is detected, the overload processing flow is automatically triggered, and natural language processing technology based on deep learning is further introduced to perform semantic parsing and interactive analysis on market demand information and arrival product information, so as to intelligently generate corresponding overload processing opinions according to market demand. In this way, it is possible to quickly respond to changes in market demand, improve the efficiency and accuracy of overload processing, thereby reducing inventory and cost problems caused by overload, and optimizing the operational efficiency of the clothing supply chain.

Description

Technical Field

[0001] The present application relates to the technical field of clothing supply management, and more specifically, to a clothing quick response supplier delivery overload processing system. Background Art

[0002] A clothing quick response supplier is a supplier in the clothing industry that can quickly adjust production and distribution strategies according to market demand to ensure that products can reach customers in the shortest possible time. However, in actual operations, due to various uncertainties that may arise in production planning, logistics and order processing, it is common for the quantity of goods delivered to exceed the maximum delivery quantity of the order.

[0003] Overloading not only increases the complexity of inventory management, but may also lead to unnecessary cost increases and waste of resources. Traditional overloading processing usually relies on manual verification and manual recording, which is not only time-consuming and labor-intensive, but also prone to errors. In addition, in traditional methods, when deciding how to deal with overloaded goods, companies usually rely on manual experience judgment or simple sales data analysis. It is difficult to accurately predict which overloaded goods can be quickly converted into sales momentum and which may be accumulated due to market saturation. The lack of in-depth analysis of market demand may not only lead to waste of resources, but also affect the company's capital flow and overall operational efficiency.

[0004] Therefore, an intelligent clothing quick response supplier delivery overload handling system is expected. Summary of the invention

[0005] In order to solve the above-mentioned technical problems, the present application is proposed. The embodiment of the present application provides a clothing quick response supplier delivery overload processing system, which first synchronizes the latest purchase order information from the enterprise resource planning system, then extracts the product information by scanning the QR code of the arrived goods, and calculates the actual arrival quantity based on the arrived product information, and at the same time determines whether there is overload in combination with the purchase order information. When overload is detected, the overload processing flow is automatically triggered, and natural language processing technology based on deep learning is further introduced to perform semantic parsing and interactive analysis on market demand information and arrival product information, so as to intelligently generate corresponding overload processing opinions according to market demand. In this way, it is possible to quickly respond to changes in market demand, improve the efficiency and accuracy of overload processing, thereby reducing inventory and cost problems caused by overload, and optimizing the operational efficiency of the clothing supply chain.

[0006] Accordingly, according to one aspect of the present application, a clothing quick response supplier delivery overload processing system is provided, which includes:

[0007] A purchase order information acquisition module is used to synchronize the latest purchase order information from the enterprise resource planning system, wherein the purchase order information includes the commodity type, specification and expected receipt quantity;

[0008] A commodity information transmission module is used to scan the QR code of the delivered commodity to obtain commodity information, and transmit the commodity information to the warehouse management system in real time to obtain a collection of commodity information;

[0009] An arrival quantity statistics module is used to calculate the actual arrival quantity based on the set of commodity information;

[0010] An overloading judgment module, used to judge whether there is an overloading situation based on the comparison between the actual arrival quantity and the expected receipt quantity to obtain a judgment result;

[0011] An overfill notification module, configured to send an overfill notification in response to the judgment result that overfilling exists, wherein the overfill notification includes the overfill quantity and current inventory information;

[0012] The overloading handling opinion generating module is used to generate overloading handling opinions based on the set of commodity information.

[0013] In the above-mentioned clothing quick response supplier delivery overload processing system, the overload processing opinion generation module includes: a market demand information semantic encoding unit, used to obtain market demand information, and semantically encode the market demand information to obtain a market demand information semantic encoding vector; a product information semantic encoding unit, used to semantically encode the set of product information to obtain a product information set semantic cascade encoding vector; a semantic interaction response analysis unit, used to perform feature interaction response analysis on the market demand information semantic encoding vector and the product information set semantic cascade encoding vector to obtain a market demand-product information alignment response encoding vector; an opinion generation unit, used to generate the overload processing opinion based on the market demand-product information alignment response encoding vector.

[0014] Compared with the prior art, the clothing quick response supplier delivery overload processing system provided by the present application first synchronizes the latest purchase order information from the enterprise resource planning system, then extracts the product information by scanning the QR code of the arrived goods, and calculates the actual arrival quantity based on the arrived product information, and at the same time determines whether there is overload in combination with the purchase order information. When overload is detected, the overload processing flow is automatically triggered, and natural language processing technology based on deep learning is further introduced to perform semantic parsing and interactive analysis on market demand information and arrival product information, so as to intelligently generate corresponding overload processing opinions based on market demand. In this way, it is possible to quickly respond to changes in market demand, improve the efficiency and accuracy of overload processing, thereby reducing inventory and cost problems caused by overload, and optimizing the operational efficiency of the clothing supply chain. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a block diagram of a clothing quick response supplier delivery overload processing system according to an embodiment of the present application.

[0017] Figure 2 It is a block diagram of an overloading handling opinion generation module in the clothing quick response supplier delivery overloading handling system according to an embodiment of the present application.

[0018] Figure 3 This is a data flow diagram of an overloading handling opinion generation module in a clothing quick response supplier delivery overloading handling system according to an embodiment of the present application.

[0019] Figure 4 It is a block diagram of a semantic interaction response analysis unit in a clothing quick response supplier delivery overload processing system according to an embodiment of the present application. DETAILED DESCRIPTION

[0020] Below, the embodiments of the present application will be described in more detail in conjunction with the accompanying drawings, and the above and other purposes, features and advantages of the present application will become more apparent. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein.

[0021] In response to the defects of traditional methods for handling overloading of clothing deliveries, this application proposes a clothing quick response supplier delivery overloading handling system, which first synchronizes the latest purchase order information from the enterprise resource planning system, then extracts the product information by scanning the QR code of the arrived goods, and calculates the actual arrival quantity based on the arrived product information, and at the same time determines whether there is overloading in combination with the purchase order information. When overloading is detected, the overloading handling process is automatically triggered, and natural language processing technology based on deep learning is further introduced to perform semantic parsing and interactive analysis of market demand information and arrival product information, so as to intelligently generate corresponding overloading handling opinions based on market demand. In this way, it is possible to quickly respond to changes in market demand, improve the efficiency and accuracy of overloading handling, thereby reducing inventory and cost issues caused by overloading, and optimizing the operational efficiency of the clothing supply chain.

[0022] Figure 1FIG. 1 is a block diagram of a clothing quick response supplier delivery overload processing system according to an embodiment of the present application. Figure 1 As shown, the clothing quick response supplier delivery overloading processing system 100 includes: a purchase order information acquisition module 110, which is used to synchronize the latest purchase order information from the enterprise resource planning system, and the purchase order information includes commodity types, specifications and expected receipt quantities; a commodity information transmission module 120, which is used to scan the QR code of the arrived commodity to obtain commodity information, and transmit the commodity information to the warehouse management system in real time to obtain a collection of commodity information; an arrival quantity statistics module 130, which is used to calculate the actual arrival quantity based on the collection of commodity information; an overloading judgment module 140, which is used to judge whether there is an overloading situation based on the comparison between the actual arrival quantity and the expected receipt quantity to obtain a judgment result; an overloading notification module 150, which is used to send an overloading notification in response to the judgment result that there is overloading, and the overloading notification includes the overloading quantity and current inventory information; an overloading processing opinion generation module 160, which is used to generate overloading processing opinions based on the collection of commodity information.

[0023] In the above-mentioned clothing quick response supplier delivery overload processing system 100, the purchase order information acquisition module 110 is used to synchronize the latest purchase order information from the enterprise resource planning system, and the purchase order information includes the type, specification and expected quantity of goods received. It should be understood that the enterprise resource planning system is a comprehensive information system used to manage resources and business processes within the enterprise, which can integrate and process information from various departments within the enterprise, including but not limited to procurement, inventory, sales, finance, etc. By synchronizing the latest purchase order information from the enterprise resource planning system, reliable data support can be provided for subsequent overload judgment and processing.

[0024] In the specific implementation, firstly, a specific data interface can be built based on modern information technologies such as Web services, RESTful API, SOAP protocol or EDI (electronic data interchange), and the clothing quick response supplier delivery overload processing system can be connected to the company's enterprise resource planning system through the data interface, thereby providing two-way communication capabilities, so that purchase order information can flow smoothly between the two systems. Specifically, the choice of which data connection technology depends on the architecture and technical specifications of the existing enterprise resource planning system, as well as the company's internal security policies and compatibility requirements. For most modern enterprises, Web service interfaces are more popular due to their flexibility, scalability and cross-platform nature.

[0025] In order to ensure that the acquired purchase order information is the latest, it is necessary to establish a reasonable data synchronization mechanism and frequency. A common approach is to adopt a timed synchronization strategy, such as automatically triggering a data synchronization operation every hour or half an hour. In this way, the overload handling system will send a data request to the enterprise resource planning system at a predetermined time point to obtain the purchase order information updated since the last synchronization. Another more real-time approach is based on an event-driven synchronization mechanism. When a key state change occurs in the purchase order in the enterprise resource planning system (such as order creation, modification, confirmation of receipt quantity change, etc.), the enterprise resource planning system will actively send a notification message to the overload handling system to trigger a data synchronization operation. This approach can maximize the real-time consistency of the data in the overload handling system and the data in the enterprise resource planning system, but the system integration and development requirements are relatively high, and the enterprise resource planning system needs to have the corresponding event publishing and subscription functions. Regardless of the synchronization mechanism adopted, strict clock synchronization is required between the two systems to ensure that the timestamp of the data is accurate and avoid data synchronization errors or omissions caused by time inconsistency.

[0026] When the overloading handling system sends a data request to the enterprise resource planning system, the enterprise resource planning system needs to query and extract the purchase order information that meets the requirements from its database according to the predetermined rules and logic. This involves complex database query operations, which are usually implemented using structured query language (SQL). First, it is necessary to determine the data source table and association relationship of the query. In the database of the enterprise resource planning system, the purchase order information may be distributed in multiple related data tables, such as the order master table, order detail table, and product information table. Through reasonable table connection operations (such as inner connection, left connection, etc.), these scattered data can be integrated together to obtain complete purchase order information. Secondly, according to the needs of the overloading handling system, specific fields are filtered out, namely, information such as product type, specification, and expected receipt quantity. This can be achieved by using the SELECT keyword in the SQL query statement to specify the list of fields to be returned. At the same time, in order to improve the efficiency of data acquisition, data can also be filtered according to some conditions, such as only obtaining purchase order information for unfinished receipt, or filtering according to specific time ranges, supplier ranges, and other conditions.

[0027] The purchase order information queried by the enterprise resource planning system needs to be transmitted to the overloading processing system through the network. During the data transmission process, in order to ensure the integrity and accuracy of the data, some data verification and validation mechanisms are usually adopted. First, before the enterprise resource planning system sends the data to the overloading processing system, it will perform a completeness check on the data to ensure that all required fields have values ​​and the data format conforms to the predetermined specifications. For example, for the commodity type field, it may be required to follow specific encoding rules; for the expected receipt quantity field, it may be required to be a positive integer, etc. If a problem is found in the data, the enterprise resource planning system will perform corresponding error processing, such as recording error logs, notifying relevant personnel to repair the data, etc., and suspend sending the data to the overloading processing system until the problem is resolved. Secondly, during the data transmission process, data encryption technology is used to protect the security of the data. For example, the SSL / TLS protocol can be used to encrypt HTTP data transmission to prevent the data from being stolen or tampered with during network transmission.

[0028] When the overload handling system receives data from the enterprise resource planning system, it will verify and check the data again. This includes checking the integrity, consistency and accuracy of the data, and comparing it with the data checksum sent by the enterprise resource planning system to ensure that the data has not been damaged or tampered with during transmission. If a problem is found in the data, the overload handling system will send a data retransmission request to the enterprise resource planning system, requesting the correct data to be resent.

[0029] The verified and validated purchase order information needs to be stored and updated in the overload handling system. The overload handling system usually uses a database management system (such as MySQL, Oracle, etc.) to store this data. In addition, before storing the data, it is necessary to design a reasonable database table structure based on the characteristics of the data and business needs. For example, create a purchase order information table containing fields such as order number, product type, specification, expected receipt quantity, synchronization timestamp, etc. Insert or update the received purchase order information according to the predetermined table structure. For new purchase order information, directly perform the insert operation; for existing purchase order information, if its key data (such as expected receipt quantity) has changed, perform an update operation to ensure that the data in the overload handling system is always up to date.

[0030] At the same time, in order to improve the efficiency of data query and processing, it is also necessary to create appropriate indexes on the database table. For example, create a primary key index on the order number field, and create a joint index on the product type and specification fields, so that data can be quickly queried and filtered based on these fields in subsequent business operations.

[0031] In addition, in order to prevent data loss or damage, data backup and recovery mechanisms are essential. Enterprises need to regularly back up the purchase order information in the overload handling system, which can be done by combining full backup and incremental backup. Full backup means that the entire database is backed up regularly (such as weekly or monthly), and all purchase order information is copied to the backup storage medium (such as tape, disk array, etc.). Incremental backup is based on full backup, and only backs up the data that has changed since the last backup every day, which can reduce the backup data volume and time cost. In the event of data loss or damage, such as database failure, hardware damage, human error, etc., the backup data can be used for recovery operations. According to the degree of data loss and backup strategy, select the appropriate backup version for recovery. The recovery process needs to be strictly operated in accordance with the recovery process of the database management system to ensure that the data can be completely and accurately restored to the specified point in time to ensure the normal operation of the overload handling system and business continuity.

[0032] In the above-mentioned clothing quick response supplier delivery overloading processing system 100, the commodity information transmission module 120 is used to scan the QR code of the arrived commodity to obtain commodity information, and transmit the commodity information to the warehouse management system in real time to obtain a collection of commodity information. It should be understood that the QR code of the arrived commodity stores the unique identifier of the commodity and other key information (such as product name, specification, batch number, production date, place of origin, etc.). After using a QR code scanning device (such as a handheld barcode scanner, a fixed barcode scanner, etc.) to read the commodity information stored in the QR code on the package of the arrived commodity, the read commodity information is sent to the warehouse management system through a wireless network or wired connection, which can realize real-time monitoring and effective allocation of goods, improve warehouse operation efficiency, and also provide accurate data support for overloading judgment.

[0033] In the specific implementation, first, high-performance QR code scanning equipment is equipped in the receiving area of ​​the warehouse to ensure that the QR code information on the received goods can be read quickly and accurately. Secondly, a stable wireless or wired network connection is established between the scanning device and the overloading processing system. When the scanning is completed, the device immediately sends the read information to the overloading processing system. After receiving the information, the overloading processing system performs preliminary data cleaning and verification, removes invalid or duplicate data, and then transmits the product information to the corresponding interface address in the warehouse management system through HTTP POST request and other methods in accordance with the interface specifications agreed with the warehouse management system (such as XML format or JSON format). After receiving the data, the warehouse management system stores it in the corresponding database table, updates the inventory-related information, and returns confirmation information of successful reception to the overloading processing system. The overloading processing system records and backs up the transmitted data for subsequent traceability and query.

[0034] In this way, the rapid collection and accurate transmission of the information of the arriving goods is realized, and the information level and work efficiency of warehouse management are improved. The warehouse management system can grasp the arrival situation in real time, facilitate the timely arrangement of the storage and shelving of goods, reduce the residence time of goods in the warehouse, and improve the inventory turnover rate. At the same time, it also provides a reliable data basis for the subsequent statistics of the arrival quantity and the judgment of overloading, reduces the error rate of overloading judgment, and improves the refinement of supply chain management.

[0035] In the above-mentioned clothing quick response supplier delivery overloading processing system 100, the arrival quantity statistics module 130 is used to calculate the actual arrival quantity based on the set of commodity information. It should be understood that clarifying the actual arrival quantity is the core step in determining whether there is overloading. Only by accurately counting the actual number of goods that have arrived can it be compared with the expected number of receipts in the purchase order to determine whether overloading has occurred and the specific number of overloading, thereby providing key data basis for subsequent processing decisions. Since the set of commodity information contains detailed information on each piece of arrived commodity, the warehouse management system counts the records of the same commodity type and specification based on the received commodity information set, and accumulates the actual arrival quantity of each commodity, so as to accurately count the actual arrival of each commodity. This statistical method based on data sets can ensure the accuracy and completeness of quantity statistics and avoid omissions and errors that may occur in manual counting.

[0036] In the specific implementation, a special quantity statistics program can be developed in the warehouse management system. First, the records of all arrived goods are queried from the database table corresponding to the collection of commodity information, and grouped according to the commodity type and specification. Then, the database aggregation function (such as the COUNT function) is used to count the number of records in each group of commodities to obtain the actual arrival quantity of each commodity. Finally, the statistical results are stored in a temporary data table for comparison with the purchase order information. At the same time, in order to ensure the accuracy and timeliness of the statistical results, each time new commodity information is transmitted to the warehouse management system, the update operation of the quantity statistics program is automatically triggered to ensure that the statistical data always reflects the latest arrival situation. In addition, the statistical process and results are logged so that they can be traced and checked when problems arise.

[0037] In this way, the actual arrival quantity can be accurately and efficiently counted, providing reliable data support for overloading judgment. Compared with the traditional manual counting method, it greatly improves the speed and accuracy of quantity counting and reduces labor and time costs. At the same time, timely and accurate arrival quantity statistics enable enterprises to quickly respond to overloading situations and take corresponding measures to avoid inventory management chaos and production delays caused by overloading problems, thereby improving the overall operational efficiency and response speed of the supply chain.

[0038] In the above-mentioned clothing quick response supplier delivery overloading processing system 100, the overloading judgment module 140 is used to judge whether there is an overloading situation based on the comparison between the actual arrival quantity and the expected receipt quantity to obtain a judgment result. Specifically, if the actual arrival quantity is greater than the expected receipt quantity, the judgment result is that there is an overloading situation; if the actual arrival quantity is less than or equal to the expected receipt quantity, the judgment result is that there is no overloading situation. It should be understood that timely discovery of overloading is an important link for enterprises to control costs, optimize inventory management and ensure the smooth operation of the supply chain. By comparing the actual arrival quantity with the expected receipt quantity, it is possible to quickly and accurately judge whether there is an overloading phenomenon, so that the enterprise can take corresponding measures to deal with it in time, and avoid inventory backlogs, capital occupation and other potential supply chain risks caused by overloading. In addition, this judgment method based on data comparison is simple and direct, and can quickly and accurately obtain the overloading judgment result, providing a clear direction for the subsequent processing flow.

[0039] In the specific implementation, an overloading judgment algorithm program can be developed in the system. The program first reads the actual arrival quantity and the expected receipt quantity data from the local database or cache to ensure the consistency and accuracy of the data. Then use programming statements (such as if-else statements) to compare and judge the quantity. When the actual arrival quantity is greater than the expected receipt quantity, the overloading quantity is calculated, and the overloading judgment result (including the overloading quantity) is stored in a specific data structure (such as a structure or object containing fields such as the judgment result identifier and the overloading quantity), and the execution of the overloading notification module is triggered at the same time; when the actual arrival quantity is less than or equal to the expected receipt quantity, the judgment result that there is no overloading is stored in the corresponding data structure, and the relevant log information is recorded for subsequent query and statistical analysis. In order to ensure the reliability of the judgment result, before comparing the data, the obtained data is checked for legitimacy, such as checking whether the data is empty, whether the data format is correct, etc., to avoid incorrect judgments caused by data anomalies.

[0040] In the above-mentioned clothing quick response supplier delivery overload processing system 100, the overload notification module 150 is used to send an overload notification in response to the judgment result that overload exists. The overload notification includes the overload quantity and current inventory information. It should be understood that once it is determined that overload exists, it is crucial to notify relevant departments and personnel in a timely manner. This can enable various departments to quickly understand inventory changes, coordinate work, and jointly formulate strategies to deal with overload, avoid work delays and decision-making errors caused by poor information flow, ensure that the company's production, sales and inventory management links can be closely connected, and minimize the adverse effects of overload on corporate operations.

[0041] In the specific implementation, firstly, the notification template and the list of notification recipients are configured in the overloading processing system. After receiving the signal of overloading sent by the overloading judgment module, the system automatically queries the current inventory information (including the inventory quantity, inventory location, inventory status, etc. of various commodities), and fills the overloading quantity and current inventory information according to the format of the notification template to generate a complete overloading notification content. Then, according to the list of notification recipients, select the appropriate notification channel to send the notification. For example, if the email notification method is selected, the system will use the internal mail server of the enterprise to send the overloading notification to the mailbox of each recipient through the SMTP protocol; if the instant messaging tool notification method is selected, the system will call the open interface of the instant messaging tool to push the notification to the corresponding contact; if the SMS notification method is selected, the system will connect with the SMS gateway and convert the notification content into SMS format and send it to the recipient's mobile phone. At the same time, the sent notification is recorded and tracked to ensure that the notification has been successfully delivered, and the viewing status and feedback information of the recipient are recorded for subsequent follow-up and statistical analysis.

[0042] In this way, the overloading information is communicated quickly and accurately, and the coordination efficiency among various departments within the enterprise is improved, so that relevant departments and personnel can timely understand the overloading situation and inventory dynamics, and make preparations in advance, such as adjusting procurement plans, arranging additional storage space, and formulating promotion strategies to digest excess inventory. In addition, through timely and effective communication and coordination, work errors and cost increases caused by information lag are reduced, the overall stability and flexibility of the enterprise supply chain are guaranteed, and the enterprise's operational management level and ability to respond to emergencies are improved.

[0043] In the above-mentioned clothing quick response supplier delivery overload processing system 100, the overload processing opinion generation module 160 is used to generate overload processing opinions based on the set of commodity information. It should be understood that the traditional overload processing method based on manual experience judgment, due to the lack of in-depth analysis of market demand, is often difficult to accurately predict which overloaded goods can be quickly converted into sales momentum and which may be accumulated due to market saturation, which may lead to waste of resources. In this regard, the present application further introduces market demand information, uses deep learning algorithms, and interactively analyzes market demand information and arrival commodity information to intelligently generate corresponding overload processing opinions.

[0044] Figure 2 It is a block diagram of an overloading handling opinion generation module in the clothing quick response supplier delivery overloading handling system according to an embodiment of the present application. Figure 3 This is a data flow diagram of the overloading handling opinion generation module in the clothing quick response supplier delivery overloading handling system according to an embodiment of the present application. Figure 2 and Figure 3As shown, the overloading handling opinion generation module 160 includes: a market demand information semantic encoding unit 161, used to obtain market demand information, and semantically encode the market demand information to obtain a market demand information semantic encoding vector; a product information semantic encoding unit 162, used to semantically encode the set of product information to obtain a product information set semantic cascade encoding vector; a semantic interaction response analysis unit 163, used to perform feature interaction response analysis on the market demand information semantic encoding vector and the product information set semantic cascade encoding vector to obtain a market demand-product information alignment response encoding vector; an opinion generation unit 164, used to generate the overloading handling opinion based on the market demand-product information alignment response encoding vector.

[0045] Specifically, the market demand information semantic encoding unit 161 is used to obtain market demand information, and semantically encode the market demand information to obtain a semantic encoding vector of the market demand information. It should be understood that by obtaining market demand information, market dynamics and consumer preferences can be understood in a timely manner, so as to more accurately evaluate the market potential of overloaded goods according to market demand, thereby making more reasonable processing decisions. Next, in order to convert the market demand information described in natural language into a data form that can be processed by a computer, the present application further uses a pre-trained language model to semantically encode the market demand information to map it to a high-dimensional semantic space and generate a semantic encoding vector of the market demand information. In a specific example of the present application, a semantic encoder based on a Transformer architecture is used to semantically encode the market demand information to obtain a semantic encoding vector of the market demand information. Those of ordinary skill in the art should know that the Transformer architecture is a model architecture widely used in the field of natural language processing, which can capture the long-distance semantic dependencies between the vocabulary units in the market demand information through a self-attention mechanism, thereby effectively extracting the deep contextual semantic features of the market demand information, and providing an accurate semantic basis for subsequent overload processing.

[0046] Specifically, the product information semantic encoding unit 162 is used to semantically encode the set of product information to obtain a semantic cascade encoding vector of the product information set. In a specific example of the present application, the product information semantic encoding unit 162 is used to: use the semantic encoder based on the Transformer architecture to semantically encode each product information in the set of product information to obtain a set of product information semantic encoding vectors; cascade the set of product information semantic encoding vectors to obtain the semantic cascade encoding vector of the product information set.

[0047] It should be understood that the product information contains a detailed description of the product, including but not limited to the product name, brand, model, color, size, etc., which is of great significance for accurately evaluating the market value and potential demand of the product. Similarly, in order to convert each product information into a computer-processable data form, the present application also uses the semantic encoder based on the Transformer architecture to semantically encode each product information in the set of product information to obtain a set of product information semantic encoding vectors. It should be understood that by using the same semantic encoding model to process market demand information and product information, errors introduced by model differences can be effectively avoided, thereby improving the accuracy of overloading handling opinions. Then, considering that the arriving goods may be diverse, in order to comprehensively consider the market potential of the entire arriving goods, the present application further cascades the set of product information semantic encoding vectors to form a comprehensive description of the arriving goods, thereby obtaining a semantic cascade encoding vector of the product information set, so as to more comprehensively reflect the market adaptability and potential value of the entire arriving goods.

[0048] Specifically, the semantic interaction response analysis unit 163 is used to perform feature interaction response analysis on the market demand information semantic coding vector and the commodity information set semantic cascade coding vector to obtain a market demand-commodity information alignment response coding vector. It should be understood that, considering that the market demand information and commodity information come from different data domains, a simple linear combination or splicing may result in information loss or inaccurate representation. Therefore, the present application proposes a feature interaction response analysis method based on semantic flow field modulation, which constructs a semantic flow field by learning the semantic alignment relationship between the market demand information semantic coding vector and the commodity information set semantic cascade coding vector, thereby finding the best matching point between market demand and commodity information in a high-dimensional space, and realizing deep interaction modeling of market demand and commodity information, ensuring that information from two different data domains can be compared and analyzed at the same semantic level to improve the accuracy of overloading processing.

[0049] Figure 4 FIG. 1 is a block diagram of a semantic interaction response analysis unit in a clothing quick response supplier delivery overload processing system according to an embodiment of the present application. Figure 4As shown, the semantic interaction response analysis unit 163 includes: a semantic flow field construction subunit 1631, which is used to construct a semantic flow field between the market demand information semantic coding vector and the commodity information set semantic cascade coding vector to obtain a market demand-commodity information semantic flow field; a feature alignment subunit 1632, which is used to perform feature alignment processing on the market demand information semantic coding vector and the commodity information set semantic cascade coding vector based on the market demand-commodity information semantic flow field to obtain an aligned market demand information semantic coding vector and an aligned commodity information set semantic cascade coding vector; a semantic interaction coding subunit 1633, which is used to perform cross-domain semantic interaction coding on the aligned market demand information semantic coding vector and the aligned commodity information set semantic cascade coding vector to obtain the market demand-commodity information alignment response coding vector.

[0050] Specifically, the semantic flow field construction subunit 1631 is used to: perform point convolution processing based on the Sigmoid function on the market demand information semantic coding vector and the commodity information set semantic cascade coding vector to obtain a dimensionally modulated market demand information semantic coding vector and a dimensionally modulated commodity information set semantic cascade coding vector, wherein the dimensionally modulated market demand information semantic coding vector and the dimensionally modulated commodity information set semantic cascade coding vector have the same characteristic dimension; perform association coding on the dimensionally modulated market demand information semantic coding vector and the dimensionally modulated commodity information set semantic cascade coding vector to obtain a market demand-commodity information association coding matrix; perform multi-scale convolution and up-sampling processing on the market demand-commodity information association coding matrix to construct the market demand-commodity information semantic flow field, which is expressed as follows:

[0051]

[0052]

[0053] in, represents the semantic encoding vector of the market demand information, represents the semantic concatenated coding vector of the product information set, represents the point convolution operation, represents the Sigmoid activation function, Represents the dimension modulated market demand information semantic encoding vector, Represents the semantic concatenated coding vector of the dimensionally modulated product information set, represents the transpose of a vector, represents the matrix multiplication operation, The characteristic scale values ​​of the dimensionally modulated market demand information semantic coding vector and the dimensionally modulated commodity information set semantic cascade coding vector, represents a 5×5 convolution operation, represents a 3×3 convolution operation, represents the upsampling operation, Represents the semantic flow field of market demand-commodity information.

[0054] That is, first, by performing dimensional modulation and multi-scale convolution operations on the semantic coding vector of the market demand information and the semantic cascade coding vector of the commodity information set, the complex semantic relationship space between the two is captured, thereby constructing a flow field that comprehensively reflects the semantic association between market demand and commodity information features, so that the subsequent feature alignment process can more accurately identify and adjust the subtle differences between features, thereby laying the foundation for deeper feature interaction.

[0055] Specifically, the feature alignment subunit 1632 is used to perform feature alignment processing on the market demand information semantic coding vector and the product information set semantic cascade coding vector based on the market demand-product information semantic flow field to obtain the aligned market demand information semantic coding vector and the aligned product information set semantic cascade coding vector, which is expressed by the formula:

[0056]

[0057] in, and They respectively represent the semantic encoding vector of the aligned market demand information and the semantic cascade encoding vector of the aligned product information set.

[0058] That is, based on the constructed semantic flow field, feature alignment processing is performed on the semantic coding vector of the market demand information and the semantic cascade coding vector of the product information set, and by dynamically adjusting the feature strength and direction of each position in the semantic coding vector of the market demand information and the semantic cascade coding vector of the product information set, the semantic features of the market demand information and the semantic features of the product information from different data sources can achieve fine-grained feature matching interaction in the same semantic space.

[0059] Specifically, the semantic interaction encoding subunit 1633 is used to: perform a linear transformation on the aligned market demand information semantic encoding vector to obtain a query vector and a value vector; perform a linear transformation on the aligned product information set semantic cascade encoding vector to obtain a key vector; input the query vector, the value vector and the key vector into a fine-grained response encoding module based on a converter structure to obtain the market demand-product information alignment response encoding vector, which is expressed as:

[0060]

[0061] in, , and denote the query embedding matrix, value embedding matrix and key embedding matrix respectively, , and Represent different bias terms, , and denote the query vector, value vector, and key vector respectively, is the characteristic scale value of the key vector, represents the normalized exponential function, Represents the market demand-product information alignment response encoding vector.

[0062] Here, after the feature alignment is completed, the converter structure is further used to perform cross-domain semantic interaction encoding between the market demand information semantic encoding vector and the commodity information set semantic cascade encoding vector. Specifically, firstly, through linear transformation operations, query vectors and value vectors are constructed based on the market demand information semantic encoding vector, and key vectors are constructed based on the commodity information set semantic cascade encoding vector, so that it can better serve the query interaction task of cross-domain features. Then, the query vector, value vector and key vector are input into the fine-grained response encoding module based on the converter structure to perform deep response interaction between features. Accordingly, the converter structure is based on a multi-head self-attention mechanism and a feedforward neural network, and guides the allocation of attention weights through the similarity calculation between the query vector and the key vector, dynamically retrieves information related to market demand features from the commodity information set, and finally generates a market demand-commodity information alignment response encoding vector. In this way, not only the original market demand information semantic features and commodity information semantic features are retained, but also the understanding and modeling capabilities of the complex relationship between market demand and commodity information are enhanced through fine-grained interactive encoding, providing more accurate decision support for the handling of overloaded commodities.

[0063] Specifically, the opinion generation unit 164 is used to generate the overloading handling opinion based on the market demand-commodity information alignment response coding vector. In a specific example of the present application, the opinion generation unit 164 is used to: input the market demand-commodity information alignment response coding vector into the overloading suggestion generation module based on the classifier to obtain the overloading handling opinion, and the overloading handling opinion is used to indicate total rejection, partial rejection, partial acceptance, or total acceptance. Specifically, the classifier comprehensively considers the market demand, commodity characteristics, market trends, and commodity scale information contained in the market demand-commodity information alignment response coding vector by performing feature learning on the market demand-commodity information alignment response coding vector, identifies the degree of match between market demand and commodity information, and makes corresponding commodity acceptance decisions accordingly, providing clear suggestions for decision makers. For example, if the market demand is strong and the commodity information is highly matched with the market demand, the classifier may recommend accepting all of the goods. On the contrary, if the market demand is weak or the commodity information does not match the market demand, the classifier may recommend rejecting all or partially rejecting. In this way, receiving strategies can be flexibly adjusted to changing market conditions, allowing for efficient management of overloads.

[0064] In a preferred example of the present application, inputting the market demand-product information alignment response encoding vector into a classifier-based overloading suggestion generation module to obtain an overloading handling suggestion includes:

[0065] First, the distance between each pair of eigenvalues ​​of the market demand-product information alignment response encoding vector is calculated, such as the L2 distance, and the square root of the distance is taken to obtain the market demand-product information alignment response encoding distance representation matrix, that is,

[0066]

[0067] in, represents the market demand-product information alignment response encoding vector, , They represent the first Position and The eigenvalues ​​at the positions, represents the distance metric function, Indicates the alignment of market demand and product information in response to the encoding distance representation matrix The element value at the position;

[0068] Next, the market demand-product information alignment response encoding autocorrelation matrix of the market demand-product information alignment response encoding vector as a row vector is obtained, that is, ,in, represents the market demand-product information alignment response encoding autocorrelation matrix;

[0069] At the same time, the market demand-product information alignment response coding vector is matrix-multiplied with the market demand-product information alignment response coding distance representation matrix to obtain the market demand-product information alignment response coding primary mapping vector, that is, ,in, represents the market demand-product information alignment response encoding distance representation matrix, Represents the first-level mapping vector of market demand-product information alignment response encoding;

[0070] Secondly, the market demand-product information alignment response coding primary mapping vector is matrix-multiplied with the matrix product of the market demand-product information alignment response coding distance representation matrix and the market demand-product information alignment response coding self-correlation matrix to obtain the market demand-product information alignment response coding multi-level mapping vector

[0071] ,in, Represents a multi-level mapping vector of market demand-product information alignment response encoding;

[0072] Then, the market demand-product information alignment response coding multi-level mapping vector is interpolated with the market demand-product information alignment response coding associated eigenvector composed of the eigenvalues ​​of the market demand-product information alignment response coding auto-association matrix to obtain an optimized market demand-product information alignment response coding vector, wherein interpolation or zero padding is performed when the eigenvalue is insufficient; and the optimized market demand-product information alignment response coding vector is input into an overloading suggestion generation module based on a classifier to obtain overloading handling suggestions.

[0073] Here, considering that the market demand information semantic coding vector and the commodity information set semantic cascade coding vector represent the semantic coding features of the market demand information and the semantic cascade features of the commodity signal respectively, when performing feature interaction response based on semantic flow field modulation, the source domain content difference between the market demand information and the commodity information set will cause the fine-grained distortion of semantic fluency, resulting in the lack of instance determination of fine-grained semantic interaction features of the market demand-commodity information alignment response coding vector, thereby affecting the accuracy of the overloading handling opinions obtained through the classifier-based overloading suggestion generation module.

[0074] Therefore, through the linear target mapping representation based on the similarity distance representation matrix of the market demand-product information alignment response coding vector, the self-correlation complete similarity instantiation of the market demand-product information alignment response coding vector is instantiated with a secondary target mapping representation based on a multi-level distribution hierarchy, and the association mismatch negative influence factor is compensated by the association fusion kernel bias to improve the degree of instance determination of the eigenvalue of the market demand-product information alignment response coding vector under similarity constraints, that is, the significance of the eigenvalue as an instance for classification regression judgment, so as to improve the accuracy of the overloading handling opinions obtained by the classifier-based overloading suggestion generation module of the market demand-product information alignment response coding vector.

[0075] In summary, according to the embodiment of the present application, the clothing quick response supplier delivery overloading processing system is explained, which first synchronizes the latest purchase order information from the enterprise resource planning system, then extracts the product information by scanning the QR code of the arrived goods, and calculates the actual arrival quantity based on the arrived product information, and at the same time determines whether there is overloading in combination with the purchase order information. When overloading is detected, the overloading processing flow is automatically triggered, and natural language processing technology based on deep learning is further introduced to perform semantic parsing and interactive analysis on market demand information and arrival product information, so as to intelligently generate corresponding overloading processing opinions according to market demand. In this way, it is possible to quickly respond to changes in market demand, improve the efficiency and accuracy of overloading processing, thereby reducing inventory and cost problems caused by overloading, and optimizing the operational efficiency of the clothing supply chain.

[0076] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to be implemented by adopting the above specific details.

Claims

1. A clothing quick response supplier delivery overload processing system, characterized in that: include: A purchase order information acquisition module is used to synchronize the latest purchase order information from the enterprise resource planning system, wherein the purchase order information includes the commodity type, specification and expected receipt quantity; A commodity information transmission module is used to scan the QR code of the delivered commodity to obtain commodity information, and transmit the commodity information to the warehouse management system in real time to obtain a collection of commodity information; An arrival quantity statistics module is used to calculate the actual arrival quantity based on the set of commodity information; An overloading judgment module, used to judge whether there is an overloading situation based on the comparison between the actual arrival quantity and the expected receipt quantity to obtain a judgment result; An overfill notification module, configured to send an overfill notification in response to the judgment result that overfilling exists, wherein the overfill notification includes the overfill quantity and current inventory information; An overloading handling opinion generating module, used for generating an overloading handling opinion based on the set of commodity information; The overloading handling opinion generating module includes: a market demand information semantic encoding unit, which is used to obtain market demand information and semantically encode the market demand information to obtain a market demand information semantic encoding vector; a commodity information semantic encoding unit, which is used to semantically encode the set of commodity information to obtain a commodity information set semantic cascade encoding vector; a semantic interaction response analysis unit; and an opinion generating unit, which is used to generate the overloading handling opinion based on the market demand-commodity information alignment response encoding vector. Wherein, the semantic interaction response analysis unit includes: The semantic flow field construction subunit is used to: perform point convolution processing based on the Sigmoid function on the market demand information semantic coding vector and the commodity information set semantic cascade coding vector respectively to obtain the dimensionally modulated market demand information semantic coding vector and the dimensionally modulated commodity information set semantic cascade coding vector, wherein the dimensionally modulated market demand information semantic coding vector and the dimensionally modulated commodity information set semantic cascade coding vector have the same characteristic dimension; perform association coding on the dimensionally modulated market demand information semantic coding vector and the dimensionally modulated commodity information set semantic cascade coding vector to obtain a market demand-commodity information association coding matrix; perform multi-scale convolution and upsampling processing on the market demand-commodity information association coding matrix to construct a market demand-commodity information semantic flow field; A feature alignment subunit, used for performing feature alignment processing on the market demand information semantic coding vector and the product information set semantic cascade coding vector based on the market demand-product information semantic flow field to obtain an aligned market demand information semantic coding vector and an aligned product information set semantic cascade coding vector; The semantic interaction encoding subunit is used to perform cross-domain semantic interaction encoding on the aligned market demand information semantic encoding vector and the aligned product information set semantic cascade encoding vector to obtain a market demand-product information alignment response encoding vector.

2. The clothing quick response supplier delivery overload processing system according to claim 1 is characterized in that: The overfilling judgment module is used to: In response to the actual arrival quantity being greater than the expected receipt quantity, the judgment result is that there is an overloading situation; In response to the actual arrival quantity being less than or equal to the expected receipt quantity, the judgment result is that there is no overloading.

3. The clothing quick response supplier delivery overload processing system according to claim 2 is characterized in that: The market demand information semantic encoding unit is used to: The market demand information is semantically encoded using a semantic encoder based on a Transformer architecture to obtain a semantic encoding vector of the market demand information.

4. The clothing quick response supplier delivery overload processing system according to claim 3 is characterized in that: The commodity information semantic encoding unit is used to: Using the semantic encoder based on the Transformer architecture to semantically encode each piece of product information in the set of product information to obtain a set of product information semantic encoding vectors; The set of the commodity information semantic coding vectors is cascaded to obtain the commodity information set semantic cascade coding vector.

5. The clothing quick response supplier delivery overload processing system according to claim 4 is characterized in that: The semantic interaction encoding subunit is used to: Performing a linear transformation on the semantic encoding vector of the aligned market demand information to obtain a query vector and a value vector; Performing a linear transformation on the semantic concatenated encoding vector of the aligned product information set to obtain a key vector; The query vector, the value vector and the key vector are input into a fine-grained response encoding module based on a converter structure to obtain the market demand-product information aligned response encoding vector.

6. The clothing quick response supplier delivery overload processing system according to claim 5 is characterized in that: The opinion generating unit is used for: The market demand-product information alignment response encoding vector is input into the classifier-based overloading suggestion generation module to obtain the overloading handling opinion, and the overloading handling opinion is used to indicate total rejection, partial rejection and partial acceptance, or full acceptance.

Images