Group delivery method and computing device

Abstract

The invention provides a group delivery method and computing equipment, and the method comprises the steps: constructing a label vector knowledge base which is used for carrying out the vectorization storage of the labels of commodity groups; obtaining a natural language question which is input by a user and is used for article selection; processing the natural language question to obtain a question vector; according to the problem vector, performing preliminary screening matching in the label vector knowledge base to obtain a preliminarily screened label list; generating a tag combination result from the natural language question and the tag list through a large language model; and putting the pallets according to the label combination result. According to the technical scheme of the invention, the work difficulty related to commodity labels can be reduced, the efficiency of a commodity selection process is improved, the complexity of commodity distribution is reduced, and commodity assembly and delivery can be better carried out.

Description

Technical Field

[0001] The present invention relates to the field of e-commerce, and in particular to a method for grouping and delivering goods and a computing device. Background Art

[0002] The current e-commerce platform product selection model is mainly based on tag matching, that is, a lot of product tags are established, and then users generate a selection pool by tag screening and calling the underlying tag query. This process involves a lot of manual screening of operation tags, which requires operators to have certain tag knowledge, and the overall operation efficiency is low, especially for some new users of the selection platform.

[0003] The categories and attributes of international e-commerce are very different from those of domestic e-commerce platforms. Different e-commerce platforms have different categories. There are usually tens of thousands of categories, and at least dozens of attributes under each category. It is almost impossible to complete the category attribute mapping by manpower. In addition, international e-commerce platforms also require related work such as item weight, traffic-generating words, short titles, and translated main pictures.

[0004] To this end, a technical solution is needed that can reduce the difficulty of work related to product labeling, improve the efficiency of the product selection process, reduce the complexity of distribution, and better organize and distribute goods. Summary of the invention

[0005] The present invention aims to provide a method and computing equipment for grouping and placing goods, which can reduce the difficulty of work related to product labels, improve the efficiency of the product selection process, reduce the complexity of distribution, and better carry out grouping and placing goods.

[0006] According to one aspect of the present invention, a method for grouping and delivering goods is provided, the method comprising:

[0007] Constructing a label vector knowledge base, wherein the label vector knowledge base is used to vectorize and store labels of commodity groups;

[0008] Obtain natural language questions input by users for product selection;

[0009] Processing the natural language question to obtain a question vector;

[0010] According to the question vector, a preliminary screening match is performed in the label vector knowledge base to obtain a preliminary screening label list;

[0011] The natural language question and the label list are subjected to a large language model to generate a label combination result;

[0012] The pallet is placed according to the label combination result.

[0013] According to some embodiments, a tag management system is constructed, and the tag management system is used by operators to add, delete, modify and query tag data;

[0014] The tag data falls into the tag database pool.

[0015] According to some embodiments, the natural language question is processed to obtain a question vector, including: using a word segmentation engine and a vector engine to segment and vectorize the natural language question.

[0016] According to some embodiments, the tag comprises:

[0017] One or more of product labels, business opportunity labels, sales volume labels, prohibited and restricted sales labels, price labels, service labels and merchant labels.

[0018] According to some embodiments, the natural language question and the label list are passed through a large language model to generate a label combination result, including:

[0019] The natural language question and the label list are input into a large language model through a preset prompt format to obtain the label combination result.

[0020] According to some embodiments, the step of passing the natural language question and the tag list through a large language model to generate a tag combination result further includes:

[0021] After verifying the validity of the tag combination result, the tag combination result is output in real time and fed back to the user side.

[0022] According to some embodiments, placing a pallet according to the label combination result includes:

[0023] Aggregate the products behind each label in the form of intersection and / or union to form a pallet pool;

[0024] providing a preview of said pallet pool;

[0025] The pallet pool is delivered to the customer side.

[0026] According to some embodiments, a user-side natural language group delivery page is constructed for interaction, and the user-side natural language group delivery page supports users to input natural language through the page and generate visual tags in real time.

[0027] According to another aspect of the present invention, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the method as described above is implemented.

[0028] According to another aspect of the present invention, there is provided a computing device, comprising:

[0029] Processor; and

[0030] A memory stores a computer program, and when the computer program is executed by the processor, the method described in any one of the above items is implemented.

[0031] According to an embodiment of the present invention, a method for grouping goods is provided, which aims to improve the cache hit rate and resource utilization efficiency in a multi-threaded environment. The present invention combines multiple data architectures and algorithms, and realizes natural language grouping goods delivery by building a tag management system, a tag vector knowledge base, building a product selection reasoning engineering link, building a grouping goods delivery engineering link and a user-side natural language grouping goods delivery page, thereby reducing the difficulty of work related to product labels, improving the efficiency of the product selection process, reducing the complexity of distribution, and better performing grouping goods delivery.

[0032] According to some embodiments, the method of the present invention can help classify and annotate commodities, making commodity information more refined and structured, facilitating subsequent screening and matching, and realizing refined management and precision marketing. The label vector knowledge base converts commodity features into vector form and uses machine learning or deep learning technology to calculate similarity, thereby realizing more accurate commodity recommendations.

[0033] According to some embodiments, a product selection reasoning engineering link is constructed, combining big data analysis and AI algorithms to automatically identify sales trends and customer preferences, intelligently predict demand, and improve product selection efficiency and accuracy. The product group delivery engineering link can automatically optimize product combination strategies, adjust inventory and pricing based on real-time data, and improve supply chain efficiency.

[0034] According to some embodiments, the user-side natural language group delivery page allows users to express their needs in natural language, and the system can understand and respond quickly, provide personalized recommendations, improve shopping experience and satisfaction, and enhance user experience. The automated group delivery system reduces the need for manual operation, reduces the error rate, and speeds up decision-making, improving overall operational efficiency.

[0035] According to some embodiments, by collecting and analyzing a large amount of user behavior data, the system can continuously optimize the algorithm model, provide data-based decision support for merchants, and help them better understand the market and consumers. The method of the present invention can intelligently adjust inventory levels based on historical sales data and market trend forecasts to avoid excessive inventory or out-of-stock situations, reducing costs and risks.

[0036] According to some embodiments, the ability to efficiently and accurately group and deliver goods can serve as one of the core competitiveness of an e-commerce or retail platform, attracting and retaining more customers, increasing sales, and enhancing brand competitiveness.

[0037] It is to be understood that the foregoing general description and the following detailed description are exemplary only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for describing the embodiments are briefly introduced below.

[0039] Figure 1 A flow chart of a method for group cargo delivery according to an example embodiment is shown.

[0040] Figure 2 A schematic diagram of the flow of a group cargo delivery system according to an example embodiment is shown.

[0041] Figure 3 A schematic diagram showing an example of applying a retrieval enhancement generation process to question answering according to an example embodiment.

[0042] Figure 4 A block diagram of a computing device is shown according to an exemplary embodiment. DETAILED DESCRIPTION

[0043] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that the present invention will be comprehensive and complete and fully convey the concepts of the example embodiments to those skilled in the art. The same reference numerals in the figures represent the same or similar parts, and thus their repeated description will be omitted.

[0044] In addition, the described features, structures or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present invention. However, those skilled in the art will appreciate that the technical solution of the present invention can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known methods, devices, implementations or operations are not shown or described in detail to avoid blurring various aspects of the present invention.

[0045] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0046] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0047] It should be understood that although the terms first, second, third, etc. may be used herein to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another component. Therefore, the first component discussed below can be referred to as the second component without departing from the teachings of the present inventive concept. As used herein, the term "and / or" includes any one of the associated listed items and all combinations of one or more.

[0048] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the present invention are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0049] Those skilled in the art will appreciate that the drawings are merely schematic diagrams of example embodiments, and the modules or processes in the drawings are not necessarily necessary for implementing the present invention, and therefore cannot be used to limit the protection scope of the present invention.

[0050] With the rise of the digital economy, the e-commerce industry is a part of the digital economy. It uses information technology to realize online transactions of goods and services, greatly facilitating consumers' shopping experience. E-commerce product delivery, especially the model of combining live streaming with goods, has become an important part of the e-commerce industry, involving supply chain management, marketing, data analysis and other fields.

[0051] With the popularization of the Internet and mobile Internet, the digital economy has become a new driving force for global economic growth. Consumers' increasing demand for personalization, immediacy and interactivity has prompted e-commerce companies to explore new marketing methods, such as live streaming, to attract and retain customers. Live streaming e-commerce combines entertainment, social networking and shopping functions. With the rise of live streaming e-commerce, the live streaming function of e-commerce platforms is particularly favored by merchants and consumers through the live demonstration and explanation of the anchor, which enhances the attractiveness of products and improves conversion rates.

[0052] E-commerce product delivery requires efficient supply chain support, including inventory management, logistics distribution, after-sales service, etc., to ensure that products can quickly respond to market changes and consumer demand. E-commerce product delivery involves various computer technologies, such as big data analysis. By collecting and analyzing massive amounts of user data, e-commerce companies can gain insights into consumer preferences, predict market trends, and optimize product combinations and delivery strategies; the application of artificial intelligence and machine learning (AI technology) in e-commerce product delivery includes intelligent recommendation systems, which can personalize product recommendations based on user historical behavior and preferences to improve conversion rates. In supply chain management, the Internet of Things can track the real-time status of products, ensure inventory accuracy and logistics efficiency, and blockchain technology is used to improve supply chain transparency and security, ensuring the authenticity and traceability of product sources. Cloud computing provides powerful computing power and storage resources to support e-commerce companies' data analysis, model training, and real-time operation needs. AR / VR technology augmented reality and virtual reality technology can provide an immersive shopping experience, allowing consumers to more intuitively understand product characteristics before purchasing.

[0053] The current pain points of e-commerce product group delivery include difficulty in product selection and distribution. At present, the main product selection process is based on tag matching, that is, a lot of product tags are established, and then users generate a product selection pool by tag screening and calling the underlying tag query. This process involves a lot of manual screening of operational tags, which requires operators to have certain tag knowledge, and the overall operation efficiency is low, especially for some new users of product selection platforms. The categories and attributes of international e-commerce are very different from those in China. Different e-commerce platforms have different categories. There are generally thousands of categories, and there are at least dozens of attributes under each category. It is almost impossible to complete the category attribute mapping by manpower. In addition, foreign e-commerce platforms also require piece weight, drainage words, short titles, and translated main pictures.

[0054] To this end, the present invention proposes a method for grouping and placing goods, which can reduce the difficulty of work related to product labels, improve the efficiency of the product selection process, reduce the complexity of distribution, and better carry out grouping and placing goods.

[0055] Exemplary embodiments of the present invention are described below with reference to the accompanying drawings.

[0056] Figure 1 A flow chart of a method for group cargo delivery according to an example embodiment is shown.

[0057] See also Figure 1 ,In S101, a label vector knowledge base is constructed.

[0058] According to some embodiments, a tag management system and a tag vector knowledge base are constructed. The tag management system is used by operators to add, delete, modify and query tag data. The tag data falls into the tag database pool; the tag data is used for tag selection of pallets and link generation of pallets. The tag data is unstructured data; after the unstructured tag data is vectorized, it falls into the vector library to form a tag vector knowledge base, and the tag vector knowledge base is used to vectorize and store the tags of commodity groups.

[0059] According to some embodiments, the label management system supports operators to add, delete, check and modify labels such as goods, business opportunities, sales volume, prohibited or restricted sales, prices, services, merchants, etc. through the label management system, and the labels are stored in the label database pool, which can then be used by operators to circle pallets using labels, or by system links to generate pallets using labels.

[0060] According to some embodiments, the open source HBase (column-based non-relational database management system) middleware is used as a vector library for RAG. Through technical solutions such as vector engine and word segmentation engine, unstructured label data is vectorized and put into the HBase vector library to form a vector knowledge base of labels to support the subsequent engineering chain of product selection reasoning.

[0061] HBase is a columnar, non-relational database management system that runs on Hadoop Distributed File System (HDFS). HBase provides a fault-tolerant way to store sparse data sets, which are common in many big data use cases. It is well suited for real-time data processing or random read / write access to large amounts of data. Unlike relational database systems, HBase does not support structured query languages ​​such as SQL.

[0062] Vector databases are mainly used in image retrieval, audio retrieval, text retrieval and other fields. Their main feature is that they can efficiently store and retrieve large-scale vector data. Vector databases use special data structures and algorithms to process similarity calculations and queries between vectors. By building an index structure, vector databases can quickly find the most similar vectors to meet query requirements in various application scenarios.

[0063] The vector library is used for RAG. RAG stands for Retrieval Enhanced Generation, which refers to optimizing the output of a large language model so that it can reference authoritative knowledge bases outside the training data source before generating a response. Large language models LLM are trained with massive amounts of data and use billions of parameters to generate original output for tasks such as answering questions, translating languages, and completing sentences.

[0064] The vectorization of unstructured labeled data is achieved through technologies such as vector engines and tokenization engines. A vector engine is a new data storage technology that stores and accesses data in units of vectors. Compared with traditional storage methods (such as block storage and file storage), vector storage engines have higher performance, lower latency, and more powerful data processing capabilities. The role of the tokenization engine is to break a sentence into several words.

[0065] In S103, obtain the natural language question entered by the user for product selection, process the natural language question, and obtain a question vector.

[0066] According to some embodiments, processing the natural language question to obtain a question vector is achieved by using a tokenization engine and a vector engine to tokenize and vectorize the natural language string. Tokenization refers to the process of splitting a text into words. For different languages, the tokenization methods may be different. For example, in English, words are usually naturally separated by spaces, while in Chinese, specialized algorithms are required to determine word boundaries.

[0067] After tokenization, word embedding is performed. Word embedding is the process of converting each word into a numerical vector that can capture the semantic relationships between words. Commonly used word embedding models include Word2Vec, GloVe, and BERT, etc. If a simplified word embedding model is used, a randomly generated vector can be assigned to each word (in practical applications, these vectors are usually obtained through training with a large amount of data). Chinese word vector example: "你好" -> [0.3, -0.1, 0.9,...], "今天" -> [-0.4, 0.6, 0.2,...].

[0068] In S105, based on the question vector, perform a preliminary screening and matching in the labeled vector knowledge base to obtain a preliminary screening label list.

[0069] According to some embodiments, the label list is used to determine product selection conditions, and the label list includes business opportunity labels, sales labels, prohibited and restricted sales labels, price labels, service labels, and merchant labels.

[0070] In S107, generate a label combination result from the natural language question and the label list through a large language model.

[0071] According to some embodiments, the method of the present invention includes building a product selection reasoning engineering link. Obtain the question input by the user, use the word segmentation engine and the vector engine to segment and vectorize the question string, perform a preliminary screening match in the label vector knowledge base, obtain the preliminary screened label list, and then consult the big model with the question and the label list in the standard Prompt (big model instruction) format, generate a label combination result, and verify the validity of the generated label result; output the label combination result in real time and return the value to the user side.

[0072] Prompt is a command for the big model. The content of Prompt can be questions, descriptions, keywords, context, etc., which are used to guide the model to generate text with specific topics or content. For example, "tell a joke", "use Python to write a snake game", etc. These instructions can be words, phrases, questions or complete sentences, which are used to guide the model to generate specific answers, texts, summaries or other outputs. In essence, they are instructions sent by users to the big model.

[0073] In S109, a pallet is placed according to the label combination result.

[0074] According to some embodiments, the cargo grouping and delivery pallet supports the combination of multiple labels, aggregating the goods behind each label in the form of intersection and / or union to form a pallet pool; the cargo grouping and delivery engineering link supports the preview of the pallet pool and delivers it to the user side for use.

[0075] According to some embodiments, the method of the present invention also includes constructing a user-side natural language group cargo delivery page for interaction; the user-side natural language group cargo delivery page supports users to input a natural language sentence through the page, and real-time visual labels are generated; the generated visual label results are operated, and after confirmation, a pallet can be generated, and the pallet is delivered to the scene side (page or API interface) on the user side.

[0076] Figure 2 A schematic diagram of the flow of a group cargo delivery system according to an example embodiment is shown.

[0077] See also Figure 2 , Figure 2 The working principle and process of the group product delivery method are shown. First, the user enters a question or statement through the UI front end; then, the system analyzes the historical records and product selection conditions and builds a prompt template; then, the LLM module processes the prompt and outputs the result; then, the product selection engine maps the result to the corresponding label and returns it to the user.

[0078] The question is inputted at the UI front end on the user side, and after being processed by the large language model, label combination verification and label mapping are performed, and the product selection conditions and label results are returned to the user side. Based on the product selection conditions, the product selection results are returned to the UI front end on the user side through the product selection engine.

[0079] The tag management system supports tag management, i.e., the addition, deletion, modification, query and storage of tags. The tag types include business opportunity tags, sales tags, prohibited and restricted sales tags, price / service tags and merchant tags, etc. The original tags in the tag management system are stored in the tag vector knowledge base after vector segmentation by the vector engine and word segmentation engine, and the natural language questions raised by the user are stored in the vector knowledge base after the question characters are vectorized.

[0080] Vector knowledge base is a technology based on vector storage and processing of high-dimensional data, which is used to support efficient similarity search and information retrieval. Vector knowledge base converts data into vector form in multidimensional space and uses the characteristics of these vectors for rapid retrieval and analysis, thereby achieving efficient data processing and knowledge extraction in various applications.

[0081] The core of vector knowledge base lies in its ability to efficiently store, index and search high-dimensional data points, i.e. vectors. These vectors can contain a variety of information, including numerical features, text or image embeddings, and even complex molecular structures. This ability makes vector knowledge base increasingly important in many fields.

[0082] In the vector library, the original information is first converted into vector form, a process usually called vectorization or embedding. For example, in the field of natural language processing, text data is converted into real vectors through models such as Word2Vec, GloVe or BERT; in the field of computer vision, images are extracted through CNN (convolutional neural network) and encoded into vectors. These vectors capture the key characteristics and information of the original data, so that similar items show similar characteristics in the vector space.

[0083] When a piece of information is added to the vector library, it generally goes through steps such as information preprocessing, vectorization, storing vectors and metadata, and indexing. Information preprocessing includes, for example, removing stop words, standardization, and normalization to ensure the quality of the vector representation. Vectorization converts the preprocessed information into a vector representation by using a pre-trained model. The vector library stores the generated vectors and metadata associated with the original information. The vector library will establish an efficient index structure for the stored vectors to facilitate fast similarity searches. When a query or request is received, the system converts the query into a vector and then searches the vector library for the most similar vector. Once a matching vector is found, the system uses the metadata associated with it to obtain the complete original information from an external database and then presents the original information to the user. Metadata refers to additional information associated with the stored vector, including ID, source, timestamp, and / or information link.

[0084] The present invention utilizes technologies such as LLM (large language model), RAG (retrieval enhanced generation), cache, DB (database), SSM architecture, etc. to realize a natural language group cargo delivery system based on LLM and RAG. Using this solution, the user can input a sentence, and labels can be filtered out in real time through this sentence, and pallets can be generated through the labels.

[0085] Among them, the SSM architecture is a framework system that combines Spring, SpringMVC and Mybatis. Spring is a framework that supports rapid development of Java EE applications. SpringMVC is a lightweight Web framework of request-driven type based on Java that implements the MVC design model. Mybatis is a persistence layer framework that supports customized SQL, stored procedures and advanced mapping; DB database is a persistent device, which refers to the MySQL database in the present invention; cache is a data storage method with an access speed much higher than that of a relational database; label refers to the description and classification of the pallet; and the pallet is a pool of goods.

[0086] In the specific implementation process, there is an interactive relationship between the various parts of the label management system, and the label combination verification is to ensure that the final output product selection results have high accuracy and reliability.

[0087] Figure 3 A schematic diagram showing an example of applying a retrieval enhancement generation process to question answering according to an example embodiment.

[0088] See also Figure 3 , Figure 3 An example of the application of the RAG process to question answering is given.

[0089] For smart product selection, there are usually two solutions for label recall. The first is RAG (retrieval augmented generation), which builds a vector knowledge base, first matches the most accurate labels through the knowledge base vector, and then uses the big model to make decisions and optimize. The second is SFT (supervised fine-tuning), which is to first train the big model with knowledge data, and then directly recall it through the big model.

[0090] Table 1 Characteristics and differences of the two solutions

[0091]

[0092]

[0093] In the scenario of intelligent product selection, there are often new indicators for pallets, which require high real-time performance. In addition, a set of general engineering capabilities may be required to connect to multiple knowledge bases, which requires high flexibility. The requirements for response time are not so strict. Overall, RAG's solution is more suitable for this scenario.

[0094] Retrieval-augmented generation (RAG) is an emerging solution that enhances the generative model through retrieval, that is, by integrating data from external knowledge bases, it improves the accuracy and credibility of the model in knowledge-intensive tasks, enables continuous updating of knowledge and integration of domain-specific information, and effectively combines the intrinsic knowledge of LLM with the huge dynamic resources of external data.

[0095] RAG (Retrieval-Augmented Generation) mainly includes three key processes, namely, retrieval, augmentation, and generation. Because LLM cannot learn all the details of professional knowledge in the proprietary field, it cannot give reliable and accurate answers when asking questions about professional knowledge, and may even speak nonsense. This phenomenon is called the "hallucination" of LLM. The main solution of RAG is to enhance the effect of LLM generating answers by "retrieving" the latest information from the external knowledge base as a "plug-in". Before answering questions or generating text, query external data sources and synthesize a more content-rich prompt, thereby significantly improving the accuracy and relevance of the output. At present, RAG has become one of the popular architectures in LLM systems. Due to its high practicality and low threshold, many dialogue products are built based on RAG.

[0096] Figure 4 A block diagram of a computing device is shown according to an exemplary embodiment.

[0097] like Figure 4 As shown, computing device 30 includes processor 12 and memory 14. Computing device 30 may also include bus 22, network interface 16, and I / O interface 18. Processor 12, memory 14, network interface 16, and I / O interface 18 may communicate with each other via bus 22.

[0098] The processor 12 may include one or more general-purpose CPUs (Central Processing Units, processors), microprocessors, or application-specific integrated circuits, etc., for executing relevant program instructions. According to some embodiments, the computing device 30 may also include a high-performance graphics card (GPU) 20 for accelerating the processor 12.

[0099] The memory 14 may include a machine system readable medium in the form of a volatile memory, such as a random access memory (RAM), a read-only memory (ROM) and / or a cache memory. The memory 14 is used to store one or more programs including instructions and data. The processor 12 can read the instructions stored in the memory 14 to execute the above-mentioned method according to the embodiment of the present invention.

[0100] The computing device 30 may also communicate with one or more networks via the network interface 16. The network interface 16 may be a wireless network interface.

[0101] The bus 22 may include an address bus, a data bus, a control bus, etc. The bus 22 provides a path for exchanging information between components.

[0102] It should be noted that, in the specific implementation process, the computing device 30 may also include other components necessary for normal operation. In addition, those skilled in the art may understand that the above device may only include components necessary for implementing the embodiments of this specification, and need not include all components shown in the figure.

[0103] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a microdrive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), a network storage device, a cloud storage device, or any type of medium or device suitable for storing instructions and / or data.

[0104] An embodiment of the present invention further provides a computer program product, which includes a computer program. The computer program is operable to enable a computer to execute part or all of the steps of any one of the methods described in the above method embodiments.

[0105] Those skilled in the art can clearly understand that the technical solution of the present invention can be implemented with the help of software and / or hardware. "Unit" and "module" in this specification refer to software and / or hardware that can independently complete or cooperate with other components to complete specific functions, where the hardware can be, for example, a field programmable gate array, an integrated circuit, etc.

[0106] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0107] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0108] In the several embodiments provided by the present invention, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0109] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0110] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0111] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention.

[0112] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0113] The exemplary embodiments of the present invention are specifically shown and described above. It should be understood that the present invention is not limited to the detailed structures, configurations or implementations described herein; on the contrary, the present invention is intended to cover various modifications and equivalent configurations included in the spirit and scope of the appended clauses.

Claims

1. A method for grouping and placing goods, characterized in that: include: Constructing a label vector knowledge base, wherein the label vector knowledge base is used to vectorize and store labels of commodity groups; Obtain natural language questions input by users for product selection; Processing the natural language question to obtain a question vector; According to the question vector, a preliminary screening match is performed in the label vector knowledge base to obtain a preliminary screening label list; The natural language question and the label list are subjected to a large language model to generate a label combination result; The pallet is placed according to the label combination result.

2. The method according to claim 1, characterized in that Build a tag management system, which is used by operators to add, delete, modify and query tag data; The tag data falls into the tag database pool.

3. The method according to claim 1, characterized in that The natural language question is processed to obtain a question vector, including: using a word segmentation engine and a vector engine to segment and vectorize the natural language question.

4. The method according to claim 1, characterized in that: The tags include: One or more of product labels, business opportunity labels, sales volume labels, prohibited and restricted sales labels, price labels, service labels and merchant labels.

5. The method according to claim 1, characterized in that The natural language question and the label list are passed through a large language model to generate a label combination result, including: The natural language question and the label list are input into a large language model through a preset prompt format to obtain the label combination result.

6. The method according to claim 5, characterized in that The natural language question and the label list are passed through a large language model to generate a label combination result, further comprising: After verifying the validity of the tag combination result, the tag combination result is output in real time and fed back to the user side.

7. The method according to claim 6, characterized in that The pallet is placed according to the label combination result, including: Aggregate the products behind each label in the form of intersection and / or union to form a pallet pool; providing a preview of said pallet pool; The pallet pool is delivered to the customer side.

8. The method according to claim 1, characterized in that Also includes: A user-side natural language group delivery page is constructed for interaction, wherein the user-side natural language group delivery page supports users to input natural language through the page and generates visual tags in real time.

9. A computer program product, characterized in that The method comprises a computer program, which implements the method according to any one of claims 1 to 8 when being executed by a processor.

10. A computing device, characterized in that include: processor; as well as A memory storing a computer program, wherein when the computer program is executed by the processor, the method according to any one of claims 1 to 8 is implemented.

Images