A method of creating a product label and a system for creating the same

Abstract

A method and system for establishing product labels, relating to the field of information processing technology, includes the following steps: S1, establishing a correspondence between the standard name of a product and its attributes; S2, collecting product data and extracting textual features from the data to retain characteristic keywords, and matching these characteristic keywords with the standard name to achieve preliminary classification of the product data; S3, classifying characteristic keywords within the same category to obtain different characteristic features, and adding characteristic features containing more than a set threshold to the product attributes, thereby establishing new product attributes or improving existing ones, thus forming product labels. In this way, product attributes are updated based on product data, making the label formation up-to-date. This facilitates consumer selection and reference when addressing specific consumer needs, and is also convenient for use as a recommendation algorithm.

Description

Technical Field

[0001] This invention relates to the field of information processing technology, and in particular to a method and system for creating labels for various products. Background Technology

[0002] With the rapid development of the internet, various industries have chosen to conduct e-commerce transactions online and rely on the internet to establish exchanges of product information. This exchange includes post-sale reviews, as well as discussions and evaluations of linked and unlinked products. These reviews, discussions, and evaluations can all serve as product data. By processing and integrating this data, we can obtain tag information such as product category or performance characteristics. This tag information is of great significance for consumers' purchasing decisions and for merchants' product updates and improvements. Furthermore, this tag information can be used to implement recommendation algorithms or to create user profiles of users of such products.

[0003] However, the processing and integration of product data currently relies primarily on manual methods, mainly through after-sales customer service reviewing customer service feedback to obtain category or performance tags corresponding to the products sold. This manual method has two drawbacks: firstly, the scope of information obtained is too narrow, limited to after-sales feedback, resulting in a small amount of product data and poor accuracy and objectivity of the obtained tags; secondly, manual processing is inefficient and requires high labor costs.

[0004] Therefore, it is necessary to develop a method that facilitates the processing of massive amounts of product data to establish accurate product labels. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for establishing product labels. This method extracts characteristic keywords from product data and processes these keywords to obtain characteristic features. Characteristic features containing more than a set threshold of characteristic keywords are added to product attributes, thereby establishing new product attributes or improving existing product attributes. This ensures the consistency between product attributes and product data updates, and guarantees the relevance and accuracy of product standard names and product attributes, thus establishing up-to-date product labels.

[0006] To achieve the above objectives, the present invention adopts the following technical solution.

[0007] A method for creating a product label includes the following steps:

[0008] S1. Establish the correspondence between the standard name of the product and the product attributes;

[0009] S2. Collect data about the product and extract text features from the data to retain characteristic keywords. Use the characteristic keywords to match the standard names to achieve preliminary classification of the product data.

[0010] S3. Classify the characteristic keywords in the same category to obtain different characteristic features. Add the characteristic features containing more than a set threshold to the product attributes to create new product attributes or improve existing product attributes, thus forming product tags.

[0011] In this way, product attributes are updated based on product data, making the formation of tags up-to-date. When targeting specific consumer filtering needs, it can facilitate consumer selection and reference, and it can also be used as a recommendation algorithm.

[0012] Furthermore, step S1 also includes establishing a preliminary name correspondence between standard names and aliases through integration, and obtaining name characteristics by performing dimensionality reduction processing on information that can be obtained about product names. The name characteristics are then matched with standard names to improve the name correspondence. If no matching standard name exists, a new standard name is established. In this way, either the name characteristics are incorporated into existing standard names through matching, or a new standard name is established if no matching standard name exists, ensuring that different name characteristics have corresponding standard names.

[0013] Furthermore, similarity algorithms, machine learning models, or manual review methods are used to match name characteristics with standard names;

[0014] Furthermore, in step S1, the product attributes include category attributes, parameter attributes, and performance attributes. The category attribute refers to the category or type to which the product belongs, the parameter attribute refers to the specific parameters of the product, and the performance attribute refers to the performance of the product.

[0015] Furthermore, in step S2, the characteristic keywords include name-type characteristic keywords that can reveal the product name and attribute-type characteristic keywords that can reveal the product attributes. By matching name-type characteristic keywords with standard names, product data is classified into the standard names that match them.

[0016] Furthermore, in step S3, the characteristic keywords are processed using feature embedding methods or text feature processing methods, and then similarity matching, classification or clustering is performed to achieve characteristic classification and obtain different characteristic features.

[0017] Furthermore, it also includes the following steps:

[0018] Step S4: Select the top-ranked feature features in terms of the number of featured keywords as core features;

[0019] Step S5: Through nearest neighbor search, classify the product data into the standard name that is most similar to the core feature and its characteristic keywords to achieve product data reclassification;

[0020] Step S6: Repeat steps S4 and S5 until the classification no longer changes;

[0021] In this way, not only can product tags be generated, but corresponding product data can also be retrieved to further enrich the product tag content;

[0022] Furthermore, in step S4, the core features are converted into word vectors before nearest neighbor search is performed.

[0023] A product label creation system includes a correspondence module, a processing module, and a first matching module. The correspondence module is used to establish a correspondence between the standard name of a product and product attributes. The processing module is used to extract characteristic keywords from data about the product, match the characteristic keywords with the standard name to achieve preliminary classification of the product data, classify the characteristic keywords in the same category to obtain different characteristic features, and the first matching module adds characteristic features containing more than a set threshold to the product attributes, thereby creating new product attributes or improving existing product attributes.

[0024] Furthermore, it also includes a standard module, which is used to establish the name correspondence between standard names and aliases, and to obtain name characteristics by performing dimensionality reduction processing on information that can be obtained about product names, and to match the name characteristics with standard names to improve the name correspondence. If there is no standard name that can be matched, a new standard name is established accordingly.

[0025] Furthermore, it also includes a second matching module, which selects several feature features with the highest number of feature keywords as core features, and uses nearest neighbor search to classify the product data into standard names that are most similar to the core features and feature keywords.

[0026] A computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement a method for classifying product data.

[0027] A terminal device includes a memory and a processor, the processor being configured to implement a product data classification method upon execution based on instructions stored in the memory.

[0028] The beneficial effects of this invention are as follows:

[0029] On the one hand, by extracting characteristic keywords from product data and processing them to obtain characteristic features, characteristic features containing more than a set threshold are added to product attributes, thereby establishing new product attributes or improving existing product attributes. This ensures the consistency of product attributes and product data updates, and guarantees the relevance and accuracy of product standard names and product attributes. It also establishes up-to-date product tags, which is beneficial to recommendation algorithms.

[0030] On the other hand, preliminary classification is achieved by matching characteristic keywords in product data with standard names. Then, by leveraging the correspondence between standard product names and product attributes, the classified product data is brought into the category. Core features are obtained by extracting and processing characteristic keywords from product data. The classification of product data is adjusted by performing nearest neighbor search on characteristic keywords and core features in product data to ensure the accuracy of product data classification. Product data is directly associated with product tags, which can retrieve corresponding product data and enrich the presentation of product tags. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the process for establishing a product label according to an embodiment of the present invention. Detailed Implementation

[0032] In the description of this invention, it should be understood that the terms or positional relationships indicating orientation are based on the orientation or positional relationships shown in the accompanying drawings, and are only for the convenience of describing this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention.

[0033] The principles and features of the present invention are described below with reference to a list. The examples given are for illustrative purposes only and are not intended to limit the scope of the invention. It should be noted that the lists are all in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the present invention.

[0034] In one embodiment of the present invention, such as Figure 1 As shown, the method for creating product labels includes the following steps:

[0035] 1) By integrating products, establish a name correspondence between standard names and aliases;

[0036] This integration includes both initial manual integration and subsequent automatic integration. The automatic integration refers to performing dimensionality reduction processing on information that can be obtained about product names to obtain name characteristics, matching the name characteristics with standard names to improve the name correspondence, and establishing a new standard name if no matching standard name exists.

[0037] The information that can reveal the name of the engine oil includes name information, model information, etc. The name information includes the full name and abbreviation, as well as the standard name and alias.

[0038] The dimensionality reduction process mentioned refers to the dimensionality reduction techniques used, including dimensionality reduction techniques based on the PCA (Principal Component Analysis) concept, such as Kernel PCA, Incremental PCA, Sparse PCA, Robust PCA, NMF (Non-negative Matrix Factorization), and ICA (Independent Component Analysis).

[0039] The matching process includes using similarity algorithms, machine learning models, or manual review to match name characteristics with standard names. Similarity algorithms, such as cosine similarity and Jaccard similarity, calculate the similarity between the dimensionality-reduced name characteristics and the standard names to determine their degree of matching. Machine learning models, such as classification models and clustering models, classify or cluster the dimensionality-reduced name characteristics and standard names to determine their matching relationships. Manual review involves manually verifying and adjusting the dimensionality-reduced name characteristics and standard names to ensure the accuracy and reliability of the matching results.

[0040] Since the name correspondence can be continuously improved by collecting and processing product data, it can be well matched with the product's technological updates and obtain the latest product aliases and standard names in a timely manner.

[0041] 2) Establish the correspondence between standard names and product attributes;

[0042] The product attributes include category attributes, parameter attributes, and performance attributes. Category attributes refer to the category or type to which the product belongs. For example, engine oil can be divided into synthetic engine oil, semi-synthetic engine oil, and mineral oil. Parameter attributes refer to the specific parameters of the product. Again, using engine oil as an example, these include compatible car models, viscosity grade, engine oil specifications, and engine oil capacity. Performance attributes refer to the product's performance characteristics. Again, using engine oil as an example, these include high viscosity, good anti-wear properties, low-temperature resistance, and low price.

[0043] Similarly, product attributes include both manual and automatic integration. For automatic integration, the methods are explained below.

[0044] 3) Collect data about the product and extract text features from the data to retain characteristic keywords. Use the characteristic keywords to match the standard names to achieve preliminary classification of the product data.

[0045] Product data includes, but is not limited to, product information (such as brand, model, specifications, capacity, price, etc.), product parameters (such as viscosity, viscosity index, flash point, pour point, interpolation point, etc.), and user reviews (such as user comments, messages, articles, forum posts, etc.).

[0046] The text feature extraction mentioned above is a technique used to remove impurities and extract key information, including keyword filtering (such as TF-IDF, TextRank, LDA, etc.), rule-based matching (such as regular expressions), word frequency-based filtering (such as statistical word frequency), or machine learning-based classification, etc.

[0047] Feature keywords include name-based feature keywords that reveal the product name and attribute-based feature keywords that reveal the product attributes. Product data is categorized into matching standard names by matching name-based feature keywords with standard names. The matching method is the same as in step 1).

[0048] 4) After processing the characteristic keywords that belong to the same category using feature embedding methods or text feature processing methods, similarity matching, classification or clustering are performed to achieve characteristic classification and obtain different characteristic features;

[0049] The feature embedding methods (such as word2vec, GloVe, FastText, Doc2Vec, etc.) are used to convert features into vectors, while the text feature processing methods (such as TF-IDF, BM25, LDA bag-of-words model, n-gram, etc.) are used to convert text into vectors, and then similarity matching, classification, or clustering are performed to achieve classification. Common applications of the aforementioned methods include similarity matching, machine learning classification, deep learning classification, clustering classification, Naive Bayes classifiers, support vector machine classifiers, convolutional neural networks, and recurrent neural networks.

[0050] 5) Add features containing more than a set threshold of characteristic keywords to product attributes, thereby creating new product attributes or improving existing ones. Furthermore, select the top-ranking features by the number of characteristic keywords as core features;

[0051] By selecting a limited number of features, errors can be avoided, as features with few keywords are more likely to contain typos, inaccuracies, or subjective descriptions. This ensures the accuracy of core features and facilitates subsequent product data classification. Setting thresholds allows for the selection of product attributes that can be associated with standard names, further refining product attributes and leading to accurate and up-to-date product tags.

[0052] 6) Convert the core features into word vectors;

[0053] 7) By using nearest neighbor search, the product data is classified into standard names that are most similar to its core features and characteristic keywords to achieve product data reclassification;

[0054] The nearest neighbor search includes exact nearest neighbor and approximate nearest neighbor. The exact nearest neighbor refers to finding the data point closest to the query point in a given dataset, such as linear scan, KD tree, or Ball tree. The approximate nearest neighbor refers to finding an approximate value of the data point closest to the query point in a given dataset. The process uses some approximation algorithms to accelerate the calculation and reduce the computation time and space complexity, such as Locality Sensitive Hash (LSH), randomized KD tree, Approximate Nearest Neighbor Tree (ANN tree), Graph-based Nearest Neighbor Search, etc.

[0055] 8) Repeat steps 4)-7) until the classification no longer changes.

[0056] Based on the description of the foregoing embodiments, specific examples are provided below, such as engine oil, which is divided into three categories: ordinary mineral oil, semi-synthetic engine oil, and fully synthetic engine oil. Of course, it should be noted that this classification is not limited to engine oil, nor is it limited to the aforementioned classification methods. It can be adjusted as needed, for example: 1. Food, such as beverages, candy, snacks, convenience foods, and condiments. Beverages can be further divided into different brands and types, etc. The classification below follows the same principle. For simplicity, only the simplest classifications are listed below; 2. Cosmetics, such as skincare products, makeup, perfumes, shampoos, and shower gels; 3. Clothing, such as men's wear, women's wear, children's wear, sportswear, and underwear; 4. Home furnishings, such as furniture, home textiles, kitchenware, and bathroom supplies; 5. Electronic products, such as mobile phones, computers, tablets, cameras, and audio equipment; 6. Automotive parts, such as tires, brake pads, air filters, and spark plugs; 7. Medical devices, such as blood pressure monitors, blood glucose meters, thermometers, and masks; 8. Sports equipment, such as treadmills, fitness equipment, bicycles, and swimming equipment. In summary, this application is not limited to a single classification, as long as the category divisions are clearly distinguishable for classification purposes.

[0057] The product label creation method according to this embodiment, taking the creation of an engine oil label as an example, is explained in the form of a list.

[0058] ① Establish a mapping table of aliases, standard names, and engine oil properties;

[0059] In this embodiment, the engine oil attributes include brand, category, and performance. The table is only provided as an example for illustrative purposes; please refer to Table 1 for details.

[0060]

[0061] Table 1

[0062] ② Collect reviews about engine oil and retain the characteristic keywords of each review through word frequency analysis; as shown in Table 2:

[0063]

[0064] Table 2

[0065] ③ Using the name-type characteristic keywords obtained, they are initially classified and assigned to matching standard names; the name-type characteristic keywords include, in this embodiment, Mobil, No. 1, Kunlun, Tianrun, KR9, etc., and are matched to their corresponding standard names, as shown in Table 3:

[0066]

[0067] Table 3

[0068] ④ Classify the characteristic keywords to obtain different characteristic features;

[0069] Taking Mobil 1 in Table 3 as an example, after matching the standard name, temporary matching of oil reviews was also achieved, allowing oil reviews to be applied to the corresponding standard name. As shown in Table 3, the characteristic keywords applicable to Mobil 1 oil reviews include durability, affordability, anti-wear, cleaning, and engine protection. Since Table 3 is only a partial illustration, assuming that similar evaluations of Mobil 1 exist in other oil reviews, such as "Mobil 1 is economical," "Mobil 1 is a great deal," "Mobil 1 has excellent anti-wear properties," "Mobil 1 is very durable," etc., then "Mobil 1 is economical," "Mobil 1 is a great deal," and "affordable" can be categorized under the characteristic of affordability, while "anti-wear," "Mobil 1 has excellent anti-wear properties," and "Mobil 1 is very durable" can be categorized under the characteristic of anti-wear properties. This completes the feature classification of characteristic keywords, thereby obtaining different characteristic features under the standard name Mobil 1, as shown in Table 4.

[0070]

[0071] Table 4

[0072] ⑤ Add features containing more than 30 characteristic keywords to the product attributes to create new product attributes or improve existing ones, thus forming the associated tags for the engine oil. For example, if the number of characteristic keywords for Mobil 1 that indicate it is cheap exceeds 30, then this perception is considered to be widespread. If the original product attributes corresponding to the standard name Mobil 1 do not include the attribute of "cheap," then add the characteristic of "cheap" to the product attributes corresponding to the standard name Mobil 1. In this way, the product attributes can be continuously enriched based on actual consumer experience, making the product attributes more closely aligned with actual usage experience, and further improving the correspondence between the standard name and product attributes, as shown in Table 5.

[0073]

[0074] Table 5

[0075] To further enrich the engine oil labels, engine oil reviews are linked to the previously established engine oil labels. Since step ③ initially categorized the oils using name-based characteristic keywords, when an engine oil review mentions two different product categories, it will be categorized under two different standard names, corresponding to two different engine oil attributes, leading to incorrect label association. For example, as shown in Table 6, the engine oil review "I've used both Great Wall Jinjixing J500 and Mobil 1, but I personally feel Mobil 1 is better; its anti-wear and cleaning abilities are very strong" is incorrectly associated with Great Wall Jinjixing J500. When retrieving related labels for Great Wall Jinjixing J500, this engine oil review will also be retrieved, resulting in label errors.

[0076]

[0077] Table 6

[0078] Therefore, the following steps are continued to ensure the accurate classification of product data, that is, to ensure the accurate association between product data and labels:

[0079] ⑥ Select the top 50 features in terms of the number of included characteristic keywords as core features; taking Mobil 1 as an example, assuming there are 8,000 characteristic keywords for the feature of engine protection, 7,982 characteristic keywords for the feature of cleanliness, 7,820 characteristic keywords for the feature of wear resistance, etc., select the top 50 characteristic features in terms of the number of included characteristic keywords as core features.

[0080] ⑦ Convert its core features into word vectors. Using approximate nearest neighbor search, assign each oil review to the standard name most similar to its feature keywords, thus reclassifying the oil reviews. Repeat steps ④, ⑤, and ⑥ until the classification no longer changes, completing the classification of the oil reviews and enriching the oil label content. Continuing with the case in Table 6, using approximate nearest neighbor search, the feature vectors of anti-wear and cleaning features are closer to the core feature vector of Mobil 1. Therefore, the oil review is associated with the standard name Mobil 1, ensuring the accuracy of the product data classification for oil reviews. The classification results are shown in Table 7, thus correcting the erroneous associations shown in Table 6.

[0081]

[0082] Table 7

[0083] This embodiment also provides a product label creation system, including a standard module, a corresponding module, a processing module, a first matching module, and a second matching module;

[0084] The standard module is used to establish the name correspondence between standard names and aliases. It also performs dimensionality reduction processing on information that can be used to obtain name characteristics, matches the name characteristics with standard names to improve the name correspondence, and establishes a new standard name if no matching standard name exists.

[0085] The corresponding module is used to establish the correspondence between the standard name of a product and its attributes;

[0086] The processing module is used to extract characteristic keywords from product data, match characteristic keywords with standard names to achieve preliminary classification of product data, and classify characteristic keywords in the same category to obtain different characteristic features.

[0087] The first matching module adds features containing more than a set threshold of characteristic keywords into the product attributes, thereby creating new product attributes or improving existing product attributes.

[0088] The second matching module selects several features with the highest number of included feature keywords as core features, and uses nearest neighbor search to classify the product data into standard names that are most similar to the core features and their feature keywords.

[0089] This embodiment also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement a product labeling and classification method.

[0090] This embodiment also provides a terminal device, including a memory and a processor, wherein the processor is configured to implement a product label classification method when executing instructions stored in the memory.

[0091] The readable storage medium of this embodiment can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be—but is not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0092] The program code contained on the computer-readable medium can be transmitted using any suitable medium, including—but not limited to—wireless, wired, optical fiber, RF, etc., or any suitable combination thereof. The computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or terminal. In cases involving a remote computer, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0093] A processor may include one or more processing cores. It connects to various parts of the terminal device via various interfaces and lines, executing instructions, programs, code sets, or instruction sets stored in memory, and accessing data stored in memory to perform various functions and process data. Optionally, the processor may be implemented using at least one of the following hardware forms: Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA). The processor may integrate one or more of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the displayed content; and the modem handles wireless communication. It is understood that the modem may also be implemented separately as a communication chip, without being integrated into the processor.

[0094] The memory may include random access memory (RAM) or read-only memory (ROM). The memory can be used to store instructions, programs, code, code sets, or instruction sets. The memory may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for implementing at least one function (such as touch functionality, sound playback functionality, image playback functionality, etc.), and instructions for implementing the aforementioned method embodiments. The data storage area may also store data created during terminal use (such as phonebook data, audio / video data, chat log data, etc.).

[0095] Furthermore, it should be understood that although this specification describes embodiments, the described embodiments are merely some, not all, of the embodiments described in this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort should fall within the scope of protection of this application.

Claims

1. A method for establishing a product label, characterized in that, Includes the following steps: S1. Establish the correspondence between the standard name of the product and the product attributes; The product attributes include category attributes, parameter attributes, and performance attributes. The category attribute refers to the category or type to which the product belongs, the parameter attribute refers to the specific parameters of the product, and the performance attribute refers to the performance of the product. S2. Collect data about the product and extract text features from the data to retain characteristic keywords. Use the characteristic keywords to match the standard names to achieve preliminary classification of the product data. The aforementioned feature keywords include name-type feature keywords that can reveal the product name and attribute-type feature keywords that can reveal the product attributes. By matching name-type feature keywords with standard names, product data is classified into the standard names that match them. S3. Classify the characteristic keywords in the same category to obtain different characteristic features. Add the characteristic features containing more than a set threshold to the product attributes to create new product attributes or improve existing product attributes, thus forming product tags. After processing the characteristic keywords using feature embedding methods or text feature processing methods, similarity matching, classification or clustering are performed to achieve characteristic classification and obtain different characteristic features. S4. Select the top-ranked feature features in terms of the number of featured keywords as core features, convert the core features into word vectors, and then perform nearest neighbor search. S5. By using nearest neighbor search, the product data is classified into the standard name that is most similar to the core feature and its characteristic keyword to achieve product data reclassification. S6. Repeat steps S4 and S5 until the classification no longer changes.

2. The method for establishing product labels according to claim 1, characterized in that, Step S1 also includes establishing a preliminary name correspondence between standard names and aliases through integration, and obtaining name characteristics by performing dimensionality reduction processing on information that can be obtained about product names. The name characteristics are then matched with the standard names to improve the name correspondence. If there is no matching standard name, a new standard name is established accordingly.

3. The method for establishing product labels according to claim 1, characterized in that, The name characteristics are matched with the standard name using similarity algorithms, machine learning models, or manual review.

4. A system for establishing a product label according to any one of claims 1-3, characterized in that, The system includes a corresponding module, a processing module, a first matching module, and a second matching module. The corresponding module is used to establish the correspondence between the standard name of the product and the product attributes. The processing module is used to extract characteristic keywords from the data about the product, match the characteristic keywords with the standard name to achieve preliminary classification of the product data, classify the characteristic keywords in the same category to obtain different characteristic features, and the first matching module adds characteristic features containing more than a set threshold to the product attributes, thereby establishing new product attributes or improving existing product attributes. The second matching module selects several feature features with the highest number of feature keywords as core features, and uses nearest neighbor search to classify the product data into standard names that are most similar to the core features and feature keywords.

5. The product label creation system according to claim 4, characterized in that, It also includes a standard module, which is used to establish the name correspondence between standard names and aliases, and to obtain name characteristics by performing dimensionality reduction processing on information that can be obtained about product names. The name characteristics are then matched with standard names to improve the name correspondence. If there is no matching standard name, a new standard name is established accordingly.

6. A computer-readable storage medium storing computer instructions thereon, characterized in that, When executed by the processor, this instruction implements the method for creating a product label as described in any one of claims 1-3.

7. A terminal device, comprising a memory and a processor, characterized in that, The processor is configured to implement the method for creating a product label according to any one of claims 1-3 when executed based on instructions stored in the memory.

Images