Designated commodity detection method, designated commodity detection system, and program

Abstract

A designated commodity detection method, a designated commodity detection system, and a program are provided. [Solution] A specified product detection method including: at least one processor acquiring multiple product datasets, each of the multiple product datasets including product descriptions for multiple products, the product descriptions including descriptions of the ingredients contained in the corresponding products; acquiring one or more keywords, each of the one or more keywords indicating one or more specified ingredients; extracting one or more suspect products from the multiple products by performing a keyword search using the one or more keywords on the multiple product datasets; detecting one or more non-containing products from the one or more suspect products; and outputting a list of containing products by excluding the one or more non-containing products from the one or more suspect products.

Description

[Technical Field]

[0001] The present disclosure relates to a designated commodity detection method, a designated commodity detection system, and a program. [Background technology]

[0002] A wide variety of products are traded on e-commerce (electronic commerce) sites such as internet auctions. There is a possibility that illegal products prohibited by law may be listed on these e-commerce sites. Detecting such inappropriate listings quickly and accurately is important for protecting consumer safety.

[0003] Patent Document 1 discloses a method for checking for illegal products at the time of listing by extracting character strings contained in the listing information and comparing the extracted character strings with prohibited character strings. Prohibited character strings are, for example, "CD-ROM copy" or "User registration not permitted." By searching the listing information using keywords such as prohibited character strings, illegal products such as illegally copied software products can be detected. As a result, it becomes possible to stop the listing of illegal products. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2006-53809 Summary of the Invention [Problem to be solved by the invention]

[0005] For example, if a product is a health food and contains ingredients that are equivalent to pharmaceuticals, it will be considered an unapproved and unlicensed drug and become an illegal product. In other words, while it is legal for a certain ingredient to be contained in a pharmaceutical product, a health food that contains that ingredient may be an illegal product.

[0006] In this way, it is difficult to detect illegal products by simply searching listing information using the names of ingredients as keywords, especially when determining the suitability of a product based on the ingredients contained in it. This problem is not limited to the detection of illegal products, but can also arise when trying to detect products that contain certain specified ingredients.

[0007] The present disclosure aims to provide a designated commodity detection method, a designated commodity detection system, and a program that can detect designated commodities that contain designated ingredients. [Means for solving the problem]

[0008] A designated commodity detection method according to one aspect of the present disclosure includes: at least one processor acquiring a plurality of commodity data sets, each of which includes a product description for a plurality of commodities, the product descriptions including a description of ingredients contained in the corresponding commodity; acquiring one or more keywords, each of which indicates one or more designated ingredients; extracting one or more suspect commodities from the plurality of commodities by performing a keyword search using the one or more keywords on the plurality of commodity data sets, each of which includes at least one of the one or more designated ingredients in the corresponding product description; detecting one or more non-containing commodities that do not contain the one or more designated ingredients from the one or more suspect commodities based on the product descriptions corresponding to each of the one or more suspect commodities; and excluding the one or more non-containing commodities from the one or more suspect commodities, thereby 1 or more and outputting a list of products containing at least one of the specified ingredients.

[0009] A designated commodity detection method according to one aspect of the present disclosure includes: acquiring, by at least one processor, a plurality of commodity data sets, each of which includes a product description for a plurality of commodities, the product descriptions including a description of ingredients contained in the corresponding commodities; acquiring one or more keywords, each of which indicates one or more designated ingredients; and extracting one or more suspect commodities from the plurality of commodities by performing a keyword search using the one or more keywords on the plurality of commodity data sets, wherein each of the suspect commodities is a product description for the corresponding commodities. The method includes: including at least one of the one or more specified ingredients in the description; detecting a negative description from the product description of each of the one or more suspect products that describes non-containing ingredients that the corresponding suspect product does not contain; detecting the one or more non-containing ingredients described in each negative description; and displaying a confirmation screen on a display for determining whether each suspect product contains the one or more specified ingredients, wherein the confirmation screen is configured to display the product description of the corresponding suspect product, and the one or more non-containing ingredients are highlighted in the product description on the confirmation screen.

[0010] A designated commodity detection system according to one aspect of the present disclosure includes at least one memory that stores computer program code and at least one processor, and the at least one processor executes the computer program code to acquire a plurality of commodity data sets, each of the plurality of commodity data sets including product descriptions for a plurality of commodity products, the product descriptions including descriptions of ingredients contained in the corresponding commodity products; acquire one or more keywords, each of the one or more keywords indicating one or more designated ingredients; extract one or more suspect commodity products from the plurality of commodity data sets by performing a keyword search using the one or more keywords on the plurality of commodity data sets, each of the suspect commodity products including at least one of the one or more designated ingredients in the corresponding product description; detect one or more non-containing commodities that do not contain the one or more designated ingredients from the one or more suspect commodities based on the product descriptions corresponding to each of the one or more suspect commodities; and exclude the one or more non-containing commodities from the one or more suspect commodities, thereby 1 or more and outputting a list of products containing at least one of the specified ingredients.

[0011] A designated product detection system according to one aspect of the present disclosure includes at least one memory that stores computer program code and at least one processor, and the at least one processor executes the computer program code to acquire a plurality of product data sets, each of which includes product descriptions for a plurality of products, and each of which includes a description of ingredients contained in the corresponding product; acquire one or more keywords, each of which indicates one or more designated ingredients; and perform a keyword search using the one or more keywords on the plurality of product data sets to identify one or more designated ingredients from among the plurality of products. The system is configured to: extract the above suspect products, each of which contains at least one of the one or more specified ingredients in the corresponding product description; detect a negative description from the product description of each of the one or more suspect products that explains non-containing ingredients that the corresponding suspect product does not contain; detect the one or more non-containing ingredients explained in each negative description; and display a confirmation screen on a display for determining whether each of the suspect products contains the one or more specified ingredients, wherein the confirmation screen is configured to display the product description of the corresponding suspect product, and the one or more non-containing ingredients are highlighted in the product description on the confirmation screen.

[0012] A program according to one aspect of the present disclosure includes the following steps: acquiring, in at least one processor, a plurality of product data sets corresponding to a plurality of products, each of the product data sets including a product description of ingredients contained in the corresponding product; acquiring one or more keywords, each of the one or more keywords indicating one or more specified ingredients; and extracting one or more suspect products from the plurality of products by performing a keyword search using the one or more keywords on the plurality of product data sets, wherein each of the suspect products includes the following in the corresponding product description: 1 or moreand, from among the one or more suspected products, based on the product descriptions corresponding to each of the one or more suspected products, 1 or more By detecting one or more non-containing products that do not contain the specified ingredient, and excluding the one or more non-containing products from the one or more suspected products, 1 or more and outputting a list of products containing at least one of the specified ingredients.

[0013] A program according to one aspect of the present disclosure causes at least one processor to: acquire multiple product datasets, each of which includes product descriptions for multiple products, and which includes descriptions of ingredients contained in the corresponding products; acquire one or more keywords, each of which indicates one or more specified ingredients; extract one or more suspect products from the multiple products by performing a keyword search on the multiple product datasets using the one or more keywords, wherein each of the suspect products includes at least one of the one or more specified ingredients in the corresponding product description; detect, from the product descriptions of each of the one or more suspect products, negative descriptions describing non-containing ingredients that the corresponding suspect product does not contain; detect the one or more non-containing ingredients described in each negative description; and display a confirmation screen on a display to determine whether each of the suspect products contains the one or more specified ingredients, wherein the confirmation screen is configured to display the product descriptions of the corresponding suspect products, and the one or more non-containing ingredients are highlighted in the product descriptions on the confirmation screen. [Brief explanation of the drawings]

[0014] [Figure 1] FIG. 1 is a schematic diagram for explaining a designated commodity detection system according to an embodiment. [Figure 2] FIG. 2 is an explanatory diagram illustrating an example of a product sales page. [Figure 3]FIG. 3 is a table illustrating some of the designated ingredients for health foods. [Figure 4] FIG. 4 is a flow diagram illustrating a method for generating a classification model. [Figure 5] FIG. 5 is a flow chart showing a method for detecting a designated product. [Figure 6] FIG. 6 is an explanatory diagram illustrating a confirmation screen for detecting a designated product. DETAILED DESCRIPTION OF THE INVENTION

[0015] Examples of a designated commodity detection method, a designated commodity detection system 11, and a program will be described with reference to Figures 1 to 6. The present invention is not limited to these examples, but is defined by the claims, and is intended to include all modifications within the meaning and scope equivalent to the claims. Furthermore, numerals such as "first" and "second" used in the following description are used to distinguish the components to which they are attached, and are not used to rank the components.

[0016] [Outline of the designated product detection system] The designated commodity detection system (hereinafter simply referred to as the "detection system") 11 shown in FIG. 1 includes a designated commodity detection device (hereinafter simply referred to as the "detection device") 20. The detection device 20 is configured to detect designated commodities containing one or more designated ingredients from among multiple commodities listed on an e-commerce site or multiple commodities sold on an e-commerce site. When the one or more designated ingredients include multiple designated ingredients, the multiple designated ingredients are also referred to as a designated ingredient group. When the designated ingredient is an illegal ingredient, the detection device 20 detects illegal commodities containing the illegal ingredient from among the multiple commodities.

[0017] The detection device 20 may be realized as a computer including at least one processor 21, at least one memory 22, and a communication IF 23. The following describes a case where the detection device 20 includes one processor 21 and one memory 22. The communication IF 23 enables communication with other devices via a network.

[0018] The memory 22 stores a program 24 executed by the processor 21 and detection data 25. The program 24 includes an application and an operating system. The processor 21 performs various functions by executing processes based on the program 24. The detection data 25 includes various data used when detecting the designated product.

[0019] The detection system 11 may include a database 13. The database 13 may be held in a web server 14 that provides an e-commerce site. The web server 14 may be a computer that includes the same components as the detection device 20 (e.g., at least one processor, at least one memory, and a communication IF). Alternatively, part or all of the database 13 may be stored in the memory 22 of the detection device 20.

[0020] The detection system 11 may include a plurality of purchaser terminals 15 that can access an e-commerce site (web server 14) via a network. The e-commerce site may be, for example, a shopping mall or a stand-alone store, or an online auction site, but is not limited to these. The detection system 11 may also include a plurality of seller terminals 16 that provide products for sale on the e-commerce site via a network.

[0021] The buyer terminal 15 and the seller terminal 16 may be, for example, a personal computer or a mobile terminal such as a smartphone or tablet. Each of the buyer terminal 15 and the seller terminal 16 may have a display as an output device. Each display may have a touch panel as an input device.

[0022] The database 13 may include a seller data table, a buyer data table, and a product data table. The seller data table stores multiple seller data sets for each of multiple sellers who offer products. The buyer data table stores multiple buyer data sets for each of multiple buyers who have purchased products using the EC site and multiple EC site registrants (potential buyers). The product data table stores multiple product data sets for each of multiple products. One product data set includes one or more data items for the corresponding product. Each of the multiple product data sets corresponds to multiple products listed on the EC site or multiple products sold on the EC site.

[0023] The detection system 11 may include a learning device 30 that can communicate with the detection device 20 via a network. Similar to the detection device 20, the learning device 30 may be realized as a computer that includes at least one processor 31, at least one memory 32, and a communication IF 33. Below, a case will be described in which the learning device 30 includes one processor 31 and one memory 32.

[0024] The memory 32 stores a program 34. The program 34 includes an application and an operating system. The processor 31 executes processing based on the program 34 to realize various functions. The memory 32 may also store one or more learning models 35 for training. The one or more pieces of generated learning data 35 may be stored in the memory 22 of the detection device 20.

[0025] The detection system 11 may include a language model 17 that can communicate with the learning device 30 via a network. The language model 17 may be a large language model (LLM) or a small language model (SLM). The large language model is a language model trained using a large amount of text data. The small language model is a language model that is smaller in scale (e.g., has fewer parameters) than the large language model. The language model 17 may be a general-purpose natural language processing (NLP) model that can be adapted to various natural language processing tasks, such as information extraction, text summarization, text generation, or question and answering, depending on an input prompt.

[0026] The language model 17 is configured to generate data according to the instruction when a prompt including some instruction is input, and output the data as a completion. In this example, the learning device 30 generates a prompt and inputs it to the language model 17, and then the learning device 30 acquires the completion generated by the language model 17. Alternatively, the detection device 20 may use a language model 17 that is not included in the detection system 11 via a network.

[0027] The detection system 11 may include another information processing device for performing at least one of creating a prompt, inputting the prompt into the language model 17, and obtaining a completion, instead of the learning device 30. The other information processing device may be a computer having the same components as the detection device 20 (for example, at least one processor, at least one memory, and a communication IF).

[0028] The detection system 11 may include an operator terminal 40 that can communicate with the detection device 20 via a network. The operator who operates the operator terminal 40 may be, for example, an operator of an EC site or an administrator who manages products listed on the EC site.

[0029] Similar to the detection device 20, the worker terminal 40 may be realized as a computer including at least one processor 41, at least one memory 42, and a communication IF 43. The following describes a case where the worker terminal 40 includes one processor 41 and one memory 42. The worker terminal 40 may include a display 46. Alternatively, an external display 46 may be connected to the worker terminal 40.

[0030] [Product Dataset] Each of the plurality of products may be categorized into one of a plurality of product categories, which may include, for example, at least one of health foods, pharmaceuticals, quasi-drugs, and cosmetics, but are not limited to these.

[0031] Each product data set may include, but is not limited to, information related to the corresponding product, such as the product name, product category, listing date, sales start date, sales price, and product description. The product category may be manually entered in advance by the seller or an operator. The product description may include a description of the ingredients contained in the corresponding product.

[0032] 2, the product description may include multiple descriptions to be displayed on a sales page 50 of the product on an e-commerce site. The sales page 50 may include, for example, a tagline 51, a title 52, a detailed description 53, and one or more images 54. The one or more images 54 may be still images or videos.

[0033] If at least some of the one or more images 54 include a character string, the detection device 20 may be configured to extract the character string from the one or more images 54. The detection device 20 may add the extracted character string extracted from the one or more images 54 to a product description in the product dataset. In this case, the product description includes the character string included in each of the tagline 51, the title 52, and the detailed description 53, as well as the extracted character string.

[0034] One or more designated ingredients may be set for at least some of the multiple product categories. A product category with one or more designated ingredients set is called a designated product category. For example, if multiple designated ingredients (referred to as a "first designated ingredient group") are designated for a health food, which is one of the multiple product categories, the first designated ingredient group may be illegal ingredients that are illegal in the health food product category. Illegal ingredients are, for example, ingredients that fall under pharmaceuticals and are prohibited under the Pharmaceutical Affairs Act. If another multiple designated ingredients (referred to as a "second designated ingredient group") are designated for a pharmaceutical, the second designated ingredient group may be, for example, pharmaceutical ingredients that are not approved in the country or region where the pharmaceutical is sold.

[0035] FIG. 3 illustrates a portion of a first designated group of ingredients for a health food. The first designated group of ingredients indicates the names of pharmaceutical ingredients that would violate the Pharmaceutical Affairs Law if contained in the health food. Each of the multiple designated ingredients can be set as multiple keywords. In this case, multiple keywords may be set for one designated ingredient. For example, two or more of the katakana name, English name, kanji name, and alias name may be set as keywords for one designated ingredient. The set keywords may be included in the detection data 25 as a search keyword set.

[0036] The one or more designated ingredients may be non-illegal ingredients that are prohibited from trading on e-commerce sites. Non-compliant products containing non-compliant ingredients may include illegal goods. Furthermore, the one or more designated ingredients may be set for the purpose of detecting specific products, not just illegal ingredients. A specific product may be, for example, a product containing a specific allergy-causing substance, or a product containing an ingredient added for a specific purpose. In this case, multiple designated ingredient groups corresponding to each of multiple purposes may be designated for one designated product category (e.g., health food).

[0037] [Machine learning model for classification] If the product dataset does not include data indicating the product category, the detection device 20 or the learning device 30 may generate data indicating the product category. The data indicating the product category may be generated using a machine learning model for classification (hereinafter referred to as a "classification model"). The generated data indicating the product category may be added to the product description.

[0038] The learning device 30 may generate a classification model (a machine learning model for classification) by training a learning model before training or a pre-trained learning model with learning data. The classification model may be a binary classification model configured to output, when a product description of a certain product is input, whether the product falls into one of multiple product categories.

[0039] For example, if the multiple product categories include one product category (first category: health foods) and another product category (second category: pharmaceuticals), the first classification model may be configured to output whether a product is a health food, and the second classification model may be configured to output whether a product is a pharmaceutical. Alternatively, the first classification model may be configured to classify which of the multiple product categories a product belongs to. Alternatively, the other classification model may be configured to classify which of the multiple product categories a product belongs to, or whether a product does not belong to any of the product categories.

[0040] If the multiple products include a set of multiple types of products, the classification model may be a multi-class classification model. The multi-class classification model may be configured to output, when a product description of a product is input, which of the following three groups (classes) 1) to 3) the product belongs to.

[0041] 1) A group of target products belonging to product category 1 (target product group). 2) A group of non-eligible products that belong to a product category other than product category 1 (non-eligible product group).

[0042] 3) A group of set products that includes the target product (set product group). The set product group in 3) above may be a group of all set products, regardless of whether or not they include the target product.

[0043] The learning device 30 may use multiple product data sets acquired from the database 13 as learning data. The multiple product data sets used as learning data are referred to as multiple training product data sets (hereinafter simply referred to as "training data sets"), and product descriptions of products corresponding to the training data sets are also referred to as training product descriptions (hereinafter simply referred to as "training instructions"). The training instructions include descriptions of the ingredients contained in the corresponding products.

[0044] The classification model may be, for example, a supervised learning model based on a decision tree algorithm that performs the classification task. By way of example, the classification model may be, but is not limited to, a model using Random Forest, GBDT, XGBoost, or lightGBM.

[0045] The training data may include multiple ground truth datasets for use as ground truth data, each of which may include a training description of a corresponding product and a ground truth label indicating whether the corresponding product is a target product, a non-target product, or a bundled product.

[0046] When the number of correct answer datasets to be used as training data is insufficient, the detection device 20 or the learning device 30 may generate a correct answer dataset. For example, when a part of the training datasets acquired from the database 13 does not include data indicating a product category that will be a correct label, the detection device 20 or the learning device 30 may generate data indicating a product category.

[0047] An example will be described below in which the learning device 30 generates data indicating product categories using the language model 17. In this case, the learning device 30 generates prompts to be input into the language model 17 in order to generate multiple correct answer datasets. The prompts in this example are generated so that, as information missing from the correct answer dataset, an answer can be obtained for a certain product category, as to whether the corresponding product falls into one of the target product, non-target product, and set product.

[0048] The prompt may include a training explanation as information for obtaining an answer. In this case, language model 17 uses the training explanation as a basis for determining whether the corresponding product is a target product in a certain product category or whether the target product is part of a set of products. The prompt may include multiple training explanations to obtain answers for multiple products at once, or may include a single training explanation to obtain an answer for a single product.

[0049] In one example, the prompt includes a plurality of training descriptions for each of a plurality of products and an instruction statement for causing the language model 17 to output an answer. The answer to be output by the instruction statement is whether the plurality of products described by each of the plurality of training descriptions corresponds to a target product (health food), a non-target product (other than health food), or a set product (including health food). The instruction statement can be written, for example, as follows: "Based on the product description describing the product, please answer whether the product corresponds to a health food, a set product including health food, or another product."

[0050] When the learning device 30 inputs the prompt created in this way into the language model 17, the language model 17 outputs a completion including an answer to the instruction sentence. The learning device 30 obtains an output result corresponding to the prompt from the language model 17. Then, the learning device 30 extracts one or more products determined to be target products from among the multiple products based on the output result of the language model 17. Then, the learning device 30 adds product category data (e.g., "health food") to the training dataset for each extracted product, thereby completing the generation of the correct answer dataset.

[0051] [How to generate a classification model] An example of a method in which the learning device 30 generates a classification model will be described with reference to FIG. In step S11, the learning device 30 acquires a plurality of product data sets to be used as learning data from the database 13. Next, in step S12, the learning device 30 prepares learning data to be input to the pre-training learning model.

[0052] The preparation of the training data may include preprocessing the training data. The preprocessing may include checking for missing data in each training dataset. The checking for missing data may include, for example, checking for the presence or absence of correct labels, i.e., whether or not product categories are included.

[0053] The preparation of the learning data may include extracting, from the acquired product datasets, those containing product categories that are correct labels as training datasets. Furthermore, to check for defects, the operator may check whether the correct labels of at least some of the training datasets are correct. Additionally, the operator may create correct labels for some product datasets that do not contain correct labels.

[0054] The preparation of the training data may include dividing a plurality of training datasets into training data, validation data, and test data, in which at least one of the datasets for which an operator has generated a correct label and the dataset for which the operator has confirmed the correct label may be used as the test data.

[0055] In step S13, the learning device 30 determines whether the amount of training data (e.g., the number of training datasets) is equal to or greater than a threshold. The threshold can be set by statistically calculating the amount of data required to train a learning model. If the amount of data is less than the threshold, the learning device 30 determines that the amount of data is insufficient (step S13=NO) and proceeds to step S14 to add training data. The training data to be added can be obtained, for example, by the learning device 30 acquiring multiple new product datasets from the database 13.

[0056] Alternatively, or in addition, the additional training data can be generated by adding data indicating product categories to product descriptions that do not include a product category that is a correct label among the multiple acquired product datasets, as described above. For example, the learning device 30 can generate product category data using the language model 17, as described above.

[0057] The learning device 30 may acquire, as at least a part of the learning data, a dummy product data set generated by an operator for learning purposes. For example, an operator may generate a dummy product description by adding a description of one or more specified ingredients to an arbitrary product description. In particular, when there are few product data sets for violating products, it is useful to generate correct answer data for the violating products using the dummy product data set.

[0058] After adding the training data in step S14, the learning device 30 may return to step S12 to prepare the training data again. After the preparation of the training data is completed, the learning device 30 proceeds to step S13. If the amount of data is equal to or greater than the threshold in step S13, the learning device 30 determines that the amount of data is sufficient (step S13=YES) and proceeds to step S15.

[0059] In step S15, the learning device 30 inputs the training data into the learning model to train the learning model. This updates the weights in the learning model. In the following step S16, the learning device 30 tunes the parameters of the learning model using the validation data.

[0060] In step S17, the learning device 30 tests the learning model using test data. For example, the learning device 30 calculates the accuracy of the learning model as a result of the test. An index of accuracy may be, for example, but is not limited to, AUC (Area Under the Curve). Thereafter, the learning device 30 may determine whether the accuracy of the learning model is equal to or greater than a predetermined accuracy threshold. The accuracy threshold may be stored in the memory 32. If the accuracy of the learning model is less than the accuracy threshold, the learning device 30 may determine that the learning of the learning model is insufficient and proceed to step S14 to add training data.

[0061] When the accuracy of the learning model is equal to or greater than the threshold, the learning device 30 determines that the learning model has been properly trained. In this case, the learning device 30 proceeds to step S18, stores the trained learning model in the memory 32 as a trained classification model, and ends the process.

[0062] [Specified Item Detection Method] With reference to FIG. 5, an example will be described in which the detection device 20 detects illegal products containing multiple designated ingredients specified for health foods from among multiple products belonging to multiple product categories.

[0063] In step S21, the detection device 20 acquires a plurality of product data sets from the database 13 or the memory 22. The plurality of product data sets may include data relating to products that have not yet been put up for sale, or may include data relating to products that are currently on sale.

[0064] 6, the detection device 20 may extract text information included in each product data set as a product description. The product description may include, for example, a character string 61 extracted from a tagline 51 included in the sales page 50 shown in FIG. 2, a character string 62 extracted from a title 52, a character string 63 extracted from a detailed description 53, and a character string 64 extracted from an image 54.

[0065] In step S22, the detection device 20 acquires one or more keywords from the memory 22. In this example, the detection device 20 acquires a search keyword set including multiple keywords each indicating multiple specified ingredients. The search keyword set includes, for example, the names of various ingredients that would be illegal if contained in health foods, as shown in FIG. 3.

[0066] In step S23, the detection device 20 performs a keyword search using the multiple keywords acquired in step S22 on the multiple product data sets acquired in step S21. More specifically, the detection device 20 searches the multiple product descriptions extracted from each of the multiple product data sets using the multiple keywords, and extracts product descriptions that include at least one keyword. Products corresponding to the extracted product descriptions are referred to as suspect products. Therefore, the keyword search in step S23 extracts one or more suspect products that include at least one of the one or more specified ingredients in their product descriptions.

[0067] In step S24, the detection device 20 classifies the extracted one or more suspect products into one of three groups based on the corresponding product descriptions: a target product (health food) group, a non-target product (products other than health foods) group, or a set product group.

[0068] More specifically, the detection device 20 inputs the product description of each suspect product into a classification model trained to classify health foods, generated using the method shown in Figure 4. The classification model then outputs a classification result for the corresponding suspect product based on the input product description. The detection device 20 obtains a list of target products extracted as health foods by obtaining the classification result for each of one or more suspect products from the classification model.

[0069] The classification in step S24 may be performed before the keyword search in step S22. In this case, a keyword search is performed on multiple target products classified as health foods to extract one or more suspect products that contain at least one specified ingredient from among the health foods.

[0070] In step S25, the detection device 20 performs named entity recognition (NER) on the product descriptions of each target product, i.e., suspect products classified as health foods, and extracts ingredient names from the extracted proper nouns. As a result, one or more ingredient names are detected from the product descriptions of each suspect product.

[0071] The detection device 20 executes the processes of steps S26 to S30 for each of a plurality of suspect products. In the following, an example will be described in which the detection device 20 detects a plurality of ingredient names from one product description of one suspect product.

[0072] In step S26, detection device 20 extracts one or more designated ingredients that match the keywords from among the ingredient names in one product description. In Fig. 6, the shaded proper nouns in the product description are the extracted designated ingredients.

[0073] In step S27, the detection device 20 detects, from the product description 1, a negative description that explains the ingredients that the product 1 does not contain. The negative description is a description that explicitly states the ingredients that the product does not contain, but may also include a description that alludes to the ingredients that the product does not contain. In the example shown in Figure 6, the sentence "This product does not contain silicone, mineral oil, ethanol, petroleum-based surfactants, wheat-, egg-, or shellfish-derived ingredients, cetanol, chlorphenesin, methylisothiazolinones, UV absorbers, UV scattering materials, parabens, fragrances, or coloring agents" is a negative description 66.

[0074] The detection of negative descriptions may be realized by, for example, including a plurality of preset negative expressions such as "not used," "not used," and "not including" in the detection data 25, and having the detection device 20 execute a search for product descriptions using the negative expressions as keywords. Alternatively, the detection device 20 may detect negative descriptions using a language model for detecting negative sentences.

[0075] In step S28, the detection device 20 detects, from among the plurality of component names, the non-containing components described in the negative description 66. In Fig. 6, among the shaded designated components, the component names enclosed in square frames are non-containing components.

[0076] In step S29, the detection device 20 detects, from the product description, a positive description that explains the ingredients contained in the product. The positive description is a description that explicitly states the ingredients contained in the product, but may also include a description that alludes to the ingredients. In the example shown in FIG. 6, the sentence "Contains the latest popular ingredients: A carefully selected blend of retinol, arbutin, placenta, fullerene, plant stem cells, human ceramide (booster), vitamin A derivative, collagen, Centella asiatica extract, and glycyrrhizinic acid licorice" is a positive description 67.

[0077] The detection of positive descriptions may be realized by, for example, including a plurality of preset ingredient expressions, such as "mixture" and "ingredients," in the detection data 25, and having the detection device 20 search the product description using the ingredient expressions as keywords. Alternatively, the detection device 20 may detect positive descriptions using a language model for detecting ingredient descriptions of products.

[0078] In step S30, the detection device 20 detects, from among the plurality of ingredient names, the contained ingredients described in the affirmative explanation 67. In Fig. 6, the ingredient names displayed in a shaded manner in the affirmative explanation 67 are the contained ingredients.

[0079] In step S31, the detection device 20 detects one or more non-containing products that do not contain one or more specified ingredients from among the one or more suspect products based on the product descriptions corresponding to each of the one or more suspect products. More specifically, based on the results of steps S28 and S30, the detection device 20 determines that the corresponding suspect product is a non-containing product if all of the one or more specified ingredients included in each product description are non-containing ingredients included in the negative description 66 and one or more specified ingredients included in the product description are not included as contained ingredients in the positive description 67.

[0080] Steps S29 and S30 may be omitted. In this case, the detection device 20 may determine that the corresponding suspect product is a non-containing product if all of the one or more specified ingredients included in each product description are non-containing ingredients that are only included in the negative description 66. If the specified ingredient is not included in the negative description and the specified ingredient is included in the positive description, the detection device 20 may determine that the corresponding suspect product is a containing product.

[0081] In step S32, the detection device 20 excludes the non-containing products detected in step S31 from one or more suspect products. Then, in step S33, the detection device 20 outputs a list of violating products in which the remaining suspect products after excluding the non-containing products are designated as violating products (in this example, illegal products), and ends the process. The violating product list in this example is a list of containing products that contain at least one of the multiple specified ingredients.

[0082] The detection device 20 may be configured to display a confirmation screen 60 including the results of the detection and extraction in steps S25 to S30 on the display 46 of the worker terminal 40 in response to the work of the worker. The confirmation screen 60 is configured to display, for example, a product description of one of the multiple product data sets.

[0083] In the confirmation screen 60, one or more of the non-containing ingredients and the containing ingredients may be highlighted with different marks. The different marks may be displayed in different colors, such as red and blue, or may be displayed in different font styles, such as underlined and bold. The confirmation screen 60 may include a legend 65 that explains the display of the marks for the non-containing ingredients and the containing ingredients.

[0084] In Figure 6, the names of ingredients other than the non-containing ingredients in the negative description are highlighted as containing ingredients. Alternatively, the non-containing ingredients in the negative description and the containing ingredients in the positive description may be highlighted with different marks. Alternatively, the non-containing ingredients in the negative description, the containing ingredients in the positive description, and the other ingredient names may be displayed using different marks.

[0085] On the confirmation screen 60 for confirming the contained ingredients, or on a confirmation screen for confirming keywords separate from the confirmation screen 60, the ingredient names extracted by named entity extraction in step S25 and one or more specified ingredients extracted using keywords in step S26 may be displayed with different marks. In this case, the operator can visually find, for example, the names of specified ingredients that are not included in the keyword set, or ingredient names that do not match the keywords due to a typographical error in the description. In this way, by extracting ingredient names by named entity extraction, it is possible to check the appropriateness of keyword settings and to discover missing keywords.

[0086] If the designated product detected as a violating or illegal product is one of the above, the operator may notify the seller of the designated product and request that the product be discontinued, or the operator may directly discontinue the listing or sale of the designated product. If the designated product is a product containing a specific ingredient, the operator may continue to list or sell the designated product after adding information appropriate to its intended use. For example, the designated product may be labeled as containing the specific ingredient and sold, or the specific ingredient may be added to the product description (e.g., tagline) as a contained ingredient so that the product will be searched for by the specific ingredient.

[0087] [Effects of the present disclosure] When a product description includes a description of the product's ingredients, the description may include both positive descriptions of the ingredients the product contains and negative descriptions of the ingredients the product does not contain. When such a product description is searched using keywords, the keywords will match not only the ingredients contained in the product but also the ingredients not contained in the negative descriptions. As a result, a product that does not contain the specified ingredient will be mistakenly detected as a product that does contain the specified ingredient.

[0088] In this regard, the present disclosure extracts suspect products from among multiple products through keyword search, then detects non-containing products that do not contain the specified ingredient based on the product descriptions of the suspect products.By excluding the non-containing products from the suspect products, it becomes possible to detect only the containing products that contain the specified ingredient.

[0089] Product descriptions may include the names of other ingredients that are not included in the product but are used to explain its efficacy, such as in Figure 6, "Ingredients that are effective against dullness include niacinamide, tretinoin, and retinol. However, this product contains a particularly high-purity blend of pure hydroquinone from among our carefully selected ingredients." In this regard, detecting positive descriptions in addition to negative descriptions and explicitly detecting only products that contain the specified ingredients can more accurately detect designated products. On the other hand, excluding non-containing products from suspect products based on negative descriptions without detecting positive descriptions can more broadly extract products that may contain the specified ingredients.

[0090] If the product is a set product, the product description of the set product includes a description of each of the multiple products included in the set product. Therefore, by classifying the set products from among the multiple products, the set products can be detected separately from the target product. This improves the detection accuracy of the target product. For example, for a set product, it may be detected whether the worker specifies a specific product.

[0091] [Effects of this disclosure] According to the present disclosure, the following effects can be achieved. (1) By excluding non-containing products that do not contain the specified ingredients from among suspected products that contain the specified ingredients in their product descriptions, it is possible to more accurately detect products that contain the specified ingredients.

[0092] (2) By detecting negative descriptions in the product descriptions, it is possible to determine that the product does not contain the names of ingredients included in the negative descriptions. Therefore, if all of the specified ingredients included in the product descriptions are only included in negative descriptions, it is possible to determine that the corresponding suspect product does not contain any of the ingredients.

[0093] (3) By detecting negative and positive descriptions in the product description, it is possible to determine whether the specified ingredient in the product description is a contained ingredient contained in the suspected product or a non-contained ingredient not contained in the suspected product. Therefore, if the specified ingredient contained in the product description is only contained in the negative description and is not contained in the positive description, it is possible to determine that the corresponding suspected product is a non-contained product. Furthermore, if the specified ingredient is not contained in the negative description and is contained in the positive description, it is possible to determine that the corresponding suspected product is a contained product.

[0094] (4) By performing named entity extraction on the product description, it is possible to detect one or more ingredient names from the product description. This makes it possible to detect a specific ingredient from the ingredient names.

[0095] (5) By displaying the confirmation screen 60 on the display 46, the worker can confirm the detected negative explanation, positive explanation, contained components, and non-contained components. This allows the worker to confirm whether the detection by the detection device 20 is being performed appropriately.

[0096] (6) If a designated ingredient is an illegal ingredient in a certain product category, the system can detect illegal products from among multiple products by outputting a list of products that contain the illegal ingredient from among the products in that product category. As a result, the listing and sale of illegal products on e-commerce sites can be stopped.

[0097] (7) By setting designated ingredients for each product category, it is possible to detect products containing the designated ingredients for each product category. (7) By extracting character strings from the image 54, it becomes possible to detect the names of ingredients contained in the image 54.

[0098] (8) By classifying multiple products into target products belonging to a certain product category based on the corresponding product descriptions and non-target products belonging to other product categories, target products can be detected for each product category. This makes it possible to exclude non-target products from keyword searches and from ingredient name extraction. Furthermore, by classifying multiple products into set products, it becomes possible to perform separate detection processing for set products.

[0099] (9) By using a classification model, product classification processing can be performed more efficiently. (10) By using the language model 17, the correct answer data set can be generated more efficiently.

[0100] This embodiment can be modified as follows: This embodiment and the following modifications can be combined and implemented within the scope of technical compatibility. [Change Example 1] In the method illustrated in Fig. 5 and the system and program for implementing this method, steps S31 to S33 may be replaced by the display of a confirmation screen 60 illustrated in Fig. 6. In this case, the determination of whether the counterfeit product is a non-containing product or a containing product may be made by an operator instead of the detection device 20. Alternatively, the operator may use the confirmation screen 60 to confirm at least some of the multiple suspect products.

[0101] [Change Example 2] In the method illustrated in FIG. 5 and the system and program for implementing this method, steps S27, S28, and S32 may be omitted, and instead of step S31, the detection device 20 may detect only products that contain the specified ingredient in the positive description as violating products.

[0102] [Change Example 3] When detecting one or more suspect products by keyword search in step S23, the detection device 20 may mark the ingredient names that match the keywords in the product descriptions of each suspect product. In this case, the extraction of ingredient names by named entity extraction in step S25 and the extraction of specified ingredients in the ingredient names in step S26 may be omitted.

[0103] [Change Example 4] Products detected based on specified components are not limited to products containing undesirable components such as illegal components or allergens. For example, the technology disclosed herein can be used to detect products containing specific desirable components or products containing selectively used components (e.g., antipyretic analgesics containing specific antipyretic analgesic components, or antiallergic drugs containing specific antiallergic components).

[0104] [Change Example 5] When detecting a positive or negative description, other factors such as the part of the plant used may be included in the detection, in addition to the name of the ingredient, since in particular, only a specific part of the plant may be used as a medicine for plant-derived substances.

[0105] [Change Example 6] The method for detecting negative and positive descriptions may differ depending on the product category. For example, if there is a characteristic description method for a certain product category, a detection method based on that description method can be adopted.

[0106] [Other change examples] The flow charts and block diagrams of the present disclosure illustrate the architecture, functionality, and operation of devices, systems, methods, and programs according to embodiments of the present disclosure. Each step in the flow charts and each component in the block diagrams may correspond to a portion of computer program code, including one or more instructions for implementing logical functions. In other embodiments, some of the illustrated steps may be omitted, other steps may be included, the order of steps may be different, or some steps may be performed simultaneously. Furthermore, a flow chart described as a series of actions may be divided into several parts and executed, or multiple flow charts may be executed sequentially or in conjunction with each other. In other embodiments, some of the illustrated components may be omitted, other components may be included, or the arrangement of components may be changed. Furthermore, the functions implemented by these steps and components may be implemented in hardware, software, or a combination of hardware and software.

[0107] The systems of the present disclosure may be implemented as a single device or may be distributed across multiple devices or subsystems that cooperate to execute a program. Examples of processing systems include general purpose central processing units, application specific processors, and logic devices, as well as any other type of processing device, combination or variation thereof.

[0108] The methods disclosed herein are embodied in software including various programs and data, and are performed by a computing system (information processing system) that executes the software. The computing system includes a non-transitory computer-readable medium that stores various instructions included in the programs, and one or more processors that execute these software instructions. The computing system may be implemented as a standalone computer, or may be implemented using a client-server architecture in which software is distributed and executed by multiple computers.

[0109] Memory is a computer-readable storage medium, including non-transitory computer-readable medium, such as, but not limited to, ROM, hard disk, storage, removable media, flash memory, memory stick, optical media, magneto-optical media, and CD-ROM.

[0110] The at least one processor may include, but is not limited to, a central processing unit (CPU), a graphics processing unit (GPU), an accelerated processing unit (APU), a neural network processing unit (NPU), a microprocessor, a microcontroller, a digital signal processor (DSP), a field programmable gate array (FPGA), a complex programmable logic device (CPLD), an application-specific integrated circuit (ASIC), other processors including general-purpose processors, or any combination thereof designed to perform the functions described herein.

[0111] Communication between the devices or systems may be performed according to well-known communication protocols over one or more communication networks, which may be, for example, but not limited to, an intranet, the Internet, a local area network, a wide area network, a wireless network, a wired network, a virtual network, a software-defined network, or any other type of network or combination thereof.

[0112] The communication IF (interface) realizes a function that allows one device to communicate with other devices via a communication network. The communication IF may be, for example, but is not limited to, a LAN (Local Area Network), Wi-Fi (registered trademark), Bluetooth (registered trademark), NFC (Near Field Communication), or other wireless communication interface.

[0113] The following are some aspects that can be understood based on the above-described embodiment and modifications. [1] At least one processor acquiring a plurality of product data sets, each of the plurality of product data sets including product descriptions for a plurality of products, the product descriptions including descriptions of ingredients contained in the corresponding product; obtaining one or more keywords, each of the one or more keywords indicating one or more designated components; extracting one or more suspect products from the plurality of products by performing a keyword search using the one or more keywords on the plurality of product data sets, wherein each of the suspect products includes at least one of the one or more specified ingredients in the corresponding product description; Detecting one or more non-containing products that do not contain the one or more specified ingredients from among the one or more suspect products based on the product descriptions corresponding to each of the one or more suspect products; By excluding the one or more non-containing products from the one or more suspected products, 1 or more outputting a list of products containing at least one of the specified ingredients; A method for detecting a designated product, including:

[0114] [2] Detecting the one or more non-containing products includes, for each of the one or more suspect products: Detecting a negative description from the product description that explains an ingredient that the corresponding product does not contain; detecting one or more of the non-containing components set forth in the negative description; If all of the one or more specified ingredients included in the product description are the non-containing ingredients included only in the negative description, determining that the corresponding suspect product is the non-containing product; The designated commodity detection method according to [1] above,

[0115] [3] Detecting the one or more non-containing products includes, for each of the one or more suspect products: Detecting a negative description from the product description that explains an ingredient that the corresponding product does not contain; Detecting a positive description from the product description that explains the ingredients contained in the corresponding product; detecting one or more of the non-containing components set forth in the negative description; detecting one or more of the ingredients set forth in the affirmative statement; If all of the one or more specified ingredients included in the product description are the non-containing ingredients included in the negative description, and the one or more specified ingredients included in the product description are not included as the containing ingredients in the positive description, determine that the corresponding suspect product is the non-containing product; Including, The designated product detection method described in [1] above.

[0116] [4] displaying a confirmation screen on a display; the confirmation screen is configured to display the product description of one product data set among the plurality of product data sets; In the product description on the confirmation screen, the one or more non-containing ingredients and the one or more containing ingredients are highlighted with marks that are different from each other. A designated commodity detection method according to any one of [1] to [3] above.

[0117] [5] At least one processor acquiring a plurality of product data sets, each of the plurality of product data sets including product descriptions for a plurality of products, the product descriptions including descriptions of ingredients contained in the corresponding product; obtaining one or more keywords, each of the one or more keywords indicating one or more designated components; extracting one or more suspect products from the plurality of products by performing a keyword search using the one or more keywords on the plurality of product data sets, wherein each of the suspect products includes at least one of the one or more specified ingredients in the corresponding product description; Detecting a negative description that explains non-containing ingredients that are not contained in the corresponding suspect product from the product description of each of the one or more suspect products; detecting one or more of the non-containing components set forth in each negative statement; Displaying a confirmation screen on a display for determining whether each of the suspected products contains the one or more specified components, the confirmation screen being configured to display the product description of the corresponding suspected product, and the one or more non-containing components being highlighted in the product description on the confirmation screen; A method for detecting a designated product, including:

[0118] [6] Detecting the one or more non-containing products includes, for each of the one or more suspect products: performing named entity extraction on the product description to detect one or more ingredient names from the product description; extracting the one or more designated components from the one or more component names; Including, A designated commodity detection method according to any one of [1] to [5] above.

[0119] [7] Each of the plurality of products is classified into one of a plurality of product categories, The one or more designated ingredients are designated for one product category among the plurality of product categories, The one or more designated ingredients are illegal ingredients that are illegal in the one product category, A designated commodity detection method according to any one of [1] to [6] above.

[0120] [8] Each of the plurality of products is classified into one of a plurality of product categories, The plurality of product categories include at least one of health foods, pharmaceuticals, quasi-drugs, and cosmetics, The one or more designated ingredients are set for each product category, A designated commodity detection method according to any one of [1] to [7] above.

[0121] [9] When one of the plurality of product data sets includes one or more images, and when at least a part of the one or more images includes a character string, extracting the character string from the one or more images; The product description of the one product data set includes an extracted character string, which is the character string extracted from the one or more images. A designated commodity detection method according to any one of [1] to [8] above.

[0122]

[10] Each of the plurality of products is classified into one of a plurality of product categories, The one or more designated ingredients are designated for one product category among the plurality of product categories, classifying the plurality of products into three groups based on the product descriptions corresponding to each of the plurality of products, wherein the three groups include: A group of target products belonging to the product category 1; A group of non-target products belonging to a product category other than the product category 1; A group of set products including the target product, A designated commodity detection method according to any one of [1] to [9] above.

[0123]

[11] Classifying them into the three groups is inputting a product description for a product into a multi-class classification model, the multi-class classification model being configured to classify the product into one of the three groups based on the product description; obtaining a classification result for the one commodity from the multi-class classification model; A designated commodity detection method according to any one of [1] to

[10] above.

[0124]

[12] generating the multi-class classification model by training it with training data; generating a plurality of ground truth data sets for use as the training data; Including, the learning data includes a plurality of training product data sets; each training product data set includes training product descriptions for a corresponding product; The training product description includes a description of the ingredients of the corresponding product, Each of the correct answer datasets is the training product description of the corresponding product; a correct answer label indicating whether the corresponding product is the target product, the non-target product, or the set product; Generating the plurality of ground truth datasets includes: inputting a prompt into a language model; obtaining an output result from the language model in response to the prompt; Including, The prompt may include: a plurality of training product descriptions for each of the plurality of products; an instruction sentence for causing the language model to output an answer as to whether the plurality of products described by each of the plurality of training product descriptions corresponds to the target product, the non-target product, or the set product; A designated commodity detection method according to any one of [1] to

[11] above.

[0125]

[13] at least one memory storing computer program code; at least one processor; wherein the at least one processor executes the computer program code to acquiring a plurality of product data sets, each of the plurality of product data sets including product descriptions for a plurality of products, the product descriptions including descriptions of ingredients contained in the corresponding product; obtaining one or more keywords, each of the one or more keywords indicating one or more designated components; extracting one or more suspect products from the plurality of products by performing a keyword search using the one or more keywords on the plurality of product data sets, wherein each of the suspect products includes at least one of the one or more specified ingredients in the corresponding product description; Detecting one or more non-containing products that do not contain the one or more specified ingredients from among the one or more suspect products based on the product descriptions corresponding to each of the one or more suspect products; By excluding the one or more non-containing products from the one or more suspected products, 1 or more outputting a list of products containing at least one of the specified ingredients; A designated product detection system configured to:

[0126]

[14] at least one memory storing computer program code; at least one processor; wherein the at least one processor executes the computer program code to acquiring a plurality of product data sets, each of the plurality of product data sets including product descriptions for a plurality of products, the product descriptions including descriptions of ingredients contained in the corresponding product; obtaining one or more keywords, each of the one or more keywords indicating one or more designated components; extracting one or more suspect products from the plurality of products by performing a keyword search using the one or more keywords on the plurality of product data sets, wherein each of the suspect products includes at least one of the one or more specified ingredients in the corresponding product description; Detecting a negative description that explains non-containing ingredients that are not contained in the corresponding suspect product from the product description of each of the one or more suspect products; detecting one or more of the non-containing components set forth in each negative statement; Displaying a confirmation screen on a display for determining whether each of the suspected products contains the one or more specified components, the confirmation screen being configured to display the product description of the corresponding suspected product, and the one or more non-containing components being highlighted in the product description on the confirmation screen; A designated product detection system configured to:

[0127]

[15] At least one processor has acquiring a plurality of product data sets corresponding to a plurality of products, each of the product data sets including a product description of ingredients contained in the corresponding product; obtaining one or more keywords, each of the one or more keywords indicating one or more designated components; A keyword search is performed on the plurality of product data sets using the one or more keywords to extract one or more suspect products from the plurality of products, and each of the suspect products contains the following in the corresponding product description: 1 or more and containing at least one of the specified ingredients: From among the one or more suspected products, based on the product descriptions corresponding to each of the one or more suspected products, 1 or more Detecting one or more non-containing products that do not contain the specified ingredient; By excluding the one or more non-containing products from the one or more suspected products, 1 or more outputting a list of products containing at least one of the specified ingredients; A program to execute.

[0128]

[16] At least one processor has acquiring a plurality of product data sets, each of the plurality of product data sets including product descriptions for a plurality of products, the product descriptions including descriptions of ingredients contained in the corresponding product; obtaining one or more keywords, each of the one or more keywords indicating one or more designated components; extracting one or more suspect products from the plurality of products by performing a keyword search using the one or more keywords on the plurality of product data sets, wherein each of the suspect products includes at least one of the one or more specified ingredients in the corresponding product description; Detecting a negative description that explains non-containing ingredients that are not contained in the corresponding suspect product from the product description of each of the one or more suspect products; detecting one or more of the non-containing components set forth in each negative statement; Displaying a confirmation screen on a display for determining whether each of the suspected products contains the one or more specified components, the confirmation screen being configured to display the product description of the corresponding suspected product, and the one or more non-containing components being highlighted in the product description on the confirmation screen; A program to execute. [Explanation of symbols]

[0129] 11...designated product detection system, 13...database, 14...web server, 15...purchaser terminal, 16...seller terminal, 17...language model, 20...designated product detection device, 21,31,41...processor, 22,32,42...memory, 23,33,43...communication IF, 24,34...program, 25...detection data, 30...learning device, 35...learning data, 40...worker terminal, 46...display, 50...sales page, 51...tagline, 52...title, 53...detailed description, 54...image, 60...confirmation screen, 61,62,63,64...character string, 66...negative description, 67...positive description.

Claims

1. At least one processor acquiring a plurality of product data sets, each of the plurality of product data sets including product descriptions for a plurality of products, the product descriptions including descriptions of ingredients contained in the corresponding product; obtaining one or more keywords, each of the one or more keywords indicating one or more specified components; extracting one or more suspect products from the plurality of products by performing a keyword search using the one or more keywords on the plurality of product data sets, wherein each of the suspect products includes at least one of the one or more specified ingredients in the corresponding product description; Detecting one or more non-containing products that do not contain the one or more specified components from among the one or more suspect products based on the product descriptions corresponding to each of the one or more suspect products; outputting a list of products containing at least one of the one or more specified ingredients by excluding the one or more non-containing products from the one or more suspect products; A method for detecting a designated product, including:

2. Detecting the one or more non-containing products includes, for each of the one or more suspect products: Detecting a negative description from the product description that explains an ingredient that the corresponding product does not contain; detecting one or more of the non-containing components set forth in the negative description; If all of the one or more specified ingredients included in the product description are the non-containing ingredients included only in the negative description, determining that the corresponding suspect product is the non-containing product; The designated commodity detection method according to claim 1 , comprising:

3. Detecting the one or more non-containing products includes, for each of the one or more suspect products: Detecting a negative description from the product description that explains an ingredient that the corresponding product does not contain; Detecting a positive description from the product description that explains the ingredients contained in the corresponding product; detecting one or more of the non-containing components set forth in the negative description; detecting one or more of the ingredients set forth in the affirmative description; If all of the one or more specified ingredients included in the product description are the non-containing ingredients included in the negative description, and the one or more specified ingredients included in the product description are not included as the containing ingredients in the positive description, determine that the corresponding suspect product is the non-containing product; The designated commodity detection method according to claim 1 , comprising:

4. displaying a confirmation screen on a display; the confirmation screen is configured to display the product description of one product data set among the plurality of product data sets; In the product description on the confirmation screen, the one or more non-containing ingredients and the one or more containing ingredients are highlighted with marks that are different from each other. The designated commodity detection method according to claim 3 .

5. At least one processor acquiring a plurality of product data sets, each of the plurality of product data sets including product descriptions for a plurality of products, the product descriptions including descriptions of ingredients contained in the corresponding product; obtaining one or more keywords, each of the one or more keywords indicating one or more specified components; extracting one or more suspect products from the plurality of products by performing a keyword search using the one or more keywords on the plurality of product data sets, wherein each of the suspect products includes at least one of the one or more specified ingredients in the corresponding product description; Detecting a negative description that explains non-containing ingredients that are not contained in the corresponding suspect product from the product description of each of the one or more suspect products; detecting one or more of the non-containing components described in each negative description; Displaying a confirmation screen on a display for determining whether each of the suspected products contains the one or more specified components, the confirmation screen being configured to display the product description of the corresponding suspected product, and the one or more non-containing components being highlighted in the product description on the confirmation screen; A method for detecting a designated product, including:

6. Detecting the one or more non-containing products includes, for each of the one or more suspect products: performing named entity extraction on the product description to detect one or more ingredient names from the product description; extracting the one or more designated components from the one or more component names; Including, The designated commodity detection method according to any one of claims 1 to 4.

7. Each of the plurality of products is classified into one of a plurality of product categories, the one or more designated ingredients are designated for one product category among the plurality of product categories, The one or more designated ingredients are illegal ingredients that are illegal in the one product category. The designated commodity detection method according to any one of claims 1 to 5.

8. Each of the plurality of products is classified into one of a plurality of product categories, the plurality of product categories include at least one of health foods, pharmaceuticals, quasi-drugs, and cosmetics; The one or more designated ingredients are set for each product category. The designated commodity detection method according to any one of claims 1 to 5.

9. When one of the plurality of product data sets includes one or more images, and when at least a part of the one or more images includes a character string, extracting the character string from the one or more images; The product description of the one product data set includes an extracted character string, which is the character string extracted from the one or more images. The designated commodity detection method according to any one of claims 1 to 5.

10. Each of the plurality of products is classified into one of a plurality of product categories, the one or more designated ingredients are designated for one product category among the plurality of product categories, classifying the plurality of products into three groups based on the product descriptions corresponding to each of the plurality of products, wherein the three groups are A group of target products belonging to the one product category; a group of non-target products belonging to a product category other than the first product category; A group of set products including the target product, The designated commodity detection method according to any one of claims 1 to 5.

11. Classifying into the three groups is inputting a product description for a product into a multi-class classification model, the multi-class classification model being configured to classify the product into one of the three groups based on the product description; obtaining a classification result for the one commodity from the multi-class classification model; The designated commodity detection method according to claim 10.

12. generating the multi-class classification model by training it with training data; generating a plurality of ground truth data sets for use as the training data; Including, the learning data includes a plurality of training product data sets; each training product data set includes training product descriptions for a corresponding product; The training product description includes a description of the ingredients of the corresponding product, Each of the correct answer datasets is the training product description of the corresponding product; a correct answer label indicating whether the corresponding product is the target product, the non-target product, or the set product; Generating the plurality of ground truth datasets includes: inputting a prompt into a language model; obtaining an output result from the language model in response to the prompt; Including, The prompt may include: a plurality of training product descriptions for each of the plurality of products; an instruction sentence for causing the language model to output an answer as to whether the plurality of products described by each of the plurality of training product descriptions corresponds to the target product, the non-target product, or the set product; The designated commodity detection method according to claim 11.

13. at least one memory storing computer program code; at least one processor; wherein the at least one processor executes the computer program code to acquiring a plurality of product data sets, each of the plurality of product data sets including product descriptions for a plurality of products, the product descriptions including descriptions of ingredients contained in the corresponding product; obtaining one or more keywords, each of the one or more keywords indicating one or more specified components; extracting one or more suspect products from the plurality of products by performing a keyword search using the one or more keywords on the plurality of product data sets, wherein each of the suspect products includes at least one of the one or more specified ingredients in the corresponding product description; Detecting one or more non-containing products that do not contain the one or more specified components from among the one or more suspect products based on the product descriptions corresponding to each of the one or more suspect products; outputting a list of products containing at least one of the one or more specified ingredients by excluding the one or more non-containing products from the one or more suspect products; A designated product detection system configured to:

14. at least one memory storing computer program code; at least one processor; wherein the at least one processor executes the computer program code to acquiring a product dataset, the product dataset including product descriptions for products, the product descriptions including ingredient descriptions for the products; obtaining one or more keywords, each of the one or more keywords indicating one or more specified components; determining whether the product is a suspect product by performing a keyword search using the one or more keywords on the product description, wherein the suspect product contains at least one of the one or more specified ingredients in the product description; Detecting a negative description explaining non-containing ingredients that the suspected product does not contain from the product description of the suspected product; detecting one or more of the non-containing components set forth in the negative description; Displaying a confirmation screen on a display for determining whether the suspected product contains the one or more specified components, the confirmation screen being configured to display the product description of the suspected product, and the one or more non-containing components being highlighted in the product description on the confirmation screen; A designated product detection system configured to:

15. At least one processor acquiring a plurality of product data sets corresponding to a plurality of products, each of the product data sets including a product description of ingredients contained in the corresponding product; obtaining one or more keywords, each of the one or more keywords indicating one or more specified components; extracting one or more suspect products from the plurality of products by performing a keyword search using the one or more keywords on the plurality of product data sets, wherein each of the suspect products includes at least one of the one or more specified ingredients in the corresponding product description; Detecting one or more non-containing products that do not contain the one or more specified ingredients from among the one or more suspect products based on the product descriptions corresponding to each of the one or more suspect products; outputting a list of products containing at least one of the one or more specified ingredients by excluding the one or more non-containing products from the one or more suspect products; A program to execute.

16. At least one processor acquiring a product dataset, the product dataset including product descriptions for products, the product descriptions including ingredient descriptions for the products; obtaining one or more keywords, each of the one or more keywords indicating one or more specified components; determining whether the product is a suspect product by performing a keyword search using the one or more keywords on the product description, wherein the suspect product contains at least one of the one or more specified ingredients in the product description; Detecting a negative description explaining non-containing ingredients that the suspected product does not contain from the product description of the suspected product; detecting one or more of the non-containing components set forth in the negative description; Displaying a confirmation screen on a display for determining whether the suspected product contains the one or more specified components, the confirmation screen being configured to display the product description of the suspected product, and the one or more non-containing components being highlighted in the product description on the confirmation screen; A program to execute.

Images