KR102823051B1 - Server and method for automatic collection and stylization processing of unstructured product data - Google Patents

Abstract

One aspect of the present invention provides a server and method for automatically collecting and stylizing unstructured product data. The server for automatically collecting and stylizing unstructured product data may include at least one processor and a memory storing instructions for instructing the at least one processor to perform a plurality of operations. The plurality of operations may include an operation for receiving product information from an external server, an operation for receiving template information from an external server, an operation for extracting metadata of the received product information and template information, an operation for storing the extracted metadata in a CSV format and specifying a product category from the metadata extracted from the product information, and an operation for allocating the product information to a predetermined product page layout template based on the metadata extracted from the product information based on the specified product category.

Description

{SERVER AND METHOD FOR AUTOMATIC COLLECTION AND STYLIZATION PROCESSING OF UNSTRUCTURED PRODUCT DATA}

The present invention relates to a server and method for automatically collecting and stylizing non-standard product data, and more specifically, to a server and method for automatically collecting and stylizing non-standard product data that collects data of products being sold, stylizes the data according to a specific pattern, and reprocesses the data into a specific format.

Recently, product information on e-commerce platforms mainly exists in the form of unstructured data. This data is provided in various forms such as images, videos, and fragmented texts, and may need to be converted into a structured form to understand the characteristics and information of the product.

However, existing data processing methods have many manual elements, are time-consuming, and are prone to data errors, resulting in low efficiency.

Effectively leveraging product data requires automated tools, which may require using OCR technology to extract text from images and AI technologies such as LLM to analyze and reconstruct it.

Such technologies can enable fast and accurate processing of product data and improve user experience by providing optimized data formats for each product category.

Currently, most product data processing methods are manual, and automation technology is also applied to a limited extent. In particular, there is a lack of specialized data processing methods for each product category, which results in problems such as reduced accuracy and availability of product information. In addition, existing systems do not effectively manage errors that frequently occur during the data reprocessing process, and there are also limitations in stylization processing of data.

To solve these problems, research is needed to automatically collect unstructured product data, analyze and process it, and efficiently convert it into metadata.

Domestic registration patent No. 10-2225729

The purpose of the present invention to solve the above problems is to provide a server and method for automatically collecting and stylizing non-standard product data.

One aspect of the present invention to achieve the above object provides a server and method for automatically collecting and stylizing non-standard product data.

A product non-standard data automatic collection and stylization processing server may include at least one processor and a memory storing instructions that instruct the at least one processor to perform a plurality of operations.

The plurality of operations may include an operation of receiving product information from an external server, an operation of receiving template information from an external server, an operation of extracting metadata of the received product information and template information, an operation of saving the extracted metadata in CSV format, an operation of specifying a product category from the metadata extracted from the product information, and an operation of assigning the product information to a predefined product page layout template based on the metadata extracted from the product information based on the specified product category.

An operation of calculating a product basic information score and a product advanced information score by combining product information, and calculating a product information score from a weighted sum of the calculated product basic information score and product advanced information score, an operation of calculating a template basic structure score and a template function score by combining template information, and calculating a template information score from a weighted sum of the calculated template basic structure score and template function score, an operation of calculating a style optimization score from a weighted sum of the calculated product information score and template information score, and calculating a template complexity adjustment coefficient ( ) and may further include an operation of producing a maximum template item from the product of the base template items.

The operation of producing the maximum template items is to adjust the template complexity by averaging the color diversity index, font diversity index, image diversity index, and layout diversity index of each template from the template information. ) may further include actions that produce a result.

When using the server and method for automatically collecting and stylizing non-standard product data according to the present invention as described above, the accuracy of product information can be increased, efficiency can be improved through automation of the data processing process, and user experience can be optimized by providing customized data formats by category.

The effects obtainable from the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art to which the present invention belongs from the description below.

The above and other aspects, features and advantages of certain preferred embodiments of the present invention will become more apparent from the following description taken in conjunction with the accompanying drawings.
Figure 1 is an exemplary diagram showing the operating environment of a server for automatically collecting and stylizing non-standard product data according to one embodiment of the present invention.
Figure 2 is an example diagram explaining automatic collection and stylization processing of non-standard product data.
Figure 3 is an example diagram illustrating an example of a server for automatically collecting and stylizing non-standard product data.
Figure 4 is a hardware configuration diagram for a server for automatically collecting and stylizing non-standard product data according to Figure 1.
It should be noted that throughout the above drawings, like reference numerals are used to illustrate identical or similar elements, features and structures.

Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

In describing the embodiments, descriptions of technical contents that are well known in the technical field to which the present invention belongs and are not directly related to the present invention will be omitted. This is to convey the gist of the present invention more clearly without obscuring it by omitting unnecessary descriptions.

For the same reason, some components in the attached drawings are exaggerated, omitted, or schematically illustrated. In addition, the size of each component does not entirely reflect the actual size. The same or corresponding components in each drawing are given the same reference numbers.

The advantages and features of the present invention, and the methods for achieving them, will become clear with reference to the embodiments described in detail below together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and the present embodiments are provided only to make the disclosure of the present invention complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined only by the scope of the claims. Like reference numerals refer to like elements throughout the specification.

At this time, it will be understood that each block of the processing flow diagrams and combinations of the flow diagrams can be performed by computer program instructions. These computer program instructions can be loaded onto a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing equipment, so that the instructions executed by the processor of the computer or other programmable data processing equipment create a means for performing the functions described in the flow diagram block(s). These computer program instructions can also be stored in a computer-available or computer-readable memory that can be directed to a computer or other programmable data processing equipment to implement the functions in a specific manner, so that the instructions stored in the computer-available or computer-readable memory can also produce an article of manufacture that includes an instruction means for performing the functions described in the flow diagram block(s). Since the computer program instructions may be installed on a computer or other programmable data processing apparatus, a series of operational steps may be performed on the computer or other programmable data processing apparatus to produce a computer-executable process, so that the instructions executing the computer or other programmable data processing apparatus may also provide steps for executing the functions described in the flowchart block(s).

Additionally, each block may represent a module, segment, or portion of code that contains one or more executable instructions for performing a particular logical function(s). It should also be noted that in some alternative implementation examples, the functions mentioned in the blocks may occur out of order. For example, two blocks shown in succession may in fact be performed substantially concurrently, or the blocks may sometimes be performed in reverse order, depending on the functionality they perform.

Here, the term '~ part' used in the present embodiment means software or hardware components such as FPGA (field-programmable gate array) or ASIC (application specific integrated circuit), and the '~ part' performs certain roles. However, the '~ part' is not limited to software or hardware. The '~ part' may be configured to be in an addressable storage medium and may be configured to reproduce one or more processors. Accordingly, as an example, the '~ part' includes components such as software components, object-oriented software components, class components, and task components, and processes, functions, properties, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functions provided in the components and '~ parts' may be combined into a smaller number of components and '~ parts' or further separated into additional components and '~ parts'. Additionally, the components and '~parts' may be implemented to regenerate one or more CPUs within the device or secure multimedia card.

In specifically describing embodiments of the present invention, examples of specific systems will be mainly targeted, but the main point claimed in this specification can be applied to other communication systems and services having similar technical backgrounds without significantly departing from the scope disclosed in this specification, and this can be done at the discretion of a person skilled in the relevant technical field.

Figure 1 is an exemplary diagram showing the operating environment of a server for automatically collecting and stylizing non-standard product data according to one embodiment of the present invention.

Referring to FIG. 1, the product non-standard data automatic collection and stylization processing server (100) (hereinafter referred to as 'server (100)') can receive at least one piece of product information through a user terminal (10). At this time, the user terminal (10) is a terminal of a user who wishes to use the product non-standard data automatic collection and stylization processing server, and may be a user who wishes to register the collected product information in at least one online shopping mall.

Here, the user terminal (10) may include a communicable desktop computer, a laptop computer, a notebook, a smart phone, a tablet PC, a mobile phone, a smart watch, a smart glass, an e-book reader, a portable multimedia player (PMP), a portable game console, a navigation device, a digital camera, a digital multimedia broadcasting (DMB) player, a digital audio recorder, a digital audio player, a digital video recorder, a digital video player, a PDA (Personal Digital Assistant), etc.

In addition, the server (100) may be connected to an external big data server (110) via a wired or wireless network to receive template information. Here, the template means a pre-created product page layout template, and may be a product registration page layout template provided on the product registration page of each online shopping mall to register a product in the online shopping mall.

Here, product information may include price change index, inventory liquidity ratio, color diversity score, size option index, origin diversity index, manufacturer diversity index, sales channel diversification index, inventory turnover rate, brand awareness score, eco-friendly certification rate, user feedback response time, warranty extension rate, consumer satisfaction score, delivery efficiency index, return policy score, and patented product ratio.

Here, the template information may include template matching accuracy, data tagging automation ratio, template flexibility index, multimodal integration degree, template adjustment response time, product attribute extraction efficiency, data standardization index, template update frequency, automatic classification performance, dynamic template adjustment function, template-based content creation speed, data security index, user experience optimization score, artificial intelligence integration efficiency, API linkage index, template-based reporting efficiency, content creation automation index, metadata formatting efficiency, product comparison analysis automation degree, and interactive visualization effectiveness.

Here, the wired/wireless network may include various communication methods such as, but is not limited to, Bluetooth communication, BLE (Bluetooth Low Energy) communication, Near Field Communication (NFC), WLAN communication, Zigbee communication, Infrared Data Association (IrDA) communication, WFD (Wi-Fi Direct) communication, UWB (ultra-wideband) communication, Ant+ communication, WIFI communication, and RFID (Radio Frequency Identification) communication.

Meanwhile, when the server (100) receives product information, it can classify the product information into at least one category based on at least one attribute.

Next, the server (100) can calculate a product information score by cross-linking product information classified in each category.

When the server (100) receives template information, it can classify the template information into at least one category based on at least one attribute.

Next, the server (100) can combine template information classified into each category to calculate a template information score.

Meanwhile, the server (100) can calculate a style optimization score by combining the calculated product information score and template information score, and can calculate a maximum template item using the calculated style optimization score.

Through this, product information can be reprocessed and provided even in online shopping malls that provide different product page layouts, thereby increasing the convenience of product management for users.

Figure 2 is an example diagram illustrating an example of a server for automatically collecting and stylizing non-standard product data.

Referring to FIGS. 1 and 2, the server (100) can collect product data, map it to different layout templates according to a specific pattern, and allocate product information to generate a product page. For example, information about a product can be obtained from an original page (210) as non-standard data, and the original page (210) can include a product image (211) and at least one piece of product information (212-1, 212-2, 212-3).

When the server (100) receives product information from the original page (210), it can reprocess the information by assigning each piece of information to at least one pre-generated product page layout template (220).

For example, a product page layout template (220) may include a product image area (221) and at least one product information area (221-1, 222-2).

The server (100) can obtain template information from the product page layout template (220) and reprocess it by assigning product information to the product image area (221) and at least one product information area (221-1, 222-2) of the product page layout template (220).

For example, the server (100) can obtain original information of a product. The original information can include non-standard data such as an image of a product detail page or text of a website. This information is converted into product metadata in text form by utilizing OCR (optical character recognition) technology and LLM (language modeling) technology, and through this, the server (100) can collect and process data.

At this time, the server (100) can extract and organize the metadata of the product. The metadata of the product may include the product's name, specifications, price, brand, manufacturer, country of manufacture, usage, features, components, materials, etc., and the server (100) can collect this data and organize it in CSV format and store it. This process can be stored in a database or other storage medium and used for subsequent processing.

In addition, the server (100) can identify the product category among the collected data. The format of the metadata can be modified and provided according to each product category. For example, in the case of the 'sunscreen' category, the server (100) can delete the standard data and emphasize the usage and features. In addition, the data set is reprocessed by necessarily including the UV protection rate and full ingredient information. The data set reprocessed in this way is finally provided to the user and can be used for product sales or promotion, and can increase the efficiency of product sales and registration.

For example, when the data target is a 'men's sunscreen' product posted on an online shopping mall, the server (100) collects data such as the product's name, price, brand, manufacturer, and product category from the online shopping mall website. The server (100) converts a detailed page image file into text using OCR technology and analyzes it using LLM technology to additionally collect data such as the product's strengths, features, and directions for use. The data collected in this way is organized in CSV format, and the server (100) can delete size and weight data and emphasize directions for use and UV protection rate information according to the data processing rules of the 'sunscreen' category to generate and provide a final product dataset.

Through this, the server (100) can efficiently manage data on products sold on various online sales platforms and provide optimized information to consumers, thereby contributing to increasing the marketability of products and improving user experience.

Figure 3 is an example diagram showing the operating environment of a server for automatic collection and stylization processing of non-standard product data.

Referring to FIG. 3, the product non-standard data automatic collection and stylization processing server (hereinafter referred to as 'server (300)') may include a product information analysis unit (310), a template information analysis unit (320), and a style optimization unit (330).

The product information analysis unit (310) can receive product information from an external server based on user input product information entered by the user.

Here, product information may include price change index, inventory liquidity ratio, color diversity score, size option index, origin diversity index, manufacturer diversity index, sales channel diversification index, inventory turnover rate, brand awareness score, eco-friendly certification rate, user feedback response time, warranty extension rate, consumer satisfaction score, delivery efficiency index, return policy score, and patented product ratio.

When receiving product information, the product information analysis unit (310) can classify the received product information into at least one category based on at least one attribute.

For example, the product information analysis unit (310) may classify price fluctuation index, inventory liquidity ratio, color diversity score, size option index, country of origin diversity index, manufacturer diversity index, sales channel diversification index, inventory turnover rate, and brand awareness score as basic product information, and classify eco-friendly certification ratio, user feedback response time, warranty period extension rate, consumer satisfaction score, delivery efficiency index, return policy score, and patented product ratio as advanced product information.

The product information analysis unit (310) manages product information classified into basic product information and advanced product information by classifying them into each category, thereby enabling real-time monitoring of product information and early identification and management of any deviations from the management pattern. In addition, the convenience of information management can be increased by managing the index of information classified into each category.

Meanwhile, the categorization of product information and the calculation of product information scores can use artificial intelligence-based machine learning or deep learning models.

For example, artificial intelligence learning models including supervised learning, unsupervised learning, reinforcement learning, convolutional neural networks (CNNs), recurrent neural networks (RNNs), long short-term memory networks (LSTMs), natural language processing (NLP), decision support systems (DSS), and predictive analytics can be used.

In addition, the product information analysis unit (310) can monitor information classified into each category in real time and generate a notification (e.g., sound, vibration, light, etc.) for warning when a pre-determined deviation range (e.g., 20%) is exceeded.

Next, the product information analysis unit (310) can calculate a product information score by mutually combining product information classified in each category. Meanwhile, unless otherwise explained, the weights w1 to wn included in the mathematical formulas described below may be predetermined weight values whose sum may be 1.

For example, the product information analysis unit (310) can calculate the product basic information score as shown in the mathematical formula below.

Here, the inventory liquidity ratio is an indicator that determines the balance between supply and demand in the market for a product, and high liquidity can mean that the product is turning over quickly and responding well to demand.

For example, if Product A sells 150 units out of 300 inventory during a given period (e.g., 1 month) and Product B sells 250 units out of 500 inventory during a given period (e.g., 1 month), the inventory liquidity ratio can be calculated by dividing the number of units sold during the given period by the beginning inventory. This ratio is used as an indicator of how quickly the products are selling.

A similar result would be that both products A and B have a stock liquidity ratio of 0.5. This means that only half of the starting inventory of each product has been sold, which could indicate that the inventory turnover rates are the same.

The price fluctuation index can analyze price fluctuation trends and determine optimal price positioning. The price fluctuation index is an indicator that measures the price volatility of a product, quantifies the price responsiveness of a product to market changes, and can be calculated using price data at various points in time, and is calculated by considering both the range and frequency of price fluctuations. For example, assume that the price fluctuations at points 1 to 5 within a predetermined period (e.g., 30 days) are [100, 120, 110, 130, 125] thousand won.

That is, the standard deviation for the same product can represent the degree of price fluctuation, and as an example, the price fluctuation index can be normalized to 0 to 1 through min-max normalization within a range of predetermined minimum and maximum values. For example, if the minimum price fluctuation index is 0, the maximum price fluctuation index is 0.5, and the actual price fluctuation index is 0.09, the price fluctuation index can be normalized to 0.82 (=1-0.18).

The sales channel diversity index is an indicator that quantifies how many different sales channels a product is sold through, indicating the product's market accessibility and consumer reach. The higher the diversity of sales channels, the higher the possibility of reaching a wider market and a wider range of consumers.

For example, assuming that the same product A maintains 7 sales channels for a total of 4 sales channel types: 3 online malls, 2 offline stores, 1 social media, and 1 direct sales, the sales ratio for each channel is calculated as 50% online malls, 30% offline stores, 10% social media, and 10% direct sales.

At this time, the entropy of each sales ratio is calculated as follows.

Here, sales ratio_i can mean the sales ratio of each sales channel (i). The higher the entropy value of the total sales ratio, the more the product is sold through various sales channels, which can mean that the product has a wide market accessibility and consumer reach.

For example, the sales channel diversification index can be normalized to 0 to 1 through min-max normalization within the range of minimum and maximum values of the predetermined total sales ratio entropy. For example, if the minimum total sales ratio entropy is 0, the maximum total sales ratio entropy is 3, and the actual total sales ratio entropy is 1.32, the sales channel diversification index can be normalized to 0.44.

The product basic information score is an indicator used in platforms that collect and reprocess product data, and can be used as a reference indicator to optimize product information according to the product sales characteristic pattern and provide it to users, and to determine the optimization of templates that reprocess product information. In other words, a high product basic information score can mean that detailed information about the product is sufficient, and if the product basic information score is low and information is limited, the template can be simplified to emphasize only important information to users.

The product information analysis unit (310) can calculate the product advanced information score as shown in the mathematical formula below.

Here, the delivery efficiency index can be an indicator that reflects delivery efficiency and evaluates how quickly and accurately a product reaches a customer.

For example, if the total number of orders for product A is 1,000 during a predetermined period (e.g., 1 month), the number of on-time deliveries is 900, the maximum delivery time is 48 hours, and the average delivery delay time is 2 hours, the on-time delivery rate can be calculated as 0.9 (= 900/1000), the delivery delay rate can be calculated as 0.0417 (= 2/48), and the delivery efficiency index is calculated as 0.8625.

The return policy score can be a factor that promotes customer satisfaction and repurchase intention by reflecting the effectiveness and customer friendliness of the return policy.

For example, if Product K has a predefined return acceptance period of 30 days, an actual customer return rate of 0.95 (e.g., 95 out of 100 requests were approved) during the same period, a customer satisfaction score of 4.5/5, an average return processing time of 3 days, and a maximum return processing time of 7 days, the return policy score would be calculated as follows:

The patented product ratio can be an indicator of the uniqueness and technological superiority of a product in the market, reflecting the product's innovation and originality. For example, when the total number of products of a product seller selling at least one product is 200 and the number of products with patents is 50, the patented product ratio can be calculated as 0.25.

The product premium information score produced is an indicator of the quality of the product, and can be an indicator of how the product is perceived in the market. For example, if the product premium information score is high, it means that the product meets high quality, efficiency, and environmental responsibility in the market, and it indicates that the product has high consumer satisfaction and strong market competitiveness. If the product premium information score is low, it may mean that the product quality or delivery efficiency does not meet market expectations, or that there is a lack of eco-friendly certification, which indicates areas for improvement, and it may indicate that management and improvement are needed in certain areas.

The product information analysis unit (310) can calculate a product information score based on the calculated product basic information score and product advanced information score as shown in the following mathematical formula. Here, w1 to w2 are predetermined weights whose sum can be 1, and the calculated product basic information score and product advanced information score can be normalized values from 0 to 1 based on predetermined minimum and maximum values.

The product information score is generated by combining the generated 'basic product information' and 'advanced product information', and is an integrated index that can template and style product data. It can be technically utilized to optimize the product display method by summarizing various characteristics of the product into a single number.

For example, if the product information score is high, the product template can emphasize the unique features and advanced information of the product by including detailed and rich visual elements. This can focus on attracting consumers' attention and encouraging purchase by including information such as high-resolution images, video reviews, and detailed feature lists. On the other hand, if the product information score is low, the template can be simplified to provide only basic information to avoid confusing consumers, and use clear and concise expressions to convey only the basic features of the product, making it easy for consumers to understand the product.

The template information analysis unit (320) can receive template information from an external server based on user input template information entered by the user.

Here, the template information may include template matching accuracy, data tagging automation ratio, template flexibility index, multimodal integration degree, template adjustment response time, product attribute extraction efficiency, data standardization index, template update frequency, automatic classification performance, dynamic template adjustment function, template-based content creation speed, data security index, user experience optimization score, artificial intelligence integration efficiency, API linkage index, template-based reporting efficiency, content creation automation index, metadata formatting efficiency, product comparison analysis automation degree, and interactive visualization effectiveness.

When template information is received, the template information analysis unit (320) can classify the received template information into at least one category based on at least one attribute.

For example, the template information analysis unit (320) may classify template matching accuracy, data tagging automation ratio, template flexibility index, multimodal integration degree, template adjustment response time, product attribute extraction efficiency, data standardization index, template update frequency, and automatic classification performance as template basic structures, and may classify dynamic template adjustment function, template-based content creation speed, data security index, user experience optimization score, artificial intelligence integration efficiency, API interoperability index, template-based reporting efficiency, content creation automation index, metadata formatting efficiency, product comparison analysis automation degree, and interactive visualization effectiveness as template functions.

The template information analysis unit (320) classifies template information classified by template basic structure and template function into each category and manages it, thereby enabling real-time monitoring of template information and early identification and management of any deviations from the management pattern. In addition, the convenience of information management can be increased by managing the index of information classified by each category.

Meanwhile, the categorization of template information and the calculation of template information scores can use artificial intelligence-based machine learning or deep learning models, and for example, artificial intelligence learning models including supervised learning, unsupervised learning, reinforcement learning, convolutional neural networks (CNNs), recurrent neural networks (RNNs), long short-term memory networks (LSTMs), natural language processing (NLP), decision support systems (DSS), and predictive analysis can be used.

In addition, the template information analysis unit (320) can monitor information classified into each category in real time and generate a notification (e.g., sound, vibration, light, etc.) for warning when a predetermined deviation range (e.g., 20%) is exceeded.

Next, the template information analysis unit (320) can calculate a template information score by mutually combining the template information classified in each category. Meanwhile, unless otherwise explained, the weights w1 to wn included in the mathematical formula described below may be predetermined weight values whose sum may be 1.

For example, the template information analysis unit (320) can calculate the template basic structure score as shown in the mathematical formula below.

Here, the data tagging automation ratio indicates how efficiently the template can automatically tag data, which can increase the automation and efficiency of tasks.

That is, the template information analysis unit (320) evaluates the function of automatically tagging product images and descriptions in an online content management system, and uses image recognition and natural language processing technology to recognize product characteristics (e.g., color, size, material) and automatically tag the corresponding information in a database.

For example, if the total number of processed product data is 1,000, the number of product data attempted for automatic tagging, including success or failure of automatic tagging, is 900, and among these, the number of product data successfully tagged is 800, the automatic tagging success rate can be calculated as 0.89 (=800/900).

For example, the template information analysis unit (320) can extract features from a product image using a deep learning-based image recognition algorithm, and can recognize and classify the color, shape, and pattern of the image using, for example, Convolutional Neural Networks (CNN).

In addition, product descriptions can be analyzed based on natural language processing algorithms (NLP) to extract keywords and important information (e.g., material, size), and these can be automatically tagged to relevant fields in the product database. The accuracy of tagging can be evaluated using a predefined test data set, which contains manually tagged correct answers, and the accuracy can be measured by comparing it with automatically generated tags.

The template flexibility index can be used as a metric to evaluate how flexibly a template can be applied to different data types and scenarios.

The data standardization index is an indicator that measures how consistently a template can convert various data inputs into a standardized format. This index can be an indicator of the consistency of data and the accuracy of the processing.

For example, the template information analysis unit (320) can measure the success rate of conversion from the original format of the data to the standard format required by the template. For example, text data can mean conversion from various languages and encodings to UTF-8, date/time data can mean conversion from various date/time formats to the ISO 8601 standard, and numeric data can mean conversion from various numeric formats (e.g., using commas, periods) to a standard numeric format.

Next, the template information analysis unit (320) can calculate a data standardization index based on the standardization success rate for each data type. For example, it is assumed that 90 out of 100 attempts at text data conversion are successful, 95 out of 100 attempts at date/time data conversion are successful, and 85 out of 100 attempts at numeric data conversion are successful.

This demonstrates that the template can transform a variety of data types into a standard format in a fairly consistent manner, allowing for an objective assessment of the efficiency and consistency of the data processing template and identification of any necessary improvements.

The degree of multimodal integration can measure how well a template can integrate various data types, such as text, images, and videos. In other words, it is an indicator that measures how well a template can integrate and process various data types (such as text, images, and videos). This can evaluate the ability of a template to harmoniously combine various data types and, as a result, provide consistent information to the user.

That is, the degree of multimodal integration can be calculated as the ratio of how successfully the template can process data in a scenario where various data types are integrated, by measuring the success rate of processing each data type individually, such as text, image, and video, and how successfully the template can process data in a scenario where various data types are integrated, i.e., when the data is accurately transformed to the requirements of the template or system, or perfectly adapted to the desired format.

For example, if 95, 90, and 85 out of 100 text, image, and video data were converted normally, respectively, and in the case of a multimodal case where at least two text, image, and video data were combined, if 80 out of 100 data were successfully converted, the degree of multimodal integration is calculated as follows.

The degree of multimodal integration quantitatively evaluates the ability of a template to integrate and process various data types. A degree of multimodal integration close to 1 indicates that the template can process multimodal data as effectively as individual data types. A degree of multimodal integration less than 1 indicates that multimodal data processing is less efficient than processing individual data types, which may indicate that the integrated processing capability of the template needs to be improved.

The generated template basic structure score is an indicator that comprehensively evaluates the efficiency, flexibility, and multimodal integration capabilities of the template, which can be used to determine how well the template is designed, and reflect how well the template can adapt to and integrate various data and requirements.

For example, a high template basic structure score may indicate that the template efficiently auto-tags data, is flexible in its application to various data types, and can effectively integrate multimodal content, while a low template basic structure score may indicate that the template lacks automation, flexibility, and integration, which may indicate the need for template improvement.

The template information analysis unit (320) can calculate the template function score as shown in the mathematical formula below.

Here, the template-based content generation speed can be calculated based on the content generation completion time for each page or item when generating content by applying the content to the template based on the received product information.

For example, when the template information analysis unit (320) takes 120 seconds to generate each page, the template-based content generation speed is calculated as follows.

For example, the template-based content generation speed can be normalized to 0 to 1 through min-max normalization within a range of predetermined minimum and maximum values. For example, if the minimum template-based content generation speed is 0 page/sec, the maximum template-based content generation speed is 1 page/sec, and the actual template-based content generation speed is 0.0083 page/sec, the template-based content generation speed can be normalized to 0.9917 (=1-0.0083).

The API Interoperability Index is an indicator that measures how efficiently a template or system can interact with other external systems via APIs. This index is based on the success rate of API calls and can quantitatively evaluate the reliability and efficiency of integration.

For example, if the template information analysis unit (320) calls the API a total of 100 times and 95 of these API calls are successful, the API interconnectivity index can be calculated as 0.95.

Metadata format efficiency is an index that measures how accurately and efficiently a data processing system can convert input metadata into a standardized format. This index can quantitatively evaluate the degree to which data adheres to a standardized format.

For example, metadata items in a template can be separated into items where product information will be entered, including information about the product, such as name, description, price, manufacturer, and category.

At this time, the template information analysis unit (320) can determine whether a metadata item has been assigned to an accurate template item when applying a template according to the metadata of the received product information.

For example, if metadata for product information A is incorrectly assigned to a template item or there is no suitable product information metadata for the template item, it may be determined that metadata is not assigned correctly.

For example, if the total number of metadata items in the template is 100 and the number of mapped metadata items is 95, the metadata formatting efficiency can be calculated as 0.95.

The generated template function score can indicate the efficiency of the functional elements of the template. A high template function score indicates that the template generates content quickly and efficiently, has high integration with external systems, and has high data formatting and operational efficiency. A low template function score indicates that the template has slow content generation speed, inefficient API linking, and inefficient metadata formatting, which may indicate a need for improvement.

The template information analysis unit (320) can calculate a template information score based on the calculated template basic structure score and template function score as in the following mathematical formula. Here, w1 to w2 are predetermined weights whose sum can be 1, and the calculated template basic structure score and template function score can be normalized values from 0 to 1 based on predetermined minimum and maximum values.

The generated template information score is calculated by integrating the "template basic structure" and "template function", and can be an indicator for evaluating the overall efficiency and usability of the template in processing and providing information. In addition, by combining the structural completeness and functional efficiency of the template, it can quantitatively indicate how well the template is designed and satisfies user needs.

The style optimization unit (330) can calculate a style optimization score from a weighted sum of the calculated product information scores and template information scores.

The 'Style Optimization Score' is an indicator calculated by integrating the 'Product Information Score' and the 'Template Information Score', and can be an indicator to evaluate the comprehensive performance in the process of optimizing the style of a template.

A high style optimization score may indicate that the template effectively optimizes the style to reflect the user's needs and the product's characteristics, which may indicate its ability to maximize user experience and lead to higher user satisfaction and interest.

On the other hand, a low score may indicate that the template does not effectively convey product information or meet users' visual expectations, and may indicate a need for design and functionality improvements.

The style optimization unit (330) can limit the maximum template items using the calculated style optimization score.

Here, the basic template items may refer to the number of items in the template to which product information on the product page provided by each online shopping mall is allocated. For example, the number of items in the basic template for registering 'clothing' as a product and the number of items in the basic template for registering 'washing machine' as a product may be different from each other.

Meanwhile, the template complexity adjustment factor ( ) is an adjustment coefficient for adjusting the sensitivity of changes in the maximum number of template items according to the style optimization score, and the maximum template items calculated through this can adjust the number of items in the basic template items, thereby limiting the number of template items that can be allocated and displayed at the maximum in the basic template.

Here, the template complexity adjustment factor ( ) can be calculated as the average of the color diversity index, the font diversity index, the image diversity index, and the layout diversity index. At this time, the color diversity index, the font diversity index, the image diversity index, and the layout diversity index can be given scores for each index according to predetermined criteria.

For example, the color diversity index can be given as 0.2 when the number of colors in the template is 5 or less, 0.4 when 6 to 10 colors are used, and 0.7 when 11 or more colors are used.

The font diversity index can be awarded 0.4 points when using 1 to 2 font styles, 0.6 points when using 3 to 4 font styles, and 0.9 points when using 5 or more font styles.

Additionally, the image diversity index can be awarded 0.4 points if there are 1 to 3 types of image template items, 0.7 points if there are 4 to 6 image template items, and 1 point if there are 7 or more image template items.

Additionally, the layout diversity index can represent the layout complexity of a template, and can be derived from the layout entropy according to the diversity of each layout item.

Here, p_i represents the area ratio occupied by the i-th element in the template page. For example, when the template page consists of header, body, sidebar, and footer, and the area ratios occupied by each are 10%, 50%, 20%, and 20%, the layout diversity index is calculated as follows.

For example, the layout diversity index can be normalized to 0 to 1 through min-max normalization within a range of predetermined minimum and maximum values. For example, if the minimum layout diversity index is 0, the maximum layout diversity index is 3, and the actual layout diversity index is 1.76, the layout diversity index can be normalized to 0.59.

This may indicate that the elements of the template are arranged in a fairly diverse manner, which is likely to attract visual interest and enrich the user's browsing experience.

The basic template items are 15, the pre-calculated style optimization score is 0.8, and the template complexity adjustment factor ( ) is 0.4, the maximum template items are calculated as follows.

That is, the template complexity adjustment factor ( ) can adjust the sensitivity of changes in the maximum template item by adjusting the style optimization score calculated according to the complexity of the template, and in case of a complex template layout, the template complexity adjustment factor ( ) is calculated highly, so that the maximum template items can be significantly adjusted to allocate product information according to the prototype of the template.

Figure 4 is a hardware configuration diagram for a server for automatically collecting and stylizing non-standard product data according to Figure 1.

Referring to FIG. 4, the product non-standard data automatic collection and stylization processing server (400) may include at least one processor (processor, 410) and a memory (memory, 420) that stores instructions that instruct at least one processor (410) to perform at least one step.

Here, at least one processor (410) may mean a central processing unit (CPU), a graphics processing unit (GPU), or a dedicated processor on which methods according to embodiments of the present invention are performed. Each of the memory (420) and the storage device (460) may be configured with at least one of a volatile storage medium and a nonvolatile storage medium. For example, the memory (420) may be configured with at least one of a read only memory (ROM) and a random access memory (RAM).

In addition, the product non-standard data automatic collection and stylization processing server (400) may include a transmission/reception device (430) that performs communication via a wireless network. In addition, the product non-standard data automatic collection and stylization processing server (400) may further include an input interface device (440), an output interface device (450), a storage device (460), etc. Each component included in the product non-standard data automatic collection and stylization processing server (400) may be connected by a bus (bus, 470) and communicate with each other.

The methods according to the present invention may be implemented in the form of program commands that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program commands, data files, data structures, etc., alone or in combination. The program commands recorded on the computer-readable medium may be those specially designed and configured for the present invention or may be known and available to those skilled in the art of computer software.

Examples of computer-readable media may include hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions may include not only machine language codes generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter, and the like. The above-described hardware devices may be configured to operate with at least one software module to perform the operations of the present invention, and vice versa.

In addition, the above-described method or device may be implemented by combining all or part of its configuration or function, or may be implemented separately.

Although the present invention has been described above with reference to preferred embodiments thereof, it will be understood by those skilled in the art that various modifications and changes may be made to the present invention without departing from the spirit and field of the present invention as set forth in the claims below.

Claims (3)

As a server that automatically collects and stylizes non-standard product data,
at least one processor; and
A memory storing instructions that instruct at least one processor to perform a plurality of operations,
The above multiple actions are:
The action of receiving product information from an external server;
The action of receiving template information from an external server;
An action of extracting metadata of the received product information and template information;
An operation of saving the extracted metadata in CSV format and specifying a product category from the metadata extracted from the product information; and
An operation of allocating product information based on metadata extracted from the product information to a predetermined product page layout template based on the specified product category;
By combining the above product information,
An operation of calculating a product basic information score and a product advanced information score, and calculating a product information score from a weighted sum of the calculated product basic information score and the product advanced information score;
An operation of calculating a template basic structure score and a template function score by mutually combining the above template information, and calculating a template information score from a weighted sum of the calculated template basic structure score and the template function score;
An operation of calculating a style optimization score from a weighted sum of the calculated product information score and the template information score; and
The above style optimization score produced, template complexity adjustment factor ( ) and an operation of producing a maximum template item from the product of the basic template items; characterized in that it further includes;
A server that automatically collects and stylizes non-standard product data. delete In claim 1,
The operation of producing the above maximum template items is:
From the above template information, the template complexity adjustment factor is calculated as the average of the color diversity index, font diversity index, image diversity index, and layout diversity index of each template. ), characterized by further including an operation for producing a );
A server that automatically collects and stylizes non-standard product data.

Images