# US 2024/0202789 A1 · A system and method for detecting the duplicate product on the e-commerce platforms

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2024/0202789 A1
- Document ID: US20240202789A1
- Status: Published application
- Kind: A1
- Application number: 18/555,746
- Filed: 2022-04-12
- Published: 2024-06-20
- Earliest claimed: 2021-04-17
- Assignees: D-Market Elektronik Hizmetler Ve Ticaret Anonim Sirketi
- Applicants: D-Market Elektronik Hizmetler Ve Ticaret Anonim Sirketi
- Inventors: Osman Semih ALBAYRAK; Tolga Ahmet KALAYCI; Firat ONCU
- Web page: https://patentgrep.com/patent/US20240202789A1

## CPC

- G06Q 30/0603: {Catalogue creation or management}
- G06F 16/215: Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
- G06Q 10/08: Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

The invention is a duplicate product entry detection system and method for detecting the duplicate products in the product catalogs of the e-commerce platforms miming on the mobile devices or IT devices. The duplicate product entries can be searched by the duplicate product detection engine running on a deduplication engine server (20), the similarity calculations can be carried out, and the duplicate product detection can be performed for the finally detected product pairs.

## Claims

### Claim 1

A duplicate product entry detection method for detecting duplicate products in product catalogs of e-commerce platforms running on mobile devices or IT devices, wherein by means of:
a product catalog database server comprising data on the products in the e-commerce website,
a deduplication engine server for detecting the duplicate products,
a product similarity metrics database server comprising similarity data calculated for duplicate product pairs by the deduplication engine server,
an FTP server in which the data on the duplicate product pairs detected by the deduplication engine server is stored as a CSV file,
an application server for labeling the duplicate products,
a similar product database server comprising the data on the duplicate product pairs,
a service server for performing the operations such as searching, logging, updating, entering the operation results for the duplicate product entries,
the method comprising the following steps:
sending the product information from the product catalog database server to the deduplication engine server for the purposes of querying the duplicate products by the duplicate product detection engine, performing the similarity calculations, and detecting the final duplicate product,
sending the data on the product pairs predicted to be duplicates by the deduplication engine server to the FTP server as a CSV file to perform the product labeling operations in the form of “duplicate’ not duplicate” or “variant of each other” on the deduplication screens on the application server,
sending the similarity data calculated for the product pairs by a deduplication engine server to the similarity metrics database server to be used in the classification models,
sending the similarity metrics previously calculated for the product pairs from the similarity metrics database server to the deduplication engine server to be used to develop the classification model,
sending the data in the duplicate product file found in the FTP server and detected to be duplicate from the FTP server to the application server,
sending the data on the duplicate product files from the application server to the similar product database server,
listing the possible duplicate product entries on the deduplication screens on the application server, searching on the lists, labeling the duplicate products, and sending the labeled data to the service server,
entering the results of the labeling operation for the duplicate product entry data to the similar product database server by the service server,
retrieving the results of the tagging operation on the deduplication screens from the similar product database server by the deduplication engine server to be used by the duplicate product detection model, and
updating the product catalog database server by means of the service server to close the products detected to be duplicates.

### Claim 2 (depends on 1)

A duplicate product entry detection method according to claim 1, wherein the method comprises the following steps:
detecting the similar product pairs by the duplicate product detection engine running on the deduplication engine server using the traditional text similarity algorithms and accepting the product pairs above a certain threshold as the candidate duplicate product pairs,
labeling the candidate duplicate product list about whether it is a duplicate product or not via a duplicate product platform screen,
training the binary classification models with the labeled product pairs data and using the text similarity scores generated as the features of the model,
enriching the text-based model by using the additional variables expressing the similarities of the product images, price information and trademark characteristics.
