# US 2026/0064756 A1 · Method and system for identification of product taxonomy from product abbreviation

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

- Number: US 2026/0064756 A1
- Document ID: US20260064756A1
- Status: Published application
- Kind: A1
- Application number: 19/306,272
- Filed: 2025-08-21
- Published: 2026-03-05
- Earliest claimed: 2024-08-29
- Assignees: Tata Consultancy Services Limited
- Applicants: Tata Consultancy Services Limited
- Inventors: BAGYA LAKSHMI VASUDEVAN; KUNIKA SARASWAT; AMRITA ANAND; AKSHAT JAIN; MANASI SAMARTH PATWARDHA; MAYUR PATIDAR
- Web page: https://patentgrep.com/patent/US20260064756A1

## CPC

- G06F 16/35: Clustering; Classification
- G06F 16/367: {Ontology}
- G06F 16/951: Indexing; Web crawling techniques
- G06F 16/953: Querying, e.g. by the use of web search engines
- G06F 40/242: Dictionaries
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

The present invention generally relates to the field of taxonomy identification. Identifying product taxonomy from a product abbreviation is currently performed manually and consumes lot of time and effort. Hence, embodiments of present disclosure provide an automated method for identification of product taxonomy from product abbreviation. First, a brand name of product abbreviation is predicted using a Large Language Model (LLM) and a brand list. Then, possible expansions of the abbreviation are generated based on the brand name using acronym expansion dictionary. Among the generated possible expansions, a relevant one is identified using the LLM. Later, web scraping and web search techniques are applied on the relevant expansion to obtain associated top k matches of a supergroup, a product group and module of the predicted relevant expansion. Finally, product taxonomy is predicted based on the top k matches using a LLM augmented taxonomy classification by Retrieval Augmented Generation.

## Claims

### Claim 1

A processor implemented method comprising:
obtaining, via one or more hardware processors, a product abbreviation;
predicting, via the one or more hardware processors, a brand name using a Large Language Model (LLM) based on the product abbreviation and a brand list, wherein the brand list comprises a plurality of brand names extracted from a brand acronym dictionary;
generating, via the one or more hardware processors, a plurality of expansions of the product abbreviation in accordance with the predicted brand name using an acronym expansion dictionary;
determining, via the one or more hardware processors, a relevant expansion from among the plurality of expansions of the product abbreviation using the LLM;
applying, via the one or more hardware processors, web scraping and web search technique on the predicted relevant expansion to obtain associated top k matches of a supergroup, a product group and a module of the predicted relevant expansion; and
predicting, via the one or more hardware processors, a product taxonomy of the determined relevant expansion based on the associated top k matches using a LLM augmented taxonomy classification by Retrieval Augmented Generation (RAG).

### Claim 2 (depends on 1)

The method of claim 1, wherein each of the plurality of expansions of the product abbreviation comprises the brand name followed by one or more hierarchical levels of descriptions comprising the product taxonomy with the super group, the product group and the module associated with the product abbreviation.

### Claim 3

A system, comprising:
a memory storing instructions;
one or more Input/Output (I/O) interfaces; and
one or more hardware processors coupled to the memory via the one or more communication interfaces, wherein the one or more hardware processors are configured by the instructions to:
obtain a product abbreviation;
predict a brand name using a Large Language Model (LLM) based on the product abbreviation and a brand list, wherein the brand list comprises a plurality of brand names extracted from a brand acronym dictionary;
generate a plurality of expansions of the product abbreviation in accordance with the predicted brand name using an acronym expansion dictionary;
determine a relevant expansion from among the plurality of expansions of the product abbreviation using the LLM;
apply web scraping and web search technique on the predicted relevant expansion to obtain associated top k matches of a supergroup, a product group and module of the predicted relevant expansion; and
predict a product taxonomy of the determined relevant expansion based on the associated top k matches using a LLM augmented taxonomy classification by Retrieval Augmented Generation (RAG).

### Claim 4 (depends on 3)

The system of claim 3, wherein each of the plurality of expansions of the product abbreviation comprises the brand name followed by one or more hierarchical levels of descriptions comprising the product taxonomy with the super group, the product group and the module associated with the product abbreviation.

### Claim 5

One or more non-transitory machine-readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause:
obtaining a product abbreviation;
predicting a brand name using a Large Language Model (LLM) based on the product abbreviation and a brand list, wherein the brand list comprises a plurality of brand names extracted from a brand acronym dictionary;
generating a plurality of expansions of the product abbreviation in accordance with the predicted brand name using an acronym expansion dictionary;
determining a relevant expansion from among the plurality of expansions of the product abbreviation using the LLM;
applying web scraping and web search technique on the predicted relevant expansion to obtain associated top k matches of a supergroup, a product group and a module of the predicted relevant expansion; and
predicting a product taxonomy of the determined relevant expansion based on the associated top k matches using a LLM augmented taxonomy classification by Retrieval Augmented Generation (RAG).

### Claim 6 (depends on 5)

The one or more non-transitory machine-readable information storage mediums of claim 5, wherein the one or more instructions which when executed by the one or more hardware processors further cause the product taxonomy with the super group, the product group and the module associated with the product abbreviation.
