Method and system for creating a multimodal and multilingual product catalogue using a hybrid model

A method for creating a product catalogue, the method comprising:

• obtaining, by a computer system, data related to a product;
• extracting, by the computer system, text from the data related to the product;
• preprocessing the extracted text;
• contextually employing, by the computer system, a hybrid model based on a type of the data, wherein the hybrid model is an integration of a rule-based model, a Named Entity Recognition (NER) model, and a Generative-AI (Gen-AI) model,
• wherein the hybrid model is configured to:
• extract, using the rule-based model, one or more first attributes and one or more values from the data based related to the product based on predefined structures, patterns, and regular expressions in text data;
• extract, using the NER model, one or more second attributes and one or more corresponding values from unstructured or semi-structured text data related to the product, wherein the NER model tokenizes the text data using a tokenizer that supports multiple languages and represents tokens as word embedding vectors, and encodes entity labels for product attributes using a tagging scheme consistently across the multiple languages to identify and categorize named entities within the text data;
• generate, using the Gen-AI model, new or missing context-dependent values for one or more third attributes based on descriptions of the data related to the product;
• generating one or more attribute-value pairs and mapping the one or more attribute-value pairs to the product in a structured format;
• creating the product catalogue using the one or more attribute-value pairs;
• and
• combining outputs from the rule-based model, the NER model, and the Gen-AI model to generate a final taxonomy corresponding to the product catalogue, wherein the one or more attribute-value pairs are normalized and mapped to a predefined product taxonomy and product catalog template to create the final taxonomy, and wherein the final taxonomy is saved in a machine-readable format and made accessible via one or more application programming interfaces (APIs) for integration with online platforms.