Abstract:

This research aims to develop an effective search approach for Sustainable platforms. Sustainability is a concept that focuses on minimizing waste and optimizing resource utilization to establish a sustainable and resilient economic system. As the importance of Sustainability continues to grow, it is crucial to provide efficient and effective methods for accessing relevant products on Sustainable platforms. To achieve a sustainable platform, the proposed approach employs semantic search, which is an advanced search technique that comprehends the user’s search query’s intent and context to provide more precise and relevant results than the traditional keyword-based search. This paper presents a framework for semantic search for Sustainable platforms using the Sentence Bidirectional Encoder Representations (SBERT) from the Transformers model, which has proven to be a successful approach to semantic search. This model translates both the product catalog and user queries into vector space and applies machine learning algorithms like K-Nearest Neighbours (KNN) and cosine similarity to match queries with relevant products. The performance of this approach is evaluated using a dataset of Sustainable platform products. The results indicate that the proposed semantic search approach outperforms the traditional keyword-based search in terms of precision and recall, demonstrating its ability to support users in finding relevant products on Sustainable platforms effectively.

Keywords: Natural Language Processing, Sentence-BERT, Product Search, Cosine Similarity, Semantic Search.

Published in: IEEE International Conference on Emerging Techniques in Computational Intelligence (ICETCI 2023)