Web Of Synonyms: An Enhanced Keyword Extraction Model For Recommendation Systems

In the realm of information retrieval, the effectiveness of search engines hinges significantly upon their ability to interpret user queries accurately. Present-day systems predominantly furnish suggestions based on either product specifications or individual user profiles. However, traditional keyword extraction techniques often suffer from limitations such as ambiguity and lack of context, leading to suboptimal recommendations. Keywords serve as pivotal determinants in the recommendation process for most systems. A more sophisticated recommendation framework that incorporates the semantic subtleties of searches instead of depending only on previous data from worldwide queries has the potential to revolutionize computerized suggestion technology. Conventional keyword extraction methods often rely on the definition of a word or the contextual understanding of textual content, albeit often yielding extraneous or unrelated terms. This research explores the capacity to enhance recommendation systems to bolster e-commerce platforms’ business performance through the creation of a word synonym graph. This paper presents a novel approach to keyword extraction for recommendation systems, leveraging the vast resources of the web to identify synonyms related to user queries. The paper describes the model’s architecture and main components, which include a web scraper module, graph generator, and graph filtering using the Wu-Palmer similarity method and its results.