In the realm of data retrieval, sparse vectors serve as a pivotal representation for both documents and queries, where each element in the vector denotes a word or phrase from a predefined lexicon. In this study, multiple scoring mechanisms are introduced aimed at discerning the significance of specific terms within the context of a document extracted from an extensive textual dataset. Among these techniques, the widely employed method revolves around inverse document frequency (IDF) or Term Frequency-Inverse Document Frequency (TF-IDF), which emphasizes terms unique to a given context. Additionally, the integration of BM25 complements TF-IDF, sustaining its prevalent usage. However, a notable limitation of these approaches lies in their reliance on near-perfect matches for document retrieval. To address this issue, researchers have devised latent semantic analysis (LSA), wherein documents are densely represented as low-dimensional vectors. Through rigorous testing within a simulated environment, findings indicate a superior level of accuracy compared to preceding methodologies.