Improving Query Expansion Techniques Using Domain-Specific Thesauri

Abstract

Information retrieval systems mostly use query expansion techniques to improve search execution accuracy, focusing on user queries and intention. Standard query expansion practices almost always depend on unspecialized domain frameworks, such as WordNet or statistical co-occurrence, which misintroduce ambiguous, irrelevant terms, especially in more advanced, specialized domains. This paper studies optimizing query expansion techniques by applying specialized domain thesauri that supply prepared, contextually relevant synonyms, related terms, and superior and subordinate relations relevant to specific fields. Enhancing semantic resources will optimize search results precision and recall, which are fundamental in domain-oriented search applications, thus improving precision and accuracy. The study introduces a hybrid query expansion framework that unites classical lexical query expansion with domain-specific thesauri query expansion to capture the most relevant expansion terms. We run experiments on specialised datasets, including medical and legal, engineering corpora, and compare the retrieval results with baseline methods. The most important measures of evaluation are precision, recall, and mean average precision (MAP). Incorporation of domain-specific thesaurus fragments is effective in reducing drift and introducing newly relevant semantics in extended queries, thereby enhancing relevance. This is the resultant enhancement that is shown as compared to precision, recall, and mean average correction measures. In addition, we explain how such an approach helps preserve and disambiguate better context, which is a critical challenge in domain-sensitive information retrieval. Lastly, we examine the domain-specific usage of query expansion in expert systems, digital libraries, and specialized search engines. We discuss the fact that it is difficult to maintain and update domain thesauri regarding changing terminology. We show that domain knowledge integration is important in query expansion and suggest new research opportunities such as the development of automated thesauri and expansion plans as part of adaptive pre-revision. This underlines the idea that semantic resources in a particular domain would increase the efficiency of information search and user experience in a special situation.