A ML-based Query Expansion in Vertical Aggregated Web Search Using Pseudo-Relevance Feedback
International Journal of Innovations in Science and Technology · 08.05.2026 · 307
DOI: 10.33411/ijist/1778 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Innovations in Science and Technology, online erschienen am 08.05.2026. Verfasst von kashia Riaz und 1 weiteren Autor:innen. Die Arbeit nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The massive increase in multimedia content over the internet has posed a significant challenge to vertical search engines in instances where users submit short, ambiguous, or vocabulary-mismatched queries. Current query expansion techniques, including Pseudo-Relevance Feedback (PRF), show limited effectiveness in addressing these challenges, particularly in multimedia and cross-vertical search settings. The developed hybrid query expansion method described in this paper combines PRF, Machine Learning (ML), and Graph Theory to enhance the relevance of search results and enhance the semantic relevance of search results. The methodology consists of applying PRF to extend the first query with the help of relevant terms and then using a machine learning model to narrow the terms according to past search behavior and semantic trends. Also, graph-based techniques are utilized to determine semantic relationships among query terms, thereby improving contextual relevance of the query. The framework identifies semantically related terms and refines the query to improve retrieval relevance and adjusts the search, so the results are more accurate and relevant. This experiment used the TREC Web Track collection, which consists of 50 queries, 25,000 documents, and some artificially generated multimedia sources. The empirical findings show that there are great improvements in essential performance measures. Specifically, the given approach yields a 36.2% increase in MAP, 28.5% in nDCG@10, and 32.6% in P@10, as compared to BM25. In addition, a 30-user study has shown a 40% reduction in the query formulation time and a 25-30 percent increase in user effectiveness. The results indicate better accuracy of retrieval, lower query effort, and increased user satisfaction, especially for novice users involved in exploratory search activities. Additionally, issues of scalability, data privacy, and the computation cost are presented in the real-world application.
Zitieren
APA
Riaz, K., & Rashid, U. (2026). A ML-based Query Expansion in Vertical Aggregated Web Search Using Pseudo-Relevance Feedback. International Journal of Innovations in Science and Technology, 307. https://doi.org/10.33411/ijist/1778
Vancouver
Riaz K, Rashid U. A ML-based Query Expansion in Vertical Aggregated Web Search Using Pseudo-Relevance Feedback. International Journal of Innovations in Science and Technology. 2026:307. doi:10.33411/ijist/1778
BibTeX
@article{Riaz2026,
author = {Riaz, kashia and Rashid, Umer},
title = {A ML-based Query Expansion in Vertical Aggregated Web Search Using Pseudo-Relevance Feedback},
journal = {International Journal of Innovations in Science and Technology},
year = {2026},
pages = {307},
doi = {10.33411/ijist/1778}
}
EUVIMED