KSE-Web: An Analysis of Hybrid Retrieval and LLM-Assisted Query Expansion for Low-Resource Khmer Semantic Search
Researchers introduce KSE-Web, a framework for Khmer semantic search using hybrid retrieval and LLM-assisted query expansion.
Khmer is a low-resource language with significant retrieval challenges, including ambiguous word boundaries and limited annotated data. KSE-Web addresses these by combining hybrid retrieval methods with LLM-based query expansion, tested on a curated dataset of 3,000 cleaned documents and 300 manually reviewed search queries.