핵심 개념
The author proposes LIST, a novel technique that learns to index spatio-textual data for embedding-based spatial keyword queries, addressing issues of traditional models and deep learning methods. LIST outperforms existing methods in effectiveness and efficiency.
초록
The proliferation of spatio-textual data has led to the need for efficient spatial keyword query processing. Existing models face limitations in text relevance computation and spatial relevance assumptions. The proposed LIST method introduces a lightweight relevance model and an ANNS index to improve effectiveness and efficiency significantly. By incorporating deep learning techniques, LIST achieves superior performance compared to traditional models.
통계
"Experimental results show that LIST significantly outperforms state-of-the-art methods on effectiveness, with improvements up to 19.21% and 12.79% in terms of NDCG@1 and Recall@10."
"LIST is three orders of magnitude faster than the most effective baseline."