An efficient Top-k Closest Pair Queries over Spatial Knowledge Graph
TL;DR
A spatial entity node pair query method that combines K 2 - tree spatial indexing and DistilBERT-whitening model, which uses K 2 - tree for efficient spatial indexing and supports semantic query with DistilBERT-whitening model to mitigate the keyword-mismatch problem of exact string matching and improve the semantic recall coverage and speed of semantic query is introduced.
Abstract
In the processing and querying of spatial RDF data, how to efficiently query spatial entity node pairs that combine spatial and textual information is an important research direction. Traditional spatial entity node pair query methods are not efficient enough in terms of spatial indexing efficiency and text similarity matching. To tackle this issue, we introduce a spatial entity node pair query method that combines K 2 - tree spatial indexing and DistilBERT-whitening model, which uses K 2 - tree for efficient spatial indexing and supports semantic query with DistilBERT-whitening model to mitigate the keyword-mismatch problem of exact string matching and improve the semantic recall coverage and speed of semantic query. A set of pruning mechanisms, which are all lossless and preserve the exact top-k result of exhaustive search, is developed to lower computational overhead. In addition, a GPU-enabled parallel architecture is adopted to accelerate spatial indexing and semantic matching, leading to enhanced query throughput and reduced latency. The experimental results show that compared with the traditional spatial entity node pair query method, our proposed scheme has significantly improved the query speed, and at the same time, it can significantly reduce the spatial overhead of spatial indexing under the premise of maintaining efficient query, and the results prove the efficiency of our proposed algorithm.