Exploiting Cluster-Skipping Inverted Index for Semantic Place Retrieval

Cinar E. R., Altingövde I. S.

46th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2023, Taipei, Tayvan, 23 - 27 Temmuz 2023, ss.1981-1985 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1145/3539618.3591983
  • Basıldığı Şehir: Taipei
  • Basıldığı Ülke: Tayvan
  • Sayfa Sayıları: ss.1981-1985
  • Anahtar Kelimeler: location-based search, spatial keyword query processing
  • Orta Doğu Teknik Üniversitesi Adresli: Evet

Özet

Semantic place retrieval aims to find the top-k place entities, which are both textually relevant and spatially close to a given query, from a knowledge graph. In this work, our contribution toward improving the efficiency of semantic place retrieval is two-fold. First, we show that by applying an ad hoc yet intuitive restriction on the depth of search on the knowledge graph, it is possible to adopt IR-tree indexing scheme [7], which has been introduced for processing spatial keyword queries, for the semantic place retrieval scenario. Secondly, as a novel solution to this problem, we adapt the idea of cluster-skipping inverted index (CS-IIS) [1, 4], which has been originally proposed for retrieval over topically clustered document collections. Our experiments show that CS-IIS is comparable to IR-tree in terms of CPU time, while it yields substantial efficiency gains in terms of I/O time during query processing.