Hybrid semantic similarity distance measures using conceptual contexts for geospatial information retrieval

As the syntax based information retrieval yields very low precision, the researchers are moving to the semantic based information retrieval. The great challenge of the information retrieval system is to retrieve the related relevant information from the heterogeneous web resources. This paper aims for the retrieval of the relevant geo-spatial concepts from the geo-spatial ontology which aids application in quality testing of water and flood prediction. In the geo-spatial concept space, the retrieval of relevant geo-spatial concepts depends on the distance separating the two geo-spatial concepts and the context of the concepts. Research in geo-spatial information retrieval has used either the semantic distance between the geo-spatial concepts or the context of the concepts for computing the relevancy. This paper proposes HCC (Hybrid Conceptual Context) algorithm which aims at retrieving relevant geo-spatial concepts by considering both the semantic distance and the context on which the concepts occur in the geo-spatial space. This work computes the semantic similarity by measuring the semantic distance using Euclidean and Manhattan distance measure. This computation is carried out in two steps such as retrieval of semantically similar geo-spatial concepts and the retrieval of context dependent geo-spatial concepts. Semantic similarity computation in HCC algorithm retrieves contextually and semantically similar relevant geo-spatial concepts using Euclidean distance and Manhattan distance measures. Experiments have been done using Ordnance Survey Master Map datasource and geo-spatial ontology. The precision and recall show Manhattan based semantic similarity computation improves relevancy by 10%.