Indian Journal of Science and Technology
Year: 2015, Volume: 8, Issue: 27, Pages: 1-11
Mozhdeh Nazari Soleimandarabi1* and Seyed Abolghasem Mirroshandel2
1 Department of Computer Engineering, Rasht Branch, Islamic Azad University, Rasht, Iran; [email protected]
2 Department of Computer Engineering, University of Guilan, Rasht, Iran; [email protected]
Background: Computing semantic relatedness measures are extensively employed in the field of Natural Language Processing (NLP) and play pivotal role in Geographic Information Science (GIS). Methods/Analysis: Noteworthy, despite the significance of semantic relatedness in geographic domain, its role has been almost ignored. While most of the proposed measures in this context are only able to compute semantic similarity, in this paper, the notion of geosemantic relatedness is discussed and from which a novel approach for computing semantic relatedness of geographic terms is proposed. The proposed method utilizes term’s definition from geographic lexicon. Whereas lexical definition demarcates the boundaries of a term and provides valuable semantic space for deducting the meaning of a term, it can have prominent impact in the efficiency of the proposed approach. Furthermore, this approach exploits Wikipedia as semantic resource that has considerable performance in application of semantic relatedness. Finding: The cognitive plausibility of the proposed approach is evaluated on GeReSiD dataset. Compared to the previous state of the arts, using proposed approach results significant improvement in correlation of computed relatedness score with human judgment to 0.73. Conclusion: Additionally, the proposed approach not only prospers higher perception and adoption, but also it has greater applicability in real world problems and is confronted with fewer limitations. Moreover, the proposed method can perform disambiguation in geographic domain properly.
Keywords: Definition, Geographic Lexicon, Geo-Semantic, Geographic Semantic Relatedness, Wikipedia
Subscribe now for latest articles and news.