Lexical Semantic Similarity

In regards to computing lexical similarity the twofundamental problems are respectively concerned withhow to explore concept relationships predefined andenumerated in lexical knowledge bases and how tostatistically induce and learn context relationshipsfrom word co-occurrences. To address these problemsthis book focuses on approaching both taxonomicsimilarity through the semantic networks in WordNetand distributional similarity through syntacticallyconstrained context. The taxonomic similarity modelwe proposed outperforms most popular similaritymethods with respect to simulating human similarityjudgments. In relation to distributional similaritywe thoroughly investigated the semantic properties ofgrammatical relationships in regulating wordmeanings whereby over 80% precision can be reachedin extracting synonyms or near-synonyms. This bookprovides a systematic guidance on computing taxonomicsimilarity and distributional similarity. It isappropriate for system developers and researchersworking in language technology.