A Semantic-based String Matching Library

String matching has been widely used in more and more web data applications. Although a lot of matching techniques have been proposed, most of them are based on evaluating the similarity of strings in terms of their lexical or syntactic structures. The fact is that the same entity can be denoted by totally different expressions for which the conventional syntactic-based string matching techniques will not be effective any more. In this paper, a semantic-based string matching library (SSML) is proposed and implemented to achieve the semantic level string matching by combining all possible patterns for specific domain and appropriate matching techniques together. Particularly, two widely used domains are analyzed in details and implemented, that is, “Date” and “Person Name”. For each domain, a metric is proposed to quantify the potentiality of string matching. In general, under the framework of SSML, it is convenient and efficient to add new domains into the system.