obdemo.pdf

Consolidation of Ontologies fromWeb Sources using Sequence Semantics Haggai Roitman, Avigdor Gal Technion – Israel Institute of Technology Technion City, Haifa 32000, Israel 1 Introduction Ontologies, formal speci?cations of domains, have evolved in recent years as a lead- ing tool in representing and interpreting Web data. The inherent heterogeneity of Web resources, the vast amount of information on the Web, and its non-speci?c nature re- quires a semantically rich tool for extracting the essence of Web source content. The OntoBuilder project [5] supports the extraction of ontologies fromWeb interfaces, rang- ing from simple Search Engine forms to multiples, complex reservation systems. Ontologies from similar domains are then matched to identify ontology mappings. Given a sample form, ?lled by the user, and given a new form, from another Web site, OntoBuilder ?nds the best mapping between the two forms. This, in turn, can serve a system in automatically ?lling the ?elds, a sort of a query rewriting. Unlike systems such as Prot′ eg′ e [2] OntoBuilder enables fully-automatic ontology matching, and therefore falls within the same category as GLUE [1]. The use of on- tologies, as opposed to relational schema or XML, as an underlying data model allows a ?exible representation of metadata, that can be tailored to many different types of applications. OntoBuilder contains several unique matching algorithms, that can match concepts (terms) by their data types, constraints on value assignment, and above all, the sequencing of concepts within forms (termed precedence), capturing sequence seman- tics that re?ect business rules. 2 Overview of OntoBuilder OntoBuilder was developed using Java, which makes it portable to various platforms and operating system environments. OntoBuilder generates dictionary of terms by ex- tracting labels and ?eld names fromWeb forms, and then it recognizes unique relation- ships among terms, and utilize them in its matching algorithms. There are