ERSOM: A Structural Ontology Matching Approach Using Automatically Learned Entity Representation

Chuncheng Xiang,Baobao Chang,Zhifang Sui,Tingsong Jiang

doi:10.18653/v1/d15-1289

Abstract

As a key representation model of knowledge, ontology has been widely used in a lot of NLP related tasks, such as semantic parsing, information extraction and text mining etc. In this paper, we study the task of ontology matching, which concentrates on finding semantically related entities between different ontologies that describe the same domain, to solve the semantic heterogeneity problem. Previous works exploit different kinds of descriptions of an entity in ontology directly and separately to find the correspondences without considering the higher level correlations between the descriptions. Besides, the structural information of ontology haven’t been utilized adequately for ontology matching. We propose in this paper an ontology matching approach, named ERSOM, which mainly includes an unsupervised representation learning method based on the deep neural networks to learn the general representation of the entities and an iterative similarity propagation method that takes advantage of more abundant structure information of the ontology to discover more mappings. The experimental results on the datasets from Ontology Alignment Evaluation Initiative (OAEI1) show that ERSOM achieves a competitive performance compared to the state-of-the-art ontology matching systems. The OAEI is an international initiative organizing annual campaigns for evaluating ontology matching systems. All of the ontologies provided by OAEI are described in OWL-DL language, and like most of the other participates our ERSOM also manages the OWL ontology in its current version. OAEI: http://oaei.ontologymatching.org/

Highlights

An effective solution to the ontology heterogeneity problem is ontology matching (Euzenat et al, 2007; Shvaiko and Euzenat, 2013), whose main task is to establish semantic correspondences between entities from different ontologies
We first use Jena3 parsing the ontologies and extract descriptions for entities according to the description in section 2.1.1, we create a vocabulary based on the dataset and denote each class and property as a binary term vector
We use the learned representations to measure the similarities between classes and properties and apply the strategy presented in section 2.3 to extract final mappings

Summary

Introduction

An effective solution to the ontology heterogeneity problem is ontology matching (Euzenat et al, 2007; Shvaiko and Euzenat, 2013), whose main task is to establish semantic correspondences between entities (i.e., classes, properties or instances) from different ontologies. Entity’s label is always a specific substitution of its ID; entity’s comment is a semantic definition for its ID; a class can be characterized with its related properties, and a property is usually restricted by its domain and range These potential correlations of the descriptions are very important to measure the similarity between entities since they can be treated as some potential features describing an entity. The ontology matching system YAM++ (Ngo and Bellahsene, 2012) utilizes a global structural method but it only uses the structure information of classes and properties to create the propagation graph to find mappings between classes and properties.

Methods

Results

Conclusion