Actively Learning Ontology Matching vis User Interaction

Ontology matching plays a key role for semantic interoperability. Many methods have been proposed for automatically finding the alignment between heterogeneous ontologies. Traditional methods mainly focus on how to accurately measure the similarity between elements (e.g., concepts and properties) of the two ontologies. However, in many real-world applications, finding the matching in a completely automatic way is infeasible. Ideally, it is desirable to take advantage of a few user interactions (feedbacks) to guide the automatic algorithms. Fundamentally, we need to answer the following questions: how many interactions are sufficient for finding a matching of high accuracy? Can we actively select what kinds of feedbacks are really necessary for improving the matching performance? To address these questions,We propose an active learning framework for ontology matching, which tries to find the most informative candidate matches to query the user. The user’s feedbacks are used to: 1) correct the mistake matching and 2) propagate the supervised information to guide the entire matching process. Measures are proposed to estimate the infirmity of each matching candidate. A correct propagation algorithm is further proposed to maximize the spread of the user’s “guidance”. Experimental results on several public data sets show that the proposed approach can significantly improve the matching accuracy (about 8% better than the baseline methods).

By: Feng Shi; Juanzi Li; Jie Tang; Guotong Xie; Hanyu Li

Published in: Lecture Notes in Computer Science, volume 5823, (no ), pages 585-600 in 2009

Please obtain a copy of this paper from your local library. IBM cannot distribute this paper externally.

Questions about this service can be mailed to .