Tentative Schedule OL-2001

Online proceedings are available at CEUR Workshop Proceedings Vol-38.

9.00 - 10.40 Session I


9.00 Introduction


9.10-10.40 Full Papers

  • 9.10 - 9.40: N. Pernelle, M-C. Rousset, V. Ventos: Automatic Construction and Refinement fo Class Hierarchy over Semi-Structured Data.
  • 9.40 - 10.10: A. Delteil, C. Faron-Zucker, R. Dieng: Learning Ontologies from RDF annotations
  • 10.10 - 10.40: T. Yamaguchi: Acquiring Conceptual Relationships from Domain-Specific Texts

10.40 Coffee Break

11.00-12.45 Session II


11.00-12.00 Short/Position Papers

  • 11.00-11.20: C. Brewster, F.Ciravegna: Knowledge Acquisition for Knowledge Management
  • 11.20-11.40: M. Sintek, M. Junker, L. van Elst, A. Abecker: Position Statement FRODO
  • 11.40-12:00: M. Vargas-Vera, J. Domingue, Y.Kalfoglou, E. Motta, S. Buckingham-Shum: Template Driven Information Extraction for Populating Ontologies

12.00-12.30 Full Paper

  • 12.00-12.30: G. Heyer, U. Quasthoff, T. Wittig, C. Wolf: Learning Relations Using Collocations

12.30 - 13.50 Lunch Break

14.00 - 15.00 Invited Talk


14.00 - 15.00 Invited Talk: Alon Y. Halevy , University of Washington:


Learning to translate between ontologies

Schema matching is the problem of generating a semantic mapping between two distinct schemas. Schema matching is one of the longest standing problems in data management, but has recently gained more attention because of its relevance to several novel applications such as data integration from multiple sources, data warehousing, model management and peer-to-peer computing. In current data warehousing and data integration systems, mappings between schemas are largely constructed manually, resulting in a laborious and error-prone process, and therefore consitute a significant bottleneck in practice.

In this talk I will describe the LSD (Learning Source Descriptions) project which aims at building a generic schema matching solution. The key idea underlying LSD is that once a user has provided some manual mappings, the system should learn from these mappings to propose subsequent ones. Furthermore, no single learner is sufficient for this task because of the variety of information from which the system can learn. Hence, LSD employs a multi-strategy learning approach to schema mapping. In addition, LSD extends machine learning methods by incorporating knowledge about domain constraints (that are often available in data management applications), and exploiting the nesting structure in a schema whenever it is available. I will show experimental results on real-world data validating LSD's performance, and discuss our current work that gradually bridges the gap between schemas and ontologies.

This is joint work with Anhai Doan and Pedro Domingos.

15.00 - 15.30 Coffee Break

15.50 - 17.30 Session III

15.30 - 16.00 Full Paper

  • 15.30 - 16.00: R. Ichise, H. Takeda and S. Honiden: Rule Induction for Concept Hierarchy Alignment

16.00 - 17.40 Short/Position Paper

  • 16.00 - 16.20: R. Malyankar: Acquisition of Ontological Knowledge from Canonical Documents
  • 16.20 - 16.40: Borys Omelayenko: Learning Of Ontologies for The Web - The Analysis of Existent Approaches
  • 16.40 - 17.00: A. Termier, M. Sebag, M.-C. Rousset: Combining Statistics and Semantics for Word and Document Clustering.
  • 17.00 - 17.20: W.Koh, L. Mui: An information-theoretic approach for ontology-based interest matching
  • 17.20 - 17.40: Michael Minock: Constructing and Refining Ontologies within Explanation-Capable Agents

17.40 - 18.00 Closing Discussion