Last week I read a paper titled “Annotating and Searching Web Tables Using Entities, Types and Relationships” from the Machine Learning / IR group in IIT-Mumbai. The paper focuses on achieving the exact same sub-tasks which I am working on for interpreting and representing tables as linked data. The work at IIT-Mumbai by Limaye et al. also assigns a type to every column header, links the table cells to entities and identifies a binary relation between table columns. The goal of their work, however is to improve search and answering search queries over webtables. Whereas the focus of our work at Ebiquity is generating a linked data representation of a table and learning new facts from tables to be added to the Semantic Web.
It’s heartening to know that focus on webtables as source of knowledge is increasing by the day. After knowing about the details of the work, my adviser said “Great minds think alike !”
Their paper will appear in the proceedings of the 36th VLDB conference.
Limaye, G., Sarawagi, S., Chakrabarti, S.: Annotating and searching web tables using entities, types and relationships. In: Proc. of the 36th Int’l Conference on Very Large Databases (VLDB). (2010)