Querying the Data Web

30 Published by the IEEE Computer Society 1089-7801/10/$26.00 © 2010 IEEE IEEE INTERNET COMPUTING I n parallel with the hypertext Web’s continuous development, we’re witnessing an explosion of structured data published on the Web, with companies such as Google, Freebase, Upcoming, eBay, Yahoo, and Amazon competing to gather and publish this content in a way that encourages people to reuse it. The trend of publishing structured data on the Web is shifting the focus of Web technologies toward new paradigms of structured data retrieval. Traditional search engines can’t serve such data adequately because a keyword-based query will still be ambiguous, even though the underlying data is structured. To exploit the massive amount of structured data on the Web to its full potential, users should be able to easily query and fuse it. Unlike keyword-based information retrieval, the main challenge in structured data retrieval is that, before formulating a query, the user has to know the data’s structure and attribute labels (that is, the schema). However, in many cases, a schema can be dynamic, adding and dropping several kinds of items that have different attributes. Other data sources might be schema-free, or the schema might consist of assertions mixed up with the data. In RDF, for example, the data doesn’t have to be consistent with a certain schema or ontology, so the user has to manually investigate the RDF data itself before querying it to understand both vocabMashQL, a novel query formulation language for querying and mashing up