Indexing Dataspaces

摘要

Dataspaces are collections of heterogeneous and partially unstructured data. Unlike data-integration systems that also offer uniform access to heterogeneous data sources, datas- paces do not assume that all the semantic relationships be- tween sources are known and specified. Much of the user interaction with dataspaces involves exploring the data, and users do not have a single schema to which they can pose queries. Consequently, it is important that queries are al- lowed to specify varying degrees of structure, spanning key- word queries to more structure-aware queries. This paper considers indexing support for queries that combine keywords and structure. We describe several exten- sions to inverted lists to capture structure when it is present. In particular, our extensions incorporate attribute labels, relationships between data items, hierarchies of schema ele- ments, and synonyms among schema elements. We describe experiments showing that our indexing techniques improve query effciency by an order of magnitude compared with alternative approaches, and scale well with the size of the data.