LORE: A Lightweight Object Repository for Semistructured Data

The number of information sources accessible electronically is growing rapidly. Many of these sources store and export unstructured data in addition to or instead of structured data. In most cases, however, the unstructured data is not entirely devoid of structure, i.e., the data is semistructured. We consider data to be semistructured when there is no schema xed or known in advance and when the data may be incomplete or irregular. For example, HTML les on the World-Wide Web usually contain some structure, but often the data is irregular or incomplete. In addition, data integrated from multiple, heterogeneous information sources often is semistructured. Storing and querying semistructured data poses considerably di erent problems and requirements tha