Schema-Independent Retrieval from Heterogeneous Structured Text

We present a query language for searching collections of structured text. Documents within the collection are not required to adhere to a global schema nor are individual documents required to be structured according to any defined schema at all. Nonetheless, queries may directly reference structure across differently formatted documents. We briefly discuss application of the language to multilingual collections, relational databases, text filtering and document analysis. 1 Introduction Figure 1, a facsimile of a page from an edition of Shakespeare's Tragedie of Macbeth, demonstrates the complexity of structured text. This single page includes stage directions, speakers, speeches, the start of the play's first act, its entire first scene and the start of its second. Marginal notations indicate the act, scene and line numbers. Italics, small capitals, and two font sizes are used. The lineation of the spoken verse is consistent with its scansion. A page number appears at the bottom of ..