Type Classification of Semi-Structured Documents
Markus Tresch, Neal A. Palmer, Allen Luniewski · Very Large Data Bases · 1995
Semi-structured documents (e.g. journal art,icles, electronic mail, television programs, mail order catalogs, . ..) a.re often not explicitly typed; the only available t,ype information is the implicit structure. An explicit t,ype, however, is needed in order to a.pply objectoriented technology, like type-specific methods. In this paper, we present a.n experimental vector space cla.ssifier for determining the type of semi-structured documents. Our goal was to design a. high-performa.nce classifier in t,erms of accuracy (recall and precision), speed, and extensibility.