SUMMARY OF IMAGED DOCUMENTS WITHOUT OCR

A system is presented for creating a summary indicating the contents o f an imaged document. The summary is composed from selected regions ex tracted from the imaged document. The regions may include sentences, k ey phrases, headings, and figures. The extracts are identified without the use of optical character recognition. The imaged document is firs t processed to identify the word-bounding boxes, the reading order of words, and the location of sentence and paragraph boundaries in the te xt. The word-bounding boxes are grouped into equivalence classes to mi mic the terms in a text document. Equivalence classes representing con tent words are identified, and key phrases are identified from the set of content words. Summary sentences are selected using a statisticall y based classifier applied to a set of discrete sentence features. Eva luation of sentence selection against a set of abstracts created by a professional abstracting company is given. (C) 1998 Academic Press.