ICRA
                   INSTITUTE FOR COMPUTER-ASSISTED RESEARCH IN ASTRONOMY

SUMMARIZER for Documents (1998)

Documents usually require a lot of time to read (offers, reports, descriptions, contracts, etc.). Same applies to a search within a database. These documents require a short and precise format with the content sorted by importance in order to be able to plan decisions. Such a summary is established by someone reading the document, taking notes about what is important. It takes a lot of time, is expensive and contains the danger of a personal interpretation by the reader: he produces his views on the subject.

The module Summarizer can create such an automatic conclusion of relevant content in a document established by linguistic elements. The requirement is in the form of a structured or unstructured text by full text search or combined with relational attributes. Hits are reported as documents, Summarizer can provide a summary upon the inputs of the user:

      - Language – German, English and French

      - Presentation as a file and/or display on screen

      - Number or percentage of subjects

      - Number or percentage of relevant sentences.

The retrieved words and sentences are ordered according to their relevance in the original document shown in a window next to the original. In most cases the result can be read as a well formulated summary. Clicking on a word shows its place inside the original document.

Restrictions: poems usually do not deliver good results in the Summarizer.