Browse Prior Art Database

Intelligent Document Retrieval

IP.com Disclosure Number: IPCOM000120880D
Original Publication Date: 1991-Jun-01
Included in the Prior Art Database: 2005-Apr-02
Document File: 3 page(s) / 235K

Publishing Venue

IBM

Related People

Baber, RL: AUTHOR

Abstract

This article describes a research assistant tool for searching databases for documents relevant in subject and meaning to an initial document. (Image Omitted)

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Intelligent Document Retrieval

      This article describes a research assistant tool for
searching databases for documents relevant in subject and meaning to
an initial document.

                            (Image Omitted)

      An executive attending a scheduled meeting may have need of
documents relating to the agenda to be discussed. There is normally
at least one document in hand describing the topic; this may be
nothing more than a meeting notice, or may be a lengthy research
article.  From this original document the executive or his secretary
must retrieve related documents for study or presentation. Retrieving
documents related to a particular topic requires knowledge in several
domains:
      1) Understanding of the original document.
      2) Knowledge of related topics.
      3) Knowledge of all available related documents.
      4) Knowledge of access methods such as database
        queries.

      Information overload prevents the complete reading of lengthy
documents in preparation for a scheduled meeting or conference.

      Using intelligent document analysis tools it is possible to
automatically retrieve documents relating to a particular topic or
topics.  The Artificially Intelligent Document Analyzer (AIDA)
developed by the Computer Power Group of Garran, Australia provides
intelligent text parsing with four outputs:
      1) A document summary.
      2) A list of key phrases.
      3) A list of key words.
      4) A list of section titles.

      The entries in the list of key words are ranked according to
their importance to the central topic of the original document.
Thus, by running the original document through this tool, we have a
list of words relating more or less to the subject at hand, with the
most relevant words at the top of the list.

      There is currently a plan to use the SearchVision product to
search for documents related to certain key words.  This is done by
searching a list (compiled in advance) of all words of all documents
in a database. Connectives, such as and, or, the, etc., have been
removed from the list.  Associated with each word in this list are...