Browse Prior Art Database

Tree-Structured Word Lists for Natural Language Processing

IP.com Disclosure Number: IPCOM000036443D
Original Publication Date: 1989-Sep-01
Included in the Prior Art Database: 2005-Jan-29

Publishing Venue

IBM

Related People

Authors:
Maier, M Saviozzi, G [+details]

Abstract

In a word list (WL) the essential information, such as words of a specific natural language, is often generated by the application of morphological rules to single lemmas and then it is interpreted as bunches of words having the same stem completed by different suffixes. Such a concept is used for building a tree-structured word list (TWL) containing a significant amount of inflected words instead of lemmas, leaving the application of morphological rules to very special situations (enclitics, etc.). The application of morphological rules to any given word turns out to be excessively time consuming, even if it is an effective strategy from the space saving point of view. Moreover, the crossing of a TWL results to be flexible enough and particularly suitable for spelling correction strategies, i.e.