A multi-dimensional data organization for natural language processing

  • Authors:

  • Kam-Hoi Cheng;Waleed Faris

  • Affiliations:

  • (Correspd. Tel.: +1 713 743 3357/ Fax: +1 713 743 3335/ E-mail: khcheng@cs.uh.edu) Computer Science Department, University of Houston, Houston, Texas, 77204-3010, USA;Computer Science Department, University of Houston, Houston, Texas, 77204-3010, USA

  • Venue:
  • Journal of Computational Methods in Sciences and Engineering
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

For knowledge systems that rely on teachings from an outside source to gain their knowledge, proper data organizations are instrumental in managing and applying what has been learned. This paper describes the development of a Multi-Dimensional Data Organization (MDDO). Our MDDO stores a collection of knowledge of the same category classified by multiple orthogonal dimensions. Each dimension has a number of indices reflecting the potential attribute values of that dimension. A knowledge object is identified by a unique combination of indices from multiple dimensions whose values signify the different properties of the object itself. Our solution allows the collection of knowledge to be learned dynamically in any order and in an incremental manner. This MDDO has been used to store part of the English grammar, and we describe in this paper how it assists a learning program in both the parsing and the production of an English sentence. We also define the functionalities of the data organization, describe its organization, describe how each function is implemented, and analyze their worst-case performances.