Use of the Normalized Word Vector Approach in Document Classification for an LKMC

Abstract

In order to realize the objective of expanding library services to provide knowledge management support for small businesses, a series of requirements must be met. This particular phase of a larger research project focuses on one of the requirements: the need for a document classification system to rapidly determine the content of digital documents. Document classification techniques are examined to assess the available alternatives for realization of Library Knowledge Management Centers (LKMCs). After evaluating prominent techniques the authors opted to investigate a less well-known method, the Normalized Word Vector (NWV) approach, which has been used successfully in classifying highly unstructured documents, i.e., student essays. The authors propose utilizing the NWV approach for LKMC automatic document classification with the goal of developing a system whereby unfamiliar documents can be quickly classified into existing topic categories. This conceptual paper will outline an approach to test NWV's suitability in this area.

Automatic Classification of Research Documents using Textual Entailment

“…The idea here is to calculate the distance between a document of interest against other documents based on their contextual meaning as opposed to ordinary structural similarity observed during lexical analysis. This component observes the entailment level found in the structured text-hypothesis using word vector approach [9]. The approach utilizes geometrical primitives as directed lines built from points to represent words in a sentence.…”