Discrimination of Neutral Postures in Computer Based Work
Abstract
A biometric system recognizes users based on the way they physically interact with the system. In this work, we discover a common behavior that a typist consistently displays in non-trivial computer work. We sought to demonstrate three objectives: first, compelling proof that a user can be actively recognized over the course of a lengthy task via a neutral posture struck multiple times in that task; two, a sensing concept for capturing the neutral posture, and, third, an objective method for determine the level of work performed by each typist. This thesis develops a model for hand tracking using a simple ellipse to describe the neutral posture where a typist pauses before typing. Initial results of a group of 10 users indicate that the neutral posture can be established with only a few seconds of training data and can perform with approximately 92.1% accuracy. Analysis of the typed text determined the complexity of the typists' work using Bloom's Taxonomy - a taxonomy based on verb usage; parsed verb phrases indicated the level of competency that the users endeavored to demonstrate. This competency or expertise may further distinguish users and their performance in their most engaging work.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2013
- Accession Number
- ADA582542
Entities
People
- Alanna R. Keith
Organizations
- Air Force Institute of Technology