Character-level Error Analysis for Text Entry
Speaker
Scott MacKenzie
Associate Professor, Department of Computer Science and Engineering, York University in Toronto
When
-
Where
Newell-Simon Hall 1305 (Michael Mauldin Auditorium)
Description
This talk describes and presents a technique to analyze character-level errors in evaluations of text entry methods. The technique uses an algorithm for sequence comparisons to generate the set of optimal alignments between the presented and transcribed text. Per-character errors, categorized as insertions, substitutions, or deletions, are obtained by analyzing the alignments and applying a weighting factor. A detailed example using a real data set is given. Software tools to facilitate the analysis are also demonstrated.
Speaker's Bio
Scott MacKenzie is a computer scientist specializing in human-computer interaction (especially human input to computing systems and human performance measurement and modeling). He is an Associate Professor in the Department of Computer Science and Engineering at York University in Toronto. MacKenzie’s research interests include human performance measurement, prediction, and modeling for human-computer interaction. Applying these to new input devices and interactive techniques for advanced computing is the primary focus of his research. He is best known for his extensive work with Fitts’ law, as a psychomotor model for human movement that is widely used in human-computer interaction. This work has recently been incorporated into an ISO standard for evaluation computer pointing devices (ISO9241-9). More recently, MacKenzie has investigated multi-modal interfaces (e.g., with tactile and force feedback), multi-degree-of-freedom input devices (e.g., a two-ball mouse), and handwriting recognition and other alphanumeric entry techniques for mobile computing systems. The latter often employ a stylus, but may also use small, physical keyboards, one-handed keyboards, or soft keyboards.
Speaker's Website
http://www.yorku.ca/mack/
Host
Brad Myers