START Conference Manager    

Predicting Task Difficulty for Different Task Types

Jingjing Liu, Jacek Gwizdka, Chang Liu and Nick Belkin

(Submission #173)


Abstract

This paper reports our investigations of how user behaviors differ between difficult and easy tasks, as well as how this difference varies in different types of tasks. We also report whether and how behavioral variables that can predict task difficulty vary across task types. We also explored how whole task-level user behaviors and task section-level behaviors differ in task difficulty prediction. Data were collected in a controlled lab experiment with 48 participants, each completing 6 search tasks of three types: single-fact finding, multiple-fact finding and multiple-piece information gathering. Results show that task type affects the relationships between task difficulty and user behaviors and that predicting task difficulty should take account of task type. Results also show that both task-level and section-level user behaviors can serve as task difficulty predictors. Task-level variables present higher prediction accuracy but section-level factors can be used in real-time prediction. These findings can help search systems predict task difficulty and accordingly adapt search for users.

Categories

Program Track:  Track 3 - Information Systems, Interactivity and Design
Submission Type:  Research Paper

START Conference Manager (V2.56.8 - Rev. 1261)