[dbjapan] 講演会：視線計測を用いた検索タスク推定と認知負荷, July 19 13:30-15:00, Room2009

[Noriko Kando <nkando [at] nii.ac.jp>]

日本データベース学会の皆さま


NIIの神門と申します。
はなはだ急なお知らせで恐縮ですが、明日7月19日13:30より
NIIにて、下記の講演会を行います。

Title:
Using Eye Movements to Infer Search Task Properties, Task
Difficulty, and Cognitive Effort During Information Search

Speaker:  Michael J. Cole, Rutgers University

Date:  Tuesday, July 19th, at 13:30-15:00

Place:  Room 2009, 20F, NII

どなたでも、ご関心のある方々の参加を歓迎いたします。
また、周囲の方々にもご紹介いただければ幸いです。

よろしくお願いいたします

神門 典子
================================================

Dear all,

You are all cordially invited to the following talk by
Dr. Michael J. Cole, Rutgers University.

Title:
Using Eye Movements to Infer Search Task Properties, Task
Difficulty, and Cognitive Effort During Information Search

Speaker:  Michael J. Cole, Rutgers University

Date:  Tuesday, July 19th, at 13:30-15:00

Place:  Room 2009, 20F, NII


Abstract:

The POODLE project(*) has been working to develop personalized 
information retrieval systems by inferring aspects of the user’s 
situation as they engage in information search properties. A series of 
user studies have investigated the influence of task type, level of user 
knowledge, and individual differences, especially cognitive abilities, 
on observable behaviors with the goal of building user models that can 
be used to predict user and task properties during the search session. A 
multiple source logging systems was used that captured high level 
behaviors, such as document use, along with low-level activities 
including mouse movement and eye movements.

Eye tracking has already been used by several groups to investigate 
information retrieval problems, such as attention during SERP link 
selection processes and detection of relevance. Eye movement is 
cognitively-controlled and it provides direct evidence for user 
cognitive processing states. There is a strong foundation of research in 
cognitive science regarding cognitive processing and eye movements. With 
a user-centered perspective on information search and
retrieval, eye movement analysis is a new research frontier to study 
information search. It has the promise of making direct measurement of 
user mental states during search to infer task and user properties that 
can provide a basis for personalization of the ongoing interaction with
the search system.

We have developed a new methodology for analysis of eye movement 
patterns based on models of the reading process. I’ll describe the 
methodology and algorithmn and present recent published work showing 
users change low-level information acquisition strategies in ways that
reflect properties of their high-level task. Our most recent work has 
extended the methodology to measure some aspects of the amount of 
cognitive processing by a user while acquiring information
during realistic IR work tasks in the journalism domain. The results 
show the cognitive effort measures are well correlated with task 
properties that contribute to task difficulty. They also match with the 
participant’s assessment of the actual difficulty of their task.

One attractive element of our methodology is that it can be applied to 
eye tracking data in any (textual) information setting and used to give 
real-time estimates of cognitive effort, and rate of information 
acquisition. We are using these measures to explore relationships with 
query formulation and document use, and to identify sub-task segments 
and task stages. I’ll discuss the potential to extend our cognitive 
effort modeling approach to make direct measurement of user domain 
knowledge and recognize when learning takes place during search.

(*: http://comminfo.rutgers.edu/imls/poodle)


Bio and acknowledgments:

Michael Cole is a PhD. candidate at Rutgers University in the School of 
Communication and Information. His advisor is Nick Belkin. He received 
his BS (honors) in physics from the University of Wisconsin - Madison 
and studied philosophy of physics at the University of Michigan for two 
years before entering the business world and focusing on technology 
strategy development. His dissertation work concerns automated learning 
of search task structure.

He has worked in the POoDLE project1 for three years and this work grows 
out of that effort. It owes much to the insights of the PIs (Profs. 
Belkin, Jacek Gwizdka, and Xiangmin Zhang) and the hard work of the 
Poodlers, especially Ralf Bierig, Jingjing Liu, and Chang Liu. Amongst 
his active current research is a Google-funded project to investigate 
the implicit detection of relevance judgments and the effect of affect 
on relevance judgments using physiological measurement of eye movements, 
brain electrical activity, and skin resistance.


Selected recent publications:
References:

Bierig, R., Cole, M. J., Gwizdka, J., and Belkin, N. J. (2010). A data 
analysis and modelling framework for the evaluation of interactive 
information retrieval. In Gurrin, C. et al., editors, Proceedings of 
ECIR 2010, (Milton Keynes, UK 2010), volume 5993 of Lecture Notes in 
Computer Science, pages 673–674. Springer.

Cole, M. J. (2010). Simulation of the IIR User: Beyond the Automagic. In 
Azzopardi, L., Järvelin, K., Kamps, J., and Smucker, M. D., editors, 
Proceedings of the SIGIR 2010 Workshop on the Simulation of Interaction: 
Automated Evaluation of Interactive IR, Geneva, Switzerland. IR 
Publications.

Cole, M. J., Bierig, R., Gwizdka, J., and Belkin, N. J. (2009a). A 
system to model user interaction sequences. Demonstration at ASIST 2009 
(Vancouver, Canada).

Cole, M. J., Gwizdka, J., Bierig, R., Belkin, N. J., Liu, J., Liu, C., 
Zhang, J., and Zhang, X. (2010a). Linking search tasks with low-level 
eye movement patterns. In Proceedings of the 17th European Conference on 
Cognitive Ergonomics, Delf, The Netherlands.

Cole, M. J., Liu, J., Belkin, N. J., Bierig, R., Gwizdka, J., Liu, C., 
Zhang, J., and Zhang, X. (2009b). Usefulness as the criterion for 
evaluation of interactive information retrieval. In Proceedings of
the HCIR 2009 Workshop, Washington, DC.

Cole, M. J., Zhang, X., Liu, J., Liu, C., Belkin, N. J., Bierig, R., and 
Gwizdka, J. (2010b). Are self-assessments reliable indicators of topic 
knowledge? In Proceedings of ASIST 2010, pages 30:1–30:10, Pittsburgh, 
PA. American Society for Information Science and Technology.

Academic Host: Noriko Kando, Professor, NII
Inquiry: Email to kando-secr [at] nii.ac.jp (ex. 2733)
------------------------------------------------------------------------

Attachment: abstractBioHandout.pdf
Description: Adobe PDF document