Integrating neurophysiologic relevance feedback in intent modeling for information retrieval.

Area of Science:

• Neuroscience
• Information Science
• Human-Computer Interaction

Background:

• Implicit relevance feedback offers potential for effortless information retrieval.
• Challenges include uncertainty from noisy neurophysiological signals and data representation.

Purpose of the Study:

• To develop and evaluate an integrated information retrieval system using online implicit relevance feedback from electroencephalography (EEG) and eye movements.
• To demonstrate the integration of noisy, implicit neurophysiological feedback with explicit feedback in interactive intent modeling.

Main Methods:

• Development of a fully integrated information retrieval system.
• Utilizing online implicit relevance feedback from EEG and eye movements.
• Conducting an evaluation experiment with 16 participants on realistic search tasks.

Main Results:

• The system successfully computed online neurophysiology-based relevance feedback.
• Performance was significantly better than chance in complex data domains and realistic search tasks.
• The approach demonstrated the generation of relevance feedback from brain and eye signals in a realistic scenario.

Conclusions:

• The developed system shows promise for neuroadaptive information retrieval (IR).
• Integration of noisy implicit feedback with explicit feedback is feasible for interactive intent modeling.
• Further research is needed to translate classification outcomes into measurable improvements in search task performance.