Bridging Information-Seeking Human Gaze and Machine Reading Comprehension
This work addresses improving machine reading comprehension by incorporating human-like gaze strategies, offering incremental gains for natural language processing applications.
The study analyzed human gaze patterns during reading comprehension tasks and found increased fixation times on question-relevant text parts. By mimicking this information-seeking behavior, they improved a state-of-the-art model's performance on multiple-choice question answering in English.
In this work, we analyze how human gaze during reading comprehension is conditioned on the given reading comprehension question, and whether this signal can be beneficial for machine reading comprehension. To this end, we collect a new eye-tracking dataset with a large number of participants engaging in a multiple choice reading comprehension task. Our analysis of this data reveals increased fixation times over parts of the text that are most relevant for answering the question. Motivated by this finding, we propose making automated reading comprehension more human-like by mimicking human information-seeking reading behavior during reading comprehension. We demonstrate that this approach leads to performance gains on multiple choice question answering in English for a state-of-the-art reading comprehension model.