Collections > Master's Papers > School of Information and Library Science > Predicting the Difficulty of Trivia Questions Using Text Features

In numerous contexts, including community question answering systems, school exams, and trivia competitions, a need to assess the difficulty of questions arises. This study examines what features predict difficulty in the realm of trivia questions, considering features related to readability and the question's topic as potential contributors. Using clues from the game show Jeopardy!, the study finds that features relating to a trivia question's length, the inclusion of audiovisual media, and its constituent noun and verb phrases have a significant impact on the clue's difficulty. Based on these findings, this study proposes that finding more nuanced ways to depict the amount of information in a trivia question would lead to further advancements.