作者: Stefano Mizzaro , Josiane Mothe
关键词:
摘要: Predicting if a query will be difficult for system is important to improve retrieval effectiveness by implementing specific processing. There have been several attempts predict difficulty, both automatically and manually; but without high accuracy at pre-retrieval stage. In this paper, we focus rather on understanding Why perceived humans as difficult. We ran two separated related experiments in which asked provide difficulty prediction reasons explain their prediction. Results show that: (i) can categorized into 4 classes; (ii) framed closed questions answered Likert scale; (iii) some correlate coherent way with the human predicted numerical difficulty. On basis of these results it possible derive hints provided help users when formulating queries avoid them rely wrong perception