Interest Indicators in Structured Scientific Articles

Interaction with the systems has been focus of study since many years to devise the interest of user. This information can be used to implement implicit relevance feedback and to provide personalized suggestion in recommender systems. The aim of this article is to investigate the interaction patterns that can be predictors of users' interest in the context of structured documents. The investigated predictors include time spent on a page, clicks to navigate within the document, query and result presentation overlap. Descriptive statistical and machine learning techniques are used to find the relationship between searcher's interest and their explicit given feedback. The results indicate that reading time is predictive of user interest at document level.