Exploiting Interaction Features in User Intent Understanding

Understanding user intent during a web navigation session is a challenging topic, which is drawing the attention of many researchers in this area. The achievements of such research goals would have a great impact on many internet-based applications. For instance, if a search engine had the capability of capturing user intents, it could better suite the order of search results to user needs. In this context, the research has mainly focused on the analysis of user interactions with Search Engine Result Pages (SERPs) resulting from a web query, but most methods ignore the behavior of the user during the exploration of web pages associated to the links of the SERP s/he decides to visit. In this paper we propose a novel model that analyzes user interactions on such pages, in addition to the information considered by other mentioned approaches. In particular, captured user interactions are translated into features that are part of the input of a classification algorithm aiming to determine user informational, navigational, and transactional intents. Experimental results highlight the effectiveness of the proposed model, showing how the additional analysis it performs on visited web pages contributes to enhance user intent understanding.