Web searches tend to be short and ambiguous. It is therefore not surprising that Web query disambiguation is an actively researched topic. However, most existing work relies on the existence of search engine log data in which each user's search activities are recorded over long periods of time. Such approaches may raise privacy concerns and may be difficult to implement for pragmatic reasons. In this work, we present an approach to Web query disambiguation that bases its predictions only on a short glimpse of user search activity, captured in a brief session of about 5--6 previous searches on average. Our method exploits the relations of the current search session in which the ambiguous query is issued to previous sessions in order to predict the user's intentions and is based on Markov logic. We present empirical results that demonstrate the effectiveness of our proposed approach on data collected form a commercial general-purpose search engine.