A key challenge for personalized mobile search is to tailor the answers to the specific user by considering her contextual situation. To adapt the retrieved items to user's context, this paper presents a preference-enabled querying mechanism for personalized mobile search. By exploiting the user's dialogue history, we infer the weighted user preferences and interests. To further compute personalized answers, we aim to continuously collect the ratings given by the user's friends regarding relevant topics from stream-based data sources such as Twitter. An experiment shows that our approach allows to compute the most relevant answers, providing an increased quality of search experience for the user.