Nowadays more and more people access the Internet with their accompanying mobile devices. Performing web search activities anytime and anywhere thus becomes common. As compared to the ones sitting in front of their PCs, mobile users often search for something which is closely relevant to his or her current context, e.g., location and time, or of specific requirements. We define this kind of web searches as the on-demand search in this work. In view of the possible difficulties encountered in practice, we propose to introduce both browsing history and localized information when handling on-demand searches of a mobile user. Consequently, user demands can be better understood so that more personalized results can be integrated, delivered and visualized. On the other hand, features of social interaction are also addressed. For example, nearby friends could be informed and invited to join activities of common interests. Better user experience can thus be achieved among members of a mobile community.