Context-Based Content Search On Mobile User Device

In recent years, more and more people have owned mobile terminals, with the growing popularity of mobile terminals, information accessed by people on mobile terminals has become increasingly diverse. How to allow users to easily and promptly search the content viewing on mobile devices or storage has become an increasingly important task. Current common content keywords-based search is largely limited by user’s memory about the content. Most of mobile devices today are equipped with diversified sensors. And these sensors can provide rich user context (e.g. location). Compared with the details of content, some kinds of user context is much easier to be remembered.The diversity and quantity of content and the limitation of people’s memory often make it difficult to remember the content keywords accurately. But people may remember the contextual information when accessing information, most of mobile terminals now are equipped with a diversity of sensors which can provide rich user contextual information (such as GPS positioning, mobile phone activities and etc.). Therefore, compared with common content keyword-based search method, context-based content search method for mobile devices can get a better user experience, and has been attracted more and more research attention.This paper presents a context-based unstructured query processing framework, referred to as UCQP (Unstructured Context Query Parser), the framework can support user context information on search the historical content. Describing contextual information, the user may enter the keyword combination or even a complete sentence, rather than simple keywords. So the query may be unstructured. Then in order to obtain the user’s intent to obtain accurate search results, it is necessary to analyze the user queries, and additional information in queries to understand the context entities. Based on this task, UCQP composed different types of query parsers for different types of context information. Furthermore, UCQP has a similarity calculation method to calculate the similarity of the context information in user’s query and the corresponding context information of historical content, which will rank multiple of search results. The experimental results show a good precision and a fast processing speed in the prototype system based on UCQP framework.