FlashFind: Responsive search in large datasets on mobile devices

摘要

This paper introduces FlashFind, an on-device search engine for making the vast amounts of digital content stored on today's flash memory-enabled mobile devices (PMP's, PND's, e-readers, smart phones) instantly accessible through a single-widget search interface. Without network access, FlashFind provides a search experience comparable to that of standard web search engines. More than that, the search is fully incremental and prefix-based, i.e. results are shown shortly after each key press, and instead of entire words, the user mostly needs to type only a small number of characters until the desired result is found. The pilot application, implemented on Windows Mobile smart phones and on Linux, provides near-instantaneous global search-based selection among the streets, points of interest and cities in a map of Western Europe (-10 millions of entries) and is being integrated into a commercial navigation application. The patent-pending FlashFind technology is equally capable to search other coarsely-structured text-based or text-annotated spatial or non-spatial content, e.g. collections of e-books on an e-reader device. In contrast to server or desktop search solutions, FlashFind is optimized for the limitations of mobile devices in terms of CPU, main memory, and interaction constraints (e.g. having only a phone keypad). Ongoing work is targeted at fuzzy search, at new applications and device classes, and at integrating our search technology with other innovative input methods.