6. CONCLUSIONS
This paper describes the design and development of a scalable system underlying the election video search application that allows end users to search and navigate through the YouTube video material surrounding the 2008 US presidential election. It makes an inventory of videos, maintained by the election candidates, more accessible by use of automatic speech recognition, information retrieval and user interface technologies. Although these technologies are imperfect, the system provides a utility to the end user in that it allows the user to do content-based searches and navigation which, particularly for the dense long-form content, makes the information more readily available. The choice of consistently building the system on Google infrastructure allows us to scale this application to much larger corpora in the future.