This system will take questions from the user, parse them to the START database and with the returned data run a series of fractal searches based on the results of the START system.
A single search of "what time is it in florida?" can spawn up to 20 children based on the result. MILO will search for information about "time", "Florida" and from those results build a whole new dataset of questions that is then parsed to the Wikipedia, Google and other information sources. The longest running fractal search of MILO was ran on June 6'th 2010. The fractal search ran for over 19 hours and returned more than 3GB of data.
This demonstration by far is not fully complete nor fully compatible, and it was written to provide a public showing of how data that is mined from an outside source can be passed into “LIBBY” for vocal output based on 100% user input. The data is collected from the user, passed to the START database, the answer is passed back to MILO and then stripped of extra garbage then passed into the libby framework and prepared for the web browser.
MILO (the online version) is using the data source provided from the MIT START Natural Language Question Answering System that is maintained by the InfoLab group in the stata center.
Read the answer for "Where are you now? here " or try MILO out yourself! Click Here.