PeARS (Peer-to-peer Agent for Reciprocated Search) is a search engine that you can run locally from your browser. It allows you to 'index' pages (i.e. to produce a computer-readable representation of the pages' content, essential to the search process), and to search pages that you or your friends have indexed. Search happens entirely on your machine, meaning that no one knows what you are searching and when.
We are delighted to announce that thanks to the generosity of the NLnet Foundation, we have been able to launch a new project, dedicated to making the PeARS indexer smaller, more efficient, and more accurate.
The new project, codename Fruit Fly, is investigating a state-of-the-art algorithm derived from a biological model to bring Web indexing to the masses. Inspired by the olfactory system of the species Drosophila Melanogaster (the common fruit fly), the indexer is a shallow, partially connected neural network which transforms the text of Web documents into a compact, searchable representation. Given its tiny size and simplicity, a fruit fly model can be stored and run on any device with great efficiency. Using a genetic algorithm, we are currently looking for the fruit fly individuals that generate the best possible document representations.
Intrigued? Then read the whole story.
join now and start contributing