The Rise of Collective Intelligence
Transparent Concept Search Query Cloud
The demo application used to showcase PD semantic search architecture is called AcmeBrain. When a searcher uses AcmeBrain to search, their search begins when they enter query text, click the “search” button and receive back a QueryCloud. A QueryCloud is a semantically generated set of terms and phrases that are generated from a users search query. The example to the right is an actual QueryCloud generated from the search term "oncology." The query term was automatically sent the most relevant Brain in the Brain Library (in this case there was a Brain library comprised of 17 PD Brains) and that Brain generated a set of semantically related terms and phrases. This is done dynamically without the shackles of lexicons, synonym lists or thesauri. The terms and phrases in this QueryCloud may now be transformed into optimized searches for virtually any index internal to a company or on the broad web. Users can manipulate a QueryCloud by clicking on the words in the cloud to increase the font size (more relevant), decrease the font size (less relevant), delete a term or even add one that is not already in the cloud.
The steps in generating a QueryCloud are shown above. The user types in some text and sends it to one of the PD web servers. That server uses Java Remote Method Invocation (RMI) to send the text to a “Brain Library” and gets the terms and weights for the QueryCloud (seen on the left). Instead of returning the terms and weights directly, the server returns a redirect and the client then automatically gets the QueryCloud. By using a redirect, the URL is changed and the user can bookmark that URL to get back to the same QueryCloud.