Searching for meaning, not just words

Founded by Francisco Webber in 2011, applies artificial intelligence to develop language processing tools.

Technologist: What is semantic analysis?

Francisco Webber: It’s a method to interpret queries from web users that focuses on meaning rather than the words themselves. Current search engines give you results based on key words. The most advanced versions use word associations: someone who types “flight” is probably looking for a plane ticket. But a human has to code all these suggestions beforehand. With semantic analysis, machines will autonomously understand what someone is searching for.

T. How does it work?

F.W. We’ve developed a solution that translates every word entered in a search engine into a semantic footprint, or a sequence that can find all the documents containing words with a similar semantic footprint. For example, the semantic footprint of the term “agreement reached” is very similar to the semantic footprint of “signed contract” because their meanings are the same, even if the words are different.

T. What can this be used for?

F.W. Before, if a company was doing a market study and wanted to filter Twitter messages to focus only on those relating to smartphones, it would have to make a long list of key words. But searching for the semantic footprint of the word “smartphone” is much faster and more effective. Another example: using the semantic footprint of the word “liberal politics” can provide a personalised newsfeed.