 | ||
Profiles | ||
Zanran sept 4 2015
Zanran is a search engine for data and statistics. Zanran’s focus on finding graphs, charts and tables on the Internet, which distinguishes it from other search engines (Google, Bing, etc.). Unlike a normal search engine, the search results – graphs, tables, etc. - can be previewed by mouse-hovering over the thumbnails.
Contents
History
In 2006, the founders, Dr Yves Dassas and Jon Goldhill, started developing the technology that makes Zanran possible. A limited beta ran from November 2010. The service was launched as a public beta version in May 2011, to broadly warm reviews.
Technology
Zanran has developed two technologies specifically for this application:
- Image ‘classification’ – the ability for a computer to decide whether an image is a graph, a pie-chart etc. (as opposed to a photograph or a cartoon etc.). The Zanran algorithms work to over 95% accuracy. This is important because most images on the web are not graphs – and otherwise there would be a large number of false positives.
- Text extraction – the process of taking the most appropriate text to describe the graph. This contrasts with a normal search engine where an entire HTML page might be included.
These processes are the subject of Zanran’s UK patent. The image processing in particular takes a great deal of computer power. Zanran runs on the Amazon cloud – and uses hundreds of machines at a time.
The service is English-language only (at Dec 2011).
The company
Zanran Ltd is based in London, UK. It was financed by the founders prior to a private angel investment round in March 2010.
Other data-search on the Internet
Other specialists in the data-search space include: WolframAlpha, Infochimps, Datafiniti, DataMarket, Junar, and Timetric.