RIT Tangent Takes Lead in Formula Search Engine Racing
- By Dian Schaffhauser
- 02/05/15
The results of search engine work begun at the Rochester Institute of Technology (RIT) in New York recently swept away the competition at an information access technologies conference in Tokyo. Whereas most search engines use only text, the RIT version specifically allows users to seek out results through text as well as mathematical formulas.
Computer science graduates David Stalnaker and Nidhin Pattaniyil created Tangent, which allows users to enter expressions using images, a stylus, a finger, a mouse and the keyboard. Users can also enter a search with LaTeX, a scientific document markup language.
Tangent lets math experts and scientists search for the math in technical documents; however, the developers are hoping that it will promote math literacy by helping non-experts too. Eventually, the engine may be adapted to retrieving other non-text-based documents, including chemical diagrams, tables and figures.
"For many people, visual elements are the anchor for understanding how to organize things, especially with math," said Richard Zanibbi, associate professor of computer science in RIT's College of Computing and Information Sciences. "We can't just rely on text-based math; we need an intuitive search engine for visual math."
The search engine grew out of a project call min, which allows math expressions to be drawn on a canvas, converted to text and combined with keywords, and then sent to a search engine such as Google or Wikipedia. However, the developers found that the results from those engines weren't always relevant.
As part of a master's project, Stalnaker, who now works at Google, began developing a new kind of search engine, which would index what symbols look like visually as they appear in formulas. The text retrieval aspects were built on top of an Apache open source search engine platform called Solr and designed to index multiple collections, including databases of science articles and Wikipedia.
Pattaniyil, who now works at Comcast, extended Stalnaker's work for his master's thesis by adding the indexing and retrieval of text and expanding the query capability to support multiple formulas and text.
The merit of their efforts proved itself when Tangent was tested against other search engines during a recent NII Testbeds and Community for Information access Research conference in Japan. Tangent produced the best hits for combined text and math queries; nine out of 10 queries were found to be relevant.
Work on the search engine continues with a new group of student developers at RIT as well as the University of Waterloo.
About the Author
Dian Schaffhauser is a former senior contributing editor for 1105 Media's education publications THE Journal, Campus Technology and Spaces4Learning.