Skip to ContentSkip to Navigation
About us Latest news News News articles

Breakthrough in searching ‘unreadable’ texts

21 December 2011

Researchers from the University of Groningen have developed a fast technique to filter for relevant information. It can be applied to pieces of text that were not by any means intended to be read from start to finish. The new method now makes it possible to scan barely structured texts, loose notes and documents intended for internal use. This form of artificial intelligence was devised by Ashwin Ittoo. He expects his algorithms to find a wide application, even in search machines such as Google. He will gain his PhD on 5 January 2012 at the Faculty of Economics and Business.
For a computer it is extremely easy to find words or combinations of words in large amounts of text. However, getting a computer to understand a text and distil it into a conclusion is a technique that is still very much under development. This technique requires what is known as ‘Natural Language Processing’ (NLP), which is a branch of artificial intelligence. It was therefore quite a feat when, almost a year ago, IBM’s Watson supercomputer proved better at understanding and answering quiz questions than humans.


It is no coincidence that Ittoo, a computer scientist, has worked on techniques comparable with those used in the Watson project, because NLP formed the basis of this achievement. However, his algorithms make an enormous amount of extra information accessible. It was already possible to get a computer to understand readable texts such as news reports and academic articles, and to extract relevant information from them. However, this text analysis can now also be applied to informal notes written in telegram style that are full of spelling and grammatical errors, such as customer complaint forms or internal memos. Ittoo tested and developed his search algorithms at companies including Philips Health Care and Bang&Olufsen. Both of these companies now use prototypes of his search technique to refine their company information, regarding, for example, customer satisfaction.


Although the technique was developed at consumer electronics companies, it could have a much wider application, for example in medical files, financial documents, counterterrorism and search engines. Ittoo knows that a Google search often results in a huge amount of results, which creates the need to filter them accurately. He thinks his algorithms could be useful here. The refinement of the existing NLP techniques is not just an academic step forward but is also of commercial interest. In the United States alone, the market for this sort of text analysis is estimated at $ 835 m. Moreover, the number of users of the required software increased in 2010 by 25%. Ittoo is therefore aiming for a quick commercial application of his algorithms. They have been specifically developed for the English language but are applicable to any language as long as a good model exists for this.


Ashwin Ittoo (Mauritius, 1979) studied at the National University of Singapore and the Nanyang Technological University of Singapore, where his work included software development. He regularly publishes in international journals about applications of NLP techniques. Since 2008, he has conducted his PhD research at the Faculty of Economics and Business of the University of Groningen, where he still works as a researcher. The title of his thesis is ‘Natural language processing meets business. Algorithms for mining meaning from corporate texts.’ His supervisor is Prof. J.C. Wortmann. The thesis was partly made possible due to funding from the Ministry of Economic Affairs, Agriculture and Innovation.

Note for the press

Contact: Ashwin Ittoo, tel. 06-84952303 or 050-3637822; e-mail: r.a.ittoo

Last modified:13 March 2020 01.53 a.m.
View this page in: Nederlands

More news