The Language of the Web
Semantics expert Professor David Crystal OBE has been analysing linguistics since the 1960s. He has spent the last 12 years applying his studies of semantics to good effect on the Internet. He spoke to New Media Knowledge’s Chris Lee about his career with words.
Professor David Crystal has spent a life time in linguistics, from lecturing at University College London and Reading University to editing the Cambridge Encyclopaedia of Language. He received an OBE in 1995 for services to the English language and since then he has dedicated his time researching how semantics can be effective in Internet search and other digital platforms. He has published several books on the subject, including Language and the Internet and Txtng: The Gr8 Deb8.
In 1996 Crystal took his experience in using internal classification systems he had developed at the Cambridge Encyclopaedia of Language and applied them to the growing Internet search market.
A visionary Dutch firm called AND saw that we had a classification system which could be used as a way to make online search more accurate, which we found to be incredibly effective against the main search engines of the day,” he explained. “For example, a simple search for ‘depression’ could throw up the economic, psychotherapy, weather or geographical meaning. Our system of applying word by word and sense by sense association in the English language and assigning a category meant we could create a question and ask the Web user what type of ‘depression’ they were searching for.”
The patented system took three years and a 40-strong team of lexicographers to create. A kind of word ‘net’ was developed to capture pages relating to the category entered into the search.
“If I can predict words you’re going to use then I can search more accurately,” Crystal said.
In 2000 AND became a victim of the dotcom collapse and Crystal teamed up with his colleague Ian Saunders to found Crystal Reference Systems, having bought the rights to the technology from AND, rather than see his years of research go to waste.
“We had several options of where to apply our system then - automated document classification, search engine assistance, e-commerce search, Internet security or contextual advertising,” said Crystal. Crystal and Saunders settled on contextual advertising.
“You had situations occurring where inappropriate adverts would appear on the Internet. For example, an ad for knives would appear next to news of a Chicago stabbing because the stupid search engine would find the key word ‘knife’ and put the two together,” he explained. “It wouldn’t happen with semantics. The system would analyse all the words on a page which would link to an ad inventory and works out the appropriate ad.”
Add Pepper for Flavour
By 2005 Crystal and Saunders were unable to continue to develop on their own and Crystal Reference Systems was bought by Internet advertising network, AdPepper, the following year. Crystal now focuses on developing the company’s contextual advertising systems to help advertisers deal with sensitive issues on the Web, such as drugs, pornography, gambling and swearing, content with which many advertisers are keen for their brand to avoid.
Crystal has just researched the language of text messaging and sees social networking as an increasingly interesting field of study.
“Internet search is still very much in its infancy. All areas of the Internet are crying out for analysis, presenting new challenges and saying; ‘come and study us and see what we’re doing,’” he concluded. “Semantics has to keep up to date. Take Facebook, for example. Kids use slang, which moves very fast. We have to analyse and re-analyse to keep up.”