Posted: Mon, January 17, 2005
The Future of Search Engines
Carrying out an online search can be a frustrating experience these days, often
resulting in an excessive amount of irrelevant, incoherent and
inaccurate data that must be trawled through before finding the information you need. Indeed recent research has shown that there are 186 million unsuccessful searches every day, and as the
internet continues to expand rapidly, this problem is getting worse.
Now a Holyhead based company with a background in linguistics has released a suite of software to help users make sense of searching.
Sali Earls spoke to Ian Saunders, Managing Director of Crystal Semantics to find out more.
Can you tell me something about the history of Crystal Semantics?
I founded Crystal Semantics with Professor David Crystal, a world authority in linguistics. During the 1980s, David worked with Cambridge University to develop lexicographical content, and developed
a classification scheme to manage data from disparate sources including NASA and the Natural History Museum. In the mid 1990s, the project was taken over by the Dutch company AND and work
moved to adapting the classification process to recent Internet developments.
Teams of linguists were employed to analyse every word in the English language, defining uses and mapping the results to David's classification scheme to see both where each word belonged,
and which other words could be used in association. AND moved out of the UK in 2001, and David and I kept the classification and reference business running in Holyhead to meet the demand for
reference content, and also to develop the commercial potential of our work.
Crystal Semantics was launched at the end of 2003 thanks to external investment from the Welsh Assembly Government and Finance Wales to bring in staff and resources. We now have a team
of 13, which has enabled us to bring our Textonomy products to market over the last few months
How can your products help to
make sense of searching?
When using search engines, people often have to adapt their search terms several times to find what they need. This is because current solutions are technology driven, so problems with searches are
often down to algorithms. We have developed a range of software approached from a linguistic perspective in order to enhance the use of search engines, thereby saving time.
Using our technology, search engines would not look at keywords when bringing back search results, rather at the context and concepts behind webpages. For example, if you currently look for
the word "depression", you will find millions of results covering subjects from mental health issues, to climate, economics and geography. The "sense engine" behind our products automatically
categorises the various senses of the word enabling the user to refine the search simply to only see results relating to the sense they mean. Our Textonomy products offer a quality search, and can
bring about the next generation of search engines with relevance and understanding.
Do your products only have applications for search engine companies?
No our products can also bring about benefits for ecommerce sites. Imagine you are looking for a mobile phone on an electronics site. Currently you may not find the item you are looking for, as the
description you type into the search box may differ to the company's catalogue description, as products can be described as mobile phones, mobiles, cellphones, cellular phones, etc. We can offer
solutions to online shops enabling them to have an index of terms to describe their products, and not miss any sales.
Our products can also work with online advertising, making sure they are targeted and relevant to the search terms applied. Currently a lot of online contextual advertising is based on particular
keywords. This method is fundamentally flawed as words can easily be taken out of context. The results can be disastrous for brands, for example advertisements selling knives appearing alongside a
news story reporting a stabbing! Our Textonomy Advance product analyses the content of both web pages and ads to ensure that both are appearing in the correct context. This enables marketers to
deliver the right message at the right time to their target audience.
Your products have been developed from a linguistic standpoint, but what technologies have you used for development?
We're an internet company, so we have used new and emerging standards including XML and other technology that has emerged within the last 12 to 18 months.
What's the next step for Crystal Semantics?
We launched Textonomy Advance, our contextual advertising product, in the UK and New York in November and we have spent time in Silicon Valley discussing the potential with search engine
companies. We have developed an enabling technology - we have no plans to create a new search engine ourselves, but are speaking to other organisations about using our solutions to enhance
their own products.
Over the next three years, we have plans in place to grow by an additional 41 staff to boost the business and develop our product suite. We will also be opening a US division to meet the needs
there.
For a company with a background in linguistics and reference, Crystal Semantics have come a long way in a short time. With plans in place to work with some of the world's largest internet
companies, and an aggressive expansion plan, this is a company going places, and we wish them luck with their plans.
In the future, when you're conducting a web search and you quickly find what you need, think back to this article and the Holyhead company that made your surfing a whole lot easier!
Find out more about Crystal Semantics at www.crystalsemantics.co.uk
|