Publications

On the 26th and 27th of April, the first European Text Analytics Summit took place in Amsterdam (www.textanalyticsnews.com/europe07/)

After two successful summits on this topic in the United States, the organizers decided it was time to bring the summit across the ocean. The conference attracted almost 100 visitors, of which about 80% were vendors and 20% end users. So, although the market for text mining is rapidly growing, it seems that it is still hard to convince large crowds of potential end users of its value.

However, some of the biggest vendors from both Europe and the U.S. attended to present what is currently state-of-the-art in the text mining business. Also, attendees learned that security is a hot topic in the EU for which a lot of funding is available. As a side effect, the text mining market is expected to continue growing. Some interesting real world applications were presented, especially from the pharmaceutical domain.

One of the reasons that text mining is booming business, is that companies simply can't do without any more. Gigabytes of new text come into existence every day: for example, it would take 5 years to read the scientific material that is produced every 24 hours. Clearly an impossible endeavor. Furthermore, in the discussion on the maturity of the technology, one of the speakers stated that a technology is mature enough if it brings business value. This is clearly already the case for this field and will only become more and more so.

As with many data mining and machine learning techniques, a very common application is forecasting the future. This can be useful in many domains, for example:

• reputation control (Can we predict what will be written about our company by journalists and bloggers? And if yes, can we influence this?);
• biomedical (Can we predict what would be a good drug?) and, of course,
• security (Can we foresee what security threats will occur when and where?).

Let me conclude with some guidelines for those not familiar with the text analytics business, but who would like start a venture in this direction.

First, have a good look around to find out what's already being done and find yourself a nice new niche. Second, adopt some technology that does the job for you and start building your application. The next point is important: your application should definitely be a full-blown web-based enterprise solution and it should look real slick. Keep it simple, it should just work: end users don't want to know about all the state-of-the-art stuff that's going on behind the scenes. Meanwhile, find yourself some big partners that are willing to support you, either with technology or as client. Once you have some good references and your application is ready for customization for clients, you're ready to roll!

What's this?