Archive for December, 2009

Why Venture Capital Needs to Pay Close Attention to the Linguistic Web

December 9, 2009

It’s not about investing in such or another company.  It’s not about investing in Natural Language Processing startups. As soon as the Linguistic Web comes into existence, all software businesses will have to adjust.  Therefore, the Venture Capital industry should  be fully aware of this highly anticipated development in the immediate future.  The success of every new Internet startup may depend on its ability to integrate with the nascent Linguistic Web.

The moment one of the serious linguistic technologies, such as Powerset, would become freely available (for example, by being included in a browser), the reality of the web would be instantly changed. Linguistic Parsers belong in a browser, not in proprietary data-mining services.

Alternatively, as long as all the powerful linguistic technologies are kept proprietary, any startup that wants to develop natural language applications has to start from the very beginning and develop their own language processing system. Presently available public domain systems are simply not strong enough for practical applications. Building a Natural Language platform is clearly way too big a challenge for the typical application oriented startup company. 

How likely is it that one of the linguistic platform developers will make their advanced technologies freely available for everyone to use?   Not very.  Likewise, for some reason, the browser companies apparently are not rushing into including linguistic technology.  Does this mean that there is nothing to be done?  Not at all. 

There are a few companies concentrating on the building of sophisticated linguistic systems, and they may decide to make the technology available in one of the formats such as JavaScript, Silverlight, Java Applet, or Flash.

Of course, any serious linguistic platform must have a solid, scienticfic basis. We’re not talking here about linguistically amatuer experiments by developers who typically are not familiar with contemporary theoretical linguistics beyond the level achieved, at best, in the early seventies. Any project which is not implementing the most recent scientific advancements in theoretical linguistics would simply be totally inadequate. This would be comparable to trying to reach the moon by practicing your high-jump skills. You may be the most awesome athlete with the best trainers, motivation, and steroids, but ya ain’t gonna make it to the moon.

Since the necessary technology exists, the big switch to Natural Language Interface is just a matter of choice.