This weekend, John Markoff of the Times penned an interesting article on the potential of the Semantic Web (aka “Web 3.0”).
Where Web 2.0 has introduced social applications and collaboration, the so-called Web 3.0 is focused on providing context and meaning to content. The Web 3.0 world Markoff describes will act more as a guide than a simple navigation system, providing answers to natural language searches.
The article describes the current debate over whether the Semantic Web will require a new infrastructure, or whether it will emerge organically, on top of existing web content. Each side presents a huge challenge. The organic growth, based upon tagging concepts like Delicious, Flickr and Digg, supposes that users will tag content making its meaning accessible to others. That works to some extent, but I’m not sure how well that model scales. Flickr has proven that we can get users to tag limited content on a large scale, but only because that tagging makes it easier to organize photos, something they need to do for their own use. Delicious has gained acclaim, but still has only a million users and many of them have only a handful of bookmarks.
It will be difficult to get users to tag content on a widespread basis. Even bloggers, seeking to get their content read by the masses, infrequently tag their content for Technorati or other search engines.
The other opinion is that we will use technology to read and understand content, automatically generating the metadata to make content understandable. While the future holds great promise for this, today’s semantic tagging technologies have their own limitations. Technology providers often point to defense/intelligence community initiatives as laying the groundwork for the semantic web. Having spent a good deal of time in that space, I can tell you that the hype typically exceeds the results. These technologies are highly effective when working with homogenous forms of content. A rules-based engine can easily be trained to understand concepts in, for example, FBI field incident reports. Apply that same engine to informal message traffic or blogs, however, and its accuracy drops way down.
I think that the Semantic Web will evolve over time; it won’t be a massive new framework, but instead will be a combination of organic (“collective intelligence”) and technology solutions. The solutions won’t try to “understand the web” as a whole, but instead will provide solutions for specific tasks or sectors. In the meantime, vendors should try to avoid over-hyping these capabilities, the way that artificial intelligence was hyped during the 90’s. Let’s try to deliver small, targeted solutions, then build upon those. Then, at some point, we can look around and say “oh, I guess all these cool new apps are what we meant by the Semantic Web”. In the meantime, I wouldn't be so quick to give it the Web 3.0 label.
Posted by: |