On January 10, 2007 at Cubberly Community College in Palo Alto, SDForum’s Emerging Technology SIGâ€™s AJ Chen and Peter Hallinan presented â€˜Shrinking The Meaningless Web for 2007â€™. Peter Rip of Crosslink Capital moderated a panel including Lew Tucker of Radar Networks, Giovanni Tummarello of Dbin, Alain Rappaport of Medstory, and Ramana Rao, founder of Inxight Software.
The Search for Meaning
It was the first time I had seen such a unique combination of venture capital, academic and entrepreneurial expertise talking about the search for meaning.
Sometimes called the semantic web, they discussed the new direction of the Internet, beyond standards or applications. Web 1.0 was the search for data. Web 2.0 is the search for information. Web 3.0 will be the search for meaning. Like in Douglas Adamâ€™s Hitchhikerâ€™s Guide, we may get an answer but have think about the question before we ask.
Rip started off describing the origins of the term semantic web in 1998 as a big idea that nobody knew quite what to do with. Searching the links to data should result in discovering new relationships and meaningful answers. Ontology, RSS, tagging and microformats are rudimentary ways of attaching meaning to data. Ideally it would be a machine-to-machine process creating a valuable experience. To do this, the data must have some sort of structure on which to build meaning. Examining social networks like MySpace gives better insight into how these structures work.
The best structures for businesses on the web will match a need to a solution. This is sometimes called a butterfly structure. Ebay and Google both match merchandise or queries for customers. Rip is looking for companies to be the building blocks for the semantic web rather than a complete solution. This step-by-step approach from the data web to the recombinant web could lead to the semantic web. It will start with enhancements to the existing apps, then development tools to create new apps, then exporting the apps to old things in a new way and then creating new apps doing totally new things.
Rip gave examples of companies he invests in that are on that road. Like.com helps users compare objects that are difficult to describe in text by extracting meaning from images. Vast.com crawls the web for the long tails and get attributes of entities like vacation homes and puts them searchable form. Pandora aggregates music rated by experts to find music you like. Teqlo.com uses web service APIs to make self-organizing experiences or super matches.
Tummarello came all the way from Italy to talk about to about semantics. He expects new mash-ups of socially created content that can do something useful. He demonstrated how a user could quickly create structured content and value at the same time without programming Java code. Customers drinking a special brand of beer should be able to find a brewery or pub close to their current location that is open. He called this a beer-to-beer network. To do this requires mash up or transport layers, trust or security layers. This RDF data or aggregation of information from different groups adds relevance to customers and increases its value. Cross posting a subject can add value.
Tucker responded to a question of how to get users to add structured data to something like Wikipedia. He said people are lazy and experts disagree on ontology. He thinks there may be wisdom in the masses. He talked about the Darwinian idea that if a trait does not get used, it goes away. Googlebase allows users to assign different attributes to cars. Some users will describe a car as having two doors, other as having four wheels. All the attributes create a larger description of a car. If most people think buying a car with wheels is important, the usage or traffic will determine that wheels are a more important attribute than doors. Frankly, Iâ€™d rather have both. Sheer numbers of people trying to solve their own problems could lead to solving larger ones.
Rappaport thinks we are past Web 3.0 and headed toward the relevant rather than popular web. His take on the wisdom of crowds is seen with his work in medicine. A doctor or patient wants to quickly find information relevant to their specific health problem. He talked of the fox and the hedgehog. The fox knows broad knowledge of where things are in the forest. This is the current popular web. The hedgehog knows its specific area and can dig deep. It is this meaningful web of the specialist he hopes to develop.
Rao saw the limitations of expert systems while developing games at Xerox PARC. He discovered that simple strategies are complex to execute. Instead he focused and had great success with developing information intelligence. He still thinks it is possible to extract meaningful information from text. The provisional, useful, purpose-oriented extraction worked well for his company Inxight. In many cases, consumer search does not yield the deep knowledge needed by a professional like a doctor. How do you extract and use such information?
In short, the future of the semantic web will occur like previous advances, with one development bouncing off another.
Copyright 2006 DJ Cline All rights reserved.