Semantics: Starting Small…

Jump starting the Semantic Web is no small task…

There are many people looking for Google (or the ‘next Google’) to begin applying semantic principles to the creation of a new, incredibly rich index of the web. While they certainly have the technical wherewithal and available cash to make a credible move in this direction, I think that realizing the true power of the Semantic Web will require a somewhat different approach than that taken in the past. It needs to be a lot more distributed.

It needs to become social…

Today, most people use a folder paradigm to organize and structure their own information. If they can’t remember where they put a file, they can search for it by some general things date, size, file name, file type, or included text. The entire process of local discovery is primitive, inefficient, and incredibly frustrating for anyone lacking a rigid approach to organization.

If the semantic web is really going to take off on a large scale, it needs to happen first on a small scale. Semantics need to become an everyday part of the way individuals deal with information at a personal level.

Every piece of content a person touches has embedded semantic detail – contacts, companies, products, locations, dates, times, and topics. These are the incredibly valuable reference points people would like to use find things, and even more importantly, to connect things together. To make that happen, semantic analysis and indexation needs to start happening on a personal level. It needs to become a fundamental component of how people interact with their own emails, office documents and even the web pages they visit.

The Semantic Web needs to begin on peoples own computers…

As opposed to starting off by having one massive organization crawling the entire web, a better approach would be having millions of copies of a small distributed component crawling individuals’ own file systems, providing them with a rich semantic experience of their personal information first.

This semantic component would be able to download multiple relevant ontologies from the web and use them to deeply index (in a private and secure way) all of the information found on users’ personal systems. These ontologies would also be used index any web sites a user visits, allowing them to create deep contextual links between information they use on the web and the information on their desktops.

And it can also leverage the social dimension of the web…

As users share files, they could also share all of this enhanced meta-data they’ve generated. Beyond that, but they could even share the ontologies they used to create that metadata, letting recipients opt-in to including in their own ontology set. As these connections between individuals are made, sharing can also start to take place around the web indicies each has generated from their own web crawling, and ranking/relevance scoring can also be introduced. This effectively lets the people you interact with and trust influence the way you discover information.

This approach has the power to profoundly change the way individuals manage and share the information they have, and discover new information that may be relevant to them. And it can happen without waiting for someone else to manifest a fully realized ‘Semantic Web’ on a global scale.

So where would an approach like this leave Google?…

They would probably be in pretty good shape. The approach I’ve described here is highly distributed, and effectively creates millions of localized ‘semantic islands’. Google could become the semantic backbone that ties these islands together. They could become the clearing house for the certification and distribution of the ontologies that power this approach. They could become the rolled up index of all of the individual web crawls done at a local level – the global semantic index that powers broad searching needs. They could provide semantic indexation for commercial sites, and integrate it with the searches done at a local level.

The one thing Google shouldn’t do is try to own it all…

While it may not look like it today, the would is going to move away from traditional search as the model for finding information. Instead, discovery will take place by using the context of something I’m looking at or working on as a magnet for additional related information. If you see a name in a PDF file, you could use it to pull a phone number from your own Outlook, a bio from their Facebook page, and headlines about them or their company from all over the web.

In a single window. With a single click. Without doing a single search.

But to get there, we’ll need to start small.

With the Semantic Web, we should keep thinking global, but start acting local…