AUG 29th 2007

More and more people are starting to discuss the Semantic Web, but few truly understand how it is different from the traditional World Wide Web. Though the Semantic Web will be realized as a layer upon the current Web, some of their basic philosophies are going to be updated significantly. This post tries to list some truth about the Semantic Web that is critical but often overlooked.

A semantic web must simultaneously be a proactive web, while the traditional Web is a reactive web. Personally, I claim the emergence of proactive web nodes to be the first identity of the Semantic Web.

Misconceptions about the Semantic Web

As we know, the Web is a network of human-created web nodes. Generally, standard web nodes are considered to be reactive individuals, whose basic behavior is reacting to remote requests. Traditionally the Web is a reactive web.

There is a common bias about realizing the Semantic Web from the current Web: by adding machine-processable semantics into web content, the current Web is going to become the Semantic Web automatically. This biased understanding contains a latent false assumption — a semantic web can be realized as a reactive web. A reactive web would not automatically become active or proactive simply by adding machine-processable semantics.

In fact, a "reactive semantic web" is so awkward that it is nearly impossible to be real. Reactive means that the machine-processable semantics in such a web are specified prior for others, but not for the producers themselves, to consume. This recognition unfortunately contradicts to the self-centric nature of human beings. Therefore, a reactive semantic web is a hopelessly awkward product.

In order to overcome this intrinsic obstacle about realizing Semantic Web, we need to adjust the basic view about the Web. Standard web nodes in a semantic web should be proactive individuals, which can not only react to remote requests, but also proactively execute local requests at remote locations. That means that this capability becomes standard for every semantic-web node, not just a special capability of some nodes. This fundamental upgrade of web nodes would enable a web composed by these nodes to be a proactive web. Within the proactive web environment, web users can be ensured that their specifications of machine-processable semantics are primarily contributed to themselves. The prevalence of this recognition is the basis of public adoption of semantic web.

Based on these discussions, a realized semantic web must simultaneously be a proactive web. The construction of proactive web nodes is a prerequisite to building a practical semantic web. Thus I claim the construction of proactive web nodes to be the first identity of the Semantic Web.

Web 2.0, the intermediate web

Web 2.0 is a low-degree active web, an initial step towards the Semantic Web

In between the reactive Web and the proactive Web, there are several intermediate stages. The currently popular Web 2.0 is one of them. Unlike the traditional reactive web or the future proactive web, Web 2.0 is a low-degree active web.

Web 2.0 is beyond a reactive web. Besides the basic capabilities owned by a standard traditional web node, a typical Web 2.0 node owns at least one extra fundamental capability of actively processing remote comments inputted by anonymous readers. This improvement is significant because by this capability a Web 2.0 node is able to automatically update its content based on a certain level of machine understanding. A web composed by nodes with this type of self-activated capability becomes an active web.

At the same time, however, the degree of self-activated capabilities in Web 2.0 is still low. In fact, the number of this degree is closely related to the level of machine understanding the respective web can process. If machines can automatically better understand web content, they can perform higher-degrees of self-activated functions. This means the introduction of higher degree self-activated capabilities into normal web nodes. In Web 2.0, most of the semantics, such as which request is with respect to which comment, are hard-coded and there are limited numbers of them. Web 2.0 is thus only a low-degree active web. Web 2.0 is an initial step towards the Semantic Web.

In a semantic web, each web node by default has a unique interpretation about the existing web. In contrast, in the traditional Web every web node by default shares a common interpretation about the existing web.

Not everyone has the same point of view

Here is another commonly biased understanding about the Semantic Web: a semantic web can be identically queried from any semantic web node. A homogeneous view from different nodes over the entire web is a common assumption in the traditional Web. For example, when we perform a query over the web, no matter which web node the query is issued on, the query result is assumed to be identical by default.

We can explain this assumption using a metaphor so that it can be better understood. When there are arbitrary web nodes A and B, John stands on Node A watching the web and Peter stands on Node B watching the web. In the traditional Web, John and Peter would get back an identical view of the web. This is because in the traditional Web, a view contains only facts but not interpretations (semantics) of the facts. By default, we assume every viewer has a common interpretation of the facts when the processing of interpretations is required. This previous assumption can, however, no longer be held generally about a semantic web.

In order to explain the reason, let us ask ourselves a question — when we ask an arbitrary question to a few randomly selected persons, do we generally expect identical answers or varied answers? Certainly in the real world, we expect to receive varied answers. The reason that answers are more likely to be varied is because different people usually think varied even about the same object.

The Semantic Web is like the real world

Ambiguity exists everywhere in a real semantic world. Instead of trying to overcome ambiguities, in real life we try to compromise with ambiguities. We humans are trying to stay in a vague way to allow more forgiveness in our society. Most of the time, a real society can keep peaceful not due to the overcoming of ambiguity (dictating a society is often unstable), but due to the allowance of forgiveness (democratic society is usually more stable). This is a fundamental rule for creating a stable semantic world.

The Semantic Web is another semantic world. Therefore, if the Semantic Web can be freely accepted by the general public, it must also hold this property of forgiveness. This requirement thus means the allowance of returning varied results for the same query when it is issued by different web nodes.

Inside a semantic web, every proactive web node must have the right to interpret the other web nodes based on its own view instead of being forced to adopt others' interpretations. At the same time, any web node can certainly FREELY (but not by force) decide to adopt any existing view and follow them unconditionally (as young children often fully adopt their parents' views). But in general case, we should not assume the existence of any global agreements on semantic understanding.

The Semantic Web is a realistic, obtainable goal

One question might be whether or not such a web can realistically be built. Can we really avoid the very difficult problems such as ontology matching? The skeptics should ask themselves another question: whether a democratic society is realistic? A dictating society is easier to build but harder to maintain. In contrast, a democratic society is harder to build but easier to be maintain. This is thus the trade-off.

Inside a proactive web, two connected web nodes communicate to each other by themselves based on shared ontological definitions. It is unnecessary to assume generic ontological agreements at any greater scale than the two connected nodes. A proactive web can be established based on simple P2P agreements. In contrast, the ontological agreements at any greater scale can help construct bigger communities among normal Web users. The reaching of greater scale ontological agreements, however, completely depends on the free will of web users.

The motivation of seeking increased communication will help construct greater scale ontological agreements. The motivation of seeking profits of invention, however, will help construct brand new ontological agreements to replace old ones. This is a secret in modern democratic society; and this is the secret to construct a practical semantic world from bottom-up.

About the author

Yihong Ding

I'm currently a Ph.D candidate in Brigham Young University with Prof. David W. Embley in Computer Science Department, Prof. Deryle W. Lonsdale in Linguistic Department, and Prof. Stephen W. Liddle in Marriott School of Management.

Trackback URL for this entry:

Spam protection by Akismet

» Semantic Web

Tracked on August 29, 2007

» Het web volgens Yihong Ding

Tracked on August 5, 2008

Comments for this entry:

No one has left a comment for this entry. Be the first!

Post a comment

  1. Spam protection by Akismet