This post is a complement to the grassroots vs. ivory towers discussion. In the previous post, James mentioned two trends of realizing the dream of the Semantic Web. While the grassroots way encourages the wider adoption of the Semantic Web, the ivory-tower way stimulates the construction of Semantic Web standards. Besides all of these, these two sides indeed play different roles in weaving the Web — the destiny of web evolution.
Weaving the Web
In 1999 Tim Berners-Lee published a book called Weaving the Web. In this book Berners-Lee introduced a brief history of the World Wide Web from its inventor's view. Despite the details of the book, the term "weave" is what we want to address here.
By weaving, we place two threads running perpendicular to one another. As the result, it produces a web of all the threads that are weaved. Web weaving is a little bit different from standard weaving due to the lack of pre-existing threads. In contrast, on the Web we first produce individual nodes (referenced by URLs or URIs), and then we link these nodes using external web links. As the result, Web weaving also produces a web — the World Wide Web. But we may call it a node-driven weaved web, in contrast to the standard thread-driven weaved webs.
Due to this difference, the WWW is distinct from standard-weaved webs. On the positive side, Web weaving has fewer prerequisites than standard weaving. Standard weaving demands pre-existing threads, which may or may not be obtained easily. Web weaving, however, does not require pre-existing threads. Web weaving allows people to produce random nodes and link them to any existing nodes randomly. There are far fewer restrictions on the Web weavers than on the standard weavers. So the WWW grows rapidly.
On the other side, node-driven weaved webs are less searchable than thread-driven weaved webs. For example, by naming the threads properly we can conduct highly efficient search algorithms to search thread-driven weaved webs. This type of algorithms is, however, less applicable on node-driven weaved webs (such as the current WWW) because we can hardly name long, uniform, single threads on this type of web.
Semantic Web: Turning a node-driven weaved web into a thread-driven weaved web
The difference between the ideal Semantic Web and the current WWW is the difference between a pure thread-driven weaved web and a pure node-driven weaved web. In the ideal Semantic Web, the individual machine-processable semantics are the named threads. Given a set of named threads, we can efficiently locate any nodes in the Semantic Web that are weaved using these threads. Moreover, these threads automatically link all the semantic-related nodes together. Whenever we dive into this thread-driven weaved web, we can swim through threads other than randomly jumping from one node to another.
These web threads are different from the normal web links. They are latent links, if we still prefer the term "link." They link web nodes implicitly, without the need for owners to assign these nodes and even without the need to notify the owners of these web nodes. These web threads present facts, just like the threads used in standard weaving. Unlike traditional web links, web threads are neither removable nor changeable. Thus, there exists a static method of web search upon this thread-driven weaved web, if only this web could have been built.
It is not easy, however, to turn a node-driven weaved web to a thread-driven weaved web. Above all, the content of a node-driven weaved web must have already been enriched enough so that it could be turned into a thread-driven weaved web. This is due to the General Law of Transformation from Quantity to Quality. These threads could not be revealed and lay upon each other until there are enough nodes on each thread. If the collection of traditional web links presents the explicit and dynamic graph of the WWW, the collection of web threads presents the implicit and static graph of the WWW. But we can discover this latent static structure of the WWW only after the explicit structure has already been clearly and exuberantly constructed.
The roles of Grassroots and Ivory Towers on Web Weaving
In order to realize this thread-driven weaved web, we need the help of both grassroots and ivory towers. In short, the role of grassroots is to fetch out these threads by their practices, and the role of ivory towers is to name and sort out these threads properly. Neither side can effectively take the duty of the other side.
In particular, the ivory-tower people must not enforce the threads in a theoretical way; this is a significant problem of the current Semantic Web research. Semantic annotation is not only about tagging, but also (and more importantly) about presenting social compromises explicitly. This is a process of evolving a new language, which is used by virtual servants of human Web users. So this language cannot be decided by a small group of people. The terms in this language are the names of the threads of the new WWW. The agreement of these terms can only be achieved by grassroots practices. This is what normal Web users must do before the final dream of the Semantic Web can eventually be realized.
About the author
Trackback URL for this entry:
Spam protection by Akismet
Post a comment