The Evolution of the World Wide Web: The Role of Bidirectional Linking (December 2015)
Abstract
At the end of the 1990s, the World Wide Web had proven to be globally scalable and searchable and it was rewarded by a truly world-wide adoption. Having solved the major technical problems, a human issue became apparent: Web search—so it turned out—was highly susceptible to commercial manipulation to a degree, that search results were rendered completely useless. The manipulation of Web sites in favour of a seemingly high relevance in the search engines of that time—a technique dubbed spamdexing—triggered many discussions and research on the special nature of the Web and its possible future.
The findings of one academic research would eventually result in the world’s most popular search engine: Google. The makers of Google succeeded in developing a search algorithm that proved to be insusceptible to commercial manipulation by determining the relevance of a Web page independently of its own content but merely based on references from other sites. In that way Google found a working analogy for the real world, in that a work is not to be considered important, if its authors say it is, but when others do.
A more fundamental criticism of that time was, that the Web was merely a collection of documents interlinked with meaningless links. In order to advance the Web, the next evolutionary step would be to promote the Web of documents to a Web of data, in which data elements could be interlinked in a meaningful way. Doing this in a standardized fashion would allow automatic software agents to perform tasks on the Web on behalf of their human users. Since the freedom of creating Web links on the Web of documents was considered key to its success, the idea to expand this freedom to a Web of data was promising.
Motivated by the still pending success of the Web of data, this thesis researches the evolution of both: the Web of documents and the Web of data. The linking capabilities of both Web trends are analysed over time and contrasted with each other. As a result, the Web of data is found to be highly susceptible to manipulation. If the Web of data is one day globally adopted, commercial manipulation will become an issue again. However, because Web linking and the constraints under which it takes place are very different in both Web trends, the approach found by Google for the Web of documents will not work on the Web of data.
To make the Web of data less prone to manipulation, this thesis discusses three approaches: bidirectional linking for affirming facts, a same-origin policy for linking, and rules for changing the evaluation context. Bidirectional linking is found to have always played an important role in the evolution of the Web, also prior to the implementation in its present form. Bidirectional linking is consequently discussed against a wider background: from information theory and visionaries of global information management, its role in the evolution of the Web of documents and data to examples of bidirectional linking on today’s Web.