semantic web

The semantic web is an ideal form of the World Wide Web. It doesn't exist in reality. Refer link semantic web to see what those who believe in it, say about it.

There are several levels of ambition or optimism about it:

The least optimistic is that a controlled reflexive intranet can consist mostly of semantic links with link types. Standards for this might become popular on more public subsets of the World Wide Web, such as the open politics web. A list of header tags and footer tags for instance could be used to decide whether a given page or article belonged on a list of search results or should be included in a compendium or publication, without human inspection or editorial judgement. In this limited sense the semantic web is a set of limits/borders.

Slightly more optimistically, using some naming conventions such as the open politics argument standards would make it easy to find all statements related to a given pivot or target or goal statement, e.g. an issue statement. This is as far as the Living Ontology Web would support: an imaginative rationality applied to a fixed list of all human command verbs, trying to make it easy to find all reasoning about say the word "arrest" or "deport" or "audit", but not to do such reasoning itself. Humans do all dispute resolution and dispute detection.

Very optimistically, some think it possible to go beyondt his to an ultra-reflexive intranet which can answer simple questions, say about which positions had valid or credible evidence attached, using only explicit semantic relations as a guide. This could be used for instance for rapid answers in a candidate portal.There might be use of market-based methods or betting mechanisms or scenarios to do answer recommendation, and even a minimal system would require a great deal of specific analysis and terminology and the underlying corpus to be organized in some standard form such as OWL.

An extended upper ontology may be someday capable of this, but only a strong web ontology applied uniformly in all languages would be likely to perform to very high standards of accuracy, e.g. those of an expert system equivalent in accuracy to an involved legislator or good researcher.

Thankfully it isn't necessary always to give an answer from pre-existing data, only to form the question and find someone to answer: Where the machine capacity to answer is outrun, the human experts could possibly be consulted without any additional editorial intervention or permission, since the semantic web knows who to trust to arbitrate what, when. A step towards this is the Answers to Questionnaires that some political party operatives collect during elections to guide policy debate and compare policy options later.

Human intervention in at least the form of a Lowest Troll would remain necessary always, if only for dispute resolution.

Most optimistically, fully automated reasoning without any humans involved would be not only possible but trustworthy: simply to state relationships between entities would permit them to be processed and reasoned about. There is no visible web but a linearized stream of questions and actions, more like an FAQ working on demand for even the most complex, high-stakes or deeply situated questions.

This latter vision has inspired some of the W3 ontology work, perhaps unwisely. Tim Berners-Lee and (in public policy) K. Eric Drexler seem to be advocates of these or similar approaches.