Our approach to the semantic web

Web 3.0, the semantic web, the web of things, all mean the same – the organisation and structuring of data based on their meaning, readable and searchable through machines, much the same way humans grasp the content on a website.

The idea behind this approach is that information, which at present is unrelated or only related based on ‘guesses’ by search engines and machines, is structured in such a way that the relation between content becomes visible, that meaning is injected into the content and that each piece of content is not seen as independent but part of a mesh of information.

The problem

In other words, if we read an article, we know that the title is the title and an author is an author. HTML offers possibilities to give a semantic meaning to a page or an article, but pure HTML does not know that content within, for example, an H2 tag could equally be a subtitle, the name of an author, a town, or anything else. Of course, this H2 tag could contain a link to the author, town, or other content, but even that would not help the machine to understand what it is referencing. Thus, a little help is needed, for example through the inclusion of a  ‘rel’ attribute, RDFa data, or micro formats.

Possible solutions

At present there are a number of different approaches, from POSH (plain old semantic html), over RDFa, to HTML5’s micro data in combination with the new semantic tags (header, footer, article, section, etc.). This raises a number of questions, especially which of these approaches should be supported, and even more importantly, which one is easiest to implement. Unfortunately, the semantic web requires some knowledge in addition to HTML. Furthermore, even if one is proficient in the use of writing semantic markup, where does the semantic information come from?

This is exactly the question we’ve been asking ourselves when we started to develop flowli. Traditional systems of categorising and storing data usually do this through the ubiquitous folder systems. While these systems are helpful in providing some order, users are constantly confronted with questions like ‘should this letter now go into a folder called ‘clients’ or ‘projects’?’

Articles become categories

This reflection led us to the decision to forego any traditional way of storing data, but rather to develop a system that would on the one hand allow multiple references to where something should go, and on the other make the system extremely fluid. Why, for example, do I need to create a folder in the first place? Very much like threaded emails, which follow a semantic and logical order, we decided ditch folders altogether, and turn the articles themselves into categories. Indeed this works just like in a family tree, where any element with children becomes a hierarchical element one level above the last generation.

Auto-semantic

However, this system does not only make it easier to create and organise content, but due to its very nature the content becomes linked and connected on a semantic level. Content is now organised based on its meaning and relationships to other content and not on some abstract level. These relationships do not apply to articles, but can be created between any piece of content within flowli. For example, a user could follow an article or an article could follow a user. The same is true for images, videos, etc.

In order to make this sematic organisation of content still easier, we’ve partnered with Strider Digital Intelligence, whose services allow flowli users to create these semantic links automatically. In the near future it will also allow deep-linking of articles, creating relations not only on a Meta level but within the content itself.

The semantic organisation of content within flowli provides a way to mirror these relations in any website running on flowli – that’s one step closer towards a truly semantic website.