Twine Gets New Injection of Capital

Dan Primack at Private Equity Hub, and Chris Morrison at VentureBeat are reporting that Radar Networks has received a new round of funding for its semantic web application Twine. VentureBeat claims that the new deal is probably somewhere in the region of $15-$20 million. It seems that Radar will be using the new cash to scale up their private beta (currently in the region of 1000 users) and to start partnering with other internet firms. Hopefully this means that those of us who have been on the waiting list for the beta may get a chance to test out the software soon.

Twine is described as a semantic 'personal data organiser'. It is an application that uses semantic technologies to organise information that is uploaded by the user and imported from other online sources so that it can be used more intelligently.

There is a lot of misunderstanding over what exactly the Semantic Web is, and what services like Twine are going to do, so in an attempt to clarify the issue I have compiled the following Semantic Web Glossary as an aid to understanding.

Semantic Web - The best explanation of the Semantic Web came from Tim Berners-Lee, who describes the paradigm in terms of a progression from the internet to the web to the 'Graph'. First there was the internet, which connected computers, and allowed cables to become invisible. Then there was the web, which connected documents, and allowed computers to become invisible. Next, there will be what he calls the 'Graph',which will connect things, and allow documents to become invisible. At each stage we are realising that it's not X, but Y that is interesting. It's not the cables themselves that are interesting, but the computers. It's not the computers that are interesting, but the documents stored on them. It's not the documents that are interesting, but the things that are referenced by them.

When I go to somebody's Facebook profile it's not that particular document that I am interested in, but the person that the profile is about. The problem is that that person has pages about them all over the internet. They may also have a twitter stream, a blog, a flickr account and a wikipedia entry. The proposal of the graph is that instead of browsing to a document you browse to a object, which could be a person, a place, a film, a particular flight, or anything else.

The technology challenge here is how to make connections between all the information on the web that is about a particular thing. At the moment computers do not have any way to understand what objects are referred to in a particular document; only humans can do that. The challenge is to create new protocols and models that allow computers to intelligently understand the contents of documents and then make connections between them.

RDF - Standing for Resource Description Framework, RDF is one of the core technologies that will enable computers to understand the contents of documents. It is called a 'metadata model' which essentially means that it is a way of structuring a document to allow meaningful information to be extracted from it by a machine. The RDF specification is based on a tripartite model called a triple, which represents resources in the form of 'subject-predicate-object' expressions. So the RDF model for the expression 'My computer is a Mac' would consist of three parts; the subject ('My computer'); the predicate ('is'); and the object ('a Mac').

OWL - Standing, somewhat inaccurately, for Web Ontology Language, is another specification that allows allows a definition of classes and a modeling of the relations between them. This is an ontological language, which means that it it concerned with defining the kinds of things that exist in web documents and applications. These 'kinds of things' are called classes. The specific members of classes are called individuals, and the relations between classes are called properties. An example of a class might be 'Book', or 'Author'. An example of an individual might be 'The Pickwick Papers', or 'Charles Dickens', and an example of a property might be 'WrittenBy'. Thus the property 'WrittenBy' defines the relationship between the classes 'Book' and 'Author'. OWL is thus a method for mapping out all the things on the web and the relationships between them. OWL ontologies draw on the information provided by the RDF schema.

SPARQL - Standing for SPARQL Protocol and RDF Query Language, SPARQL is a language used to make queries about RDF models. Because RDF allows machines to understand not just documents but the things to which they refer, SPARQL can make queries about those things. With SPARQL, for example, we could ask the question 'List all fantasy books written between 1985 and 1998 by authors who have published more than 5 books'. At the moment we wouldn't be able to get answer because there are hardly any (in comparison to the total volume of documents on the web) RDF models available, but when the whole of the web is modeled using RDF, SPARQL will be an extraordinarily powerful tool.

I hope that this brief glossary has shed some light on what the Semantic Web effort is all about, and why so many people are excited about products like Twine. If you think I've got anything wrong, or have anything else to say, please feel free to leave a comment.