The Semantic Problem
The way in which we communicate information to computers and how this information is processed has not changed much since the first computers became available. If we use a keyboard, touch-screen or voice commands, we still feed mostly text-information into specialized applications that do more or less one task, be it an address manager on a smartphone, an ERP-client running on a desktop or a web-shop on the internet. We communicate mostly with plain text. The problem is that plain text has no meaning for computers. Applications must determine the meaning of plain text information from the context in which it has been fed to them. For example, if we are using an address manager and enter the text “John Doe” into an input field that expects the name of a person, the application can assume that “John Doe” is the name of a person and therefore that text gets a meaning. Outside of this text-field most applications would have no way to tell that the 8 characters “John Doe” represent a person’s name, much less that the name represents a concrete person of interest in the current context. Computers do not understand plain text the way humans do. For them it is just a sequence of characters without any meaning. Only context, e.g. its place inside a structure or if we enter it into a designated text field gives it meaning. If computers cannot understand the meaning of information, they cannot reuse it in another context nor can they automatically analyze or even interpret it in a meaningful way. This “semantic problem” has of course been recognized many years ago. For example, the semantic web has been defined by the W3C in 1999 and is today an integral part of the World Wide Web. But its focus is on web-resources and information annotation, not on information expression itself. Havel is a semantic, expressive, universal computer language. It is capable of expressing arbitrary, complex information that both, humans and computers, can understand. Using Havel for human-to-computer or computer-to-computer communications would solve most of the issues associated with the “semantic problem”. In the example above, instead of filling a form, the user could either express the whole address in Havel or point to a place where the app can read a predefined address expressed in Havel. In a first generation semantic environment, this could automatically fill out the fields in the form. In a more advanced version, the application would not need a form in the first place; it can directly work with the Havel-expression. In later generations, there might be no need for that application at all; the address itself - or rather the address resolved in the user’s information universe - is the application (see information-centric computing for more details on this concept). An address expressed in Havel can be relatively simple, just an entity classified as a person and a bunch of textual properties, each also classified, therefore each having meaning to the application, very much like a RDF/OWL annotated record. But it could also be a complex, semantic structure where textual information is only required for representational purposes, each bit of information meticulously referenced into a global, semantic knowledge base, letting the digital version of John Doe become a part of a vast digital information universe. In any case, an application or website that is able to “speak” Havel will be able to understand a properly expressed address, be it the simple or the advanced version, as long as it knows or at least has access to the definitions of the semantic identifiers used by the address.
This post is part of a comprehensive introduction into SRDM, graph languages and Havel.
Continue reading: Universal Information
Comments