Meaning, supervenience and symbol grounding The Web is often conceived as a set of layers, with standards, languages

5.1 Meaning, supervenience and symbol grounding The Web is often conceived as a set of layers, with standards, languages

or protocols acting as platforms upon which new, richer, more expres- sive formalisms can sit. Such platforms, like TCP/IP, are deliberately intended to be as neutral as possible. The Semantic Web is an obvious example of a layered yet unprescriptive architecture [32].

Such layered representations are not reductive – that is, the upper levels are not merely shorthand for expressions at the lower levels. But there is an interesting question to do with the significance of such lay- ered representations of the architecture. In particular, the nearer to the

72 Social Aspects

top that an expression is found, the more likely it is to have meaning. By which we mean that although an expressive language needs to have formal syntax (and possibly semantics), to be significant it still needs to map onto human discourse in an intelligible way.

In the Semantic Web model, ontologies are intended to perform this mapping, and to make meaningful dialogue between human and machine possible [97], although it is important to be clear that such mappings aren’t magic: ontologies, as artificial creations, stand in just as much need of mapping onto human discourse as the structures they map [113, 289]). And in this, they are no different to other structured formalisms, such as queries [39].

One view is reminiscent of the philosophical idea of supervenience [168, 169]). One discourse or set of expressions A supervenes on another set B when a change in A entails a change in B but not vice versa. So, on a supervenience theory of the mind/brain, any change in men- tal state entails some change in brain state, but a change in brain state need not necessarily result in a change in mental state. Super- venience is a less strong concept than reduction (a reductionist the- ory of the mind/brain would mean one could deduce mental state from brain state, that psychology follows from neuroscience). And it has been thought over the years that supervenience is a good way of explaining the generation of meaning: uninterpreted material in the lower layers of discourse is organised in significant ways so that the material in the upper layers is constrained to be meaningful. It may be appropriate to think of the Web as having this sort of super- venience layering: the meaningful constructs at the top depending crucially on meaningless constructs in HTML or XML or whatever below.

If we are to see the higher levels of the Web as supervenient on the lower, then the question arises as to what the foundational levels of the Web are, and the further question of whether they have to take some particular form or other. One does not have to subscribe to the requirement for symbol grounding (i.e. the need to avoid symbol mean- ing being ‘grounded’ only in other symbols, and instead being grounded by some direct relationship with a referent – [129, 130] – a requirement that Wittgenstein, among others, denied could be fulfilled – [291]) to

5.2. Web reasoning 73 expect to see some sort of discourse of uninterpreted symbols playing

a foundational role. ‘Meaning is use’ is a well-known slogan that represents a key insight in Wittgenstein’s later philosophy of language. It clearly contains a very important insight, and applied to natural language is a powerful mes- sage to understand meaning in terms of what people use their language to do. The same insight applies to the Semantic Web, but there is a wider question of what ‘use’ consists in. In the world of machine pro- cessing and interoperability of data, much of the use or discourse is automatically generated by computers. For that reason, it is not clear that definitions in words, or code, or quite specific uses, will not suffice to pin down terms for the Semantic Web with sufficient accuracy to allow logical deduction to take place. Stability of the referents of key URIs, for example, might enable a great deal of automation in specific topic areas – a notion of science as underpinning meanings reminiscent of the theories of Hilary Putnam [233]. The fact that the Semantic Web works in the world of relational data, with machines doing much of the work, means that it isn’t necessarily incumbent upon it to solve the problems of definition and logic that have proved so resistant to analysis in the world of natural language, although new insights may

be gained from the grounding in URIs discussed in section 3.1.2 above.