It is important to distinguish data, information and knowledge (see Figure 21.1). Data are the symbols perceived by an observer through sensors. From these data emerge information, that is data with a strong semantic content as a result of an interpretation process. The observer's knowledge permits association of the semantic content with data. The following step is a process of appropriation and reasoning that leads to the integration of the information in the background knowledge of the observer.
Thus, knowledge has several roles: (1) the transformation of data to information, (2) the derivation of new information from that existing, and (3) the acquisition of new knowledge. Knowledge is thus simultaneously a result and a process.
Two types of knowledge must be considered , i.e. explicit and tacit knowledge [Polanyi, 1962; Nonaka, 1994]. Explicit knowledge is easily formalised and communicated, while tacit knowledge is highly personal and difficult to express. (Note that recent discussions make clearer this distinction, e.g. see [Br zillon, 2001a].)
Figure 21.2 presents four types of movements between these two types of knowledge:
Socialisation permits adaptation of our tacit knowledge with our interaction with others (a kind of internal knowledge creation),
Externalisation permits communication, at least partially, of our tacit knowledge,
Combination permits generation of new explicit knowledge after our interaction with the others, and
Internalisation permits assimilation of new knowledge from external sources that enriches our tacit knowledge.
(see [Nonaka, 1995] for an initial presentation and [Pomerol, 1999; Br zillon, 1999b], for a discussion on this subject).
Here, the process of externalisation interests us particularly with respect to context because it anticipates the process of proceduralisation leading to the proceduralised context, which will be discussed in the next section. Note that, more generally , the relationships between knowledge and context (see Figures 21.1 and 21.2) can be summed up as [Pomerol, 2001]:
Context and knowledge can be made explicit or implicit but both must be made explicit to be communicated (i.e. exploited),
The context may contain deep and shallow knowledge,
Contextual knowledge is task-oriented but are not limited to know-how,
Contextual knowledge permits description of the state of nature in which is made a decision or accomplished an action, and can be combined in different ways for that,
Proceduralisation of contextual knowledge is a decisive step in the execution of an action and in the triggering of the know-how,
The proceduralised context is task-oriented and highly subjective as the know-how and situated knowledge, and
The link between the proceduralised context and the corresponding action can be made explicit or implicit (as a result of the proceduralisation that gives the proceduralised context).
Making context explicit in terms of knowledge permits identification in a better way that is necessary for a system (or an object) with respect to a user at a given time according to the available data, information and knowledge.
Some results indicate that dimension of the context is infinite [McCarthy, 1993]. As a consequence, a context is always relative to another more general context, and thus can not be totally described. On our side, we first distinguish the part of the context that concerns the focus of attention (the contextual knowledge) from the knowledge that is not relevant at that phase of the focus of attention (the external knowledge). At a given step of the focus of attention, a sub-set of the contextual knowledge is mobilised, situated, organised and structured to be used at that step of the focus of attention. We call the result of this compilation of the contextual knowledge, the proceduralised context . Figure 21.3 gives a synthesis of the three types of context.
For a given focus of attention (e.g. a step of a problem solving), a static definition of context is the set of contextual elements that give a meaning to the focus of attention without intervening in it explicitly. Thus, we consider that we have not to distinguish context from the other objects concerned by the reasoning, objects entering or going out the context according the events. Moreover, the move of the focus of attention implies a dynamic dimension of context.
In the framework of communicating objects, we think that objects must share a reference, a shared source for data, information, knowledge and context. This referential will be updated regularly and shared by objects. Thus, different communicating objects (for example, inside a house) can coordinate their activities according to the knowledge that can be useful for them (their respective proceduralised context) and knowledge of other objects (a part of their contextual knowledge).
Context is knowledge, and knowledge is context. At a given moment, there are external knowledge and contextual knowledge, and a part of the contextual knowledge is compiled in the focus of attention (through the proceduralised context). Context is relative to a focus of attention (e.g. the context of knowledge use) with an organisation of the contextual knowledge around this focus (e.g. see the onion metaphor in [Br zillon, 1997]). The contextual knowledge, which is organised around the focus of attention, also has a granularity that depends on the distance to the focus. Moreover, the movement of the focus of attention implies a change in the proceduralised context, and explains the dynamic of context. As a consequence, the "reasoning" of a communicating object must account for this dynamic of context.