Declarative knowledge

In the previous section, we presented the modeling classes that allow one to describe the structure and behavior of chemical processing systems. Such a representation provides and structures the declarative knowledge about the current state of the design. 

Declarative knowledge should be completely decoupled from the procedural knowledge. 

At various times over the past 25 years, cognitive psychologists have found it convenient to postulate two kinds of knowledge procedural and declarative.1 For the most part, procedural knowledge means rule knowledge and has to do with skill acquisition and performance. Declarative knowledge is composed of concepts and facts. A schema does not exactly fit into either of these categories but instead transcends both of them. 

Schemas can be instantly modified (e.g., by the incorporation of a new piece of declarative knowledge), and they can be gradually tuned over time, as the result of repeated application. Experimental evidence suggests that schemas can be triggered unintentionally and retrieved deliberately. For instance, in problem solving studies, subjects sometimes recognize some feature in the problem that brings an entire schema into working memory immediately. On the other hand, a subject may be at a loss as to how to proceed and may deliberately and consciously attempt to recall any schemas that he can remember ever using to see if one suffices.

Schemas contain both procedural and declarative knowledge and extend well beyond the usual formulation of concept knowledge. Full assessment calls for the demonstration of all types of related schema knowledge and their interaction. 

A rather different type of production system model was developed by John Anderson (1983). His ACT architecture extended the notion of production system models in two important ways. First, Anderson hypothesized that a distinction exists between procedural and declarative knowledge, and second, he added the important mechanism of spreading activation. 

Declarative knowledge, 1775 Decomposition, 2167-2170 Decomposition-based approach (production-inventory systems), 1692 Decomposition heuristics (scheduling), 1729-1731... 

Frames, scripts, and cases are different forms of schemas that represent knowledge as typical stereotypical chunks. Rules and scripts generally represent procedural knowledge, while frames and cases represent declarative knowledge. By formally representing data acquired from knowledge-acquisition techniques, a model of the procedural and declarative knowledge required to perform a task is created. 

Object-oriented systems store information in frames and slots which contain facts, pointers to other objects, or rules. This combination allows the objects to be interrelated to each other in different ways. In objected-oriented systems, slots may inherit information from other objects rather than creating duplicative objects. The structure of object-oriented expert systems more resembles that of hypertext (discussed above) than rule-based systems. Object-oriented systems contain more declarative knowledge than rule-based systems. 

A solution to the interface and schema-deficiency problems described above is to integrate a hypertext-based user interface and online help system with the expert system. The interface and help system facilitate access to the information and provide the declarative knowledge needed to make decisions and interpret the results of the consultation. 

The tutor uses information from the student assessment and from the expert model to generate instructional sequences. At this point, if a student misses a declarative knowledge question, s/ he is transported to the hypertext node (card) that contains that information. If the student misses a structural knowledge question, then the expert model expert system is activated, generating an optimal structural path for the learner. This path is then illustrated for the learner. If the learner misses a procedural knowledge question, practice scenarios are provided with standard practice and feedback. The assumptions and sophistication of this part of the tutor needs development. 

The declarative knowledge pertinent to the topology and the electrical components of the circuit to be analyzed is represented via a hierarchy of units. 

Declarative knowledge refers to objects, properties, and events. Declarative knowledge is about the world (rather than about e activity of attaining goals). 

In short, procedural knowledge is goal-indexed, imperative, and more or less efficient, while declarative knowledge is goal-independent, descriptive, and more or less accurate. 

---