Methodology

Decision-Making in Conceptual Design

The iConceptStore^™ original cognitive methodology is based on the premise that expert knowledge consists of interrelated dynamic mental models of relevant essential parts of an application domain of interest (as internalised perceptions of external reality, augmented by imagination, including problem-solving methods of related human practice in the form of conditionally attached reasoning mechanisms. The unique declarative Conceptual Modelling Language (CML) of the iConceptStore^™ Conceptor^™ Knowledge Engineering subsystem is intended for domain experts, engaged in replication of such highly structured mental models (for the purpose of this complementary expressive role, the CML acronym naturally stands for Conceptual Mental Language as well). Since the extent of both human knowledge and understanding has vague aspects and dynamic boundaries that gradually expand, CML was designed with flexibility (to reflect different types and degree of uncertainty) and extensibility in mind. Any unknown or ill-understood matters that lie beyond those boundaries at any given point in time remain outside its current scope (better left to the so called “new” AI experts to tackle).

One of the most distinguishing features of iConceptStore^™ Conceptor^™ CML is its general-purpose design. As conceived from the outset, these generic capabilities stem from its conceptual foundation – human thoughts are concept based reflection (perceptions generalisation) of reality observed or imagined. While knowledge in a specific area of expertise may enrich the human thinking process and its result, human’s natural intelligence is not confined to any particular domain. Likewise, while the syntax of iConceptStore^™ Conceptor^™ CML is restricted to ensure well-formed and unambiguous expression of expert knowledge, its vocabulary and usage are not. In fact, they are virtually unlimited, going far beyond even those of a natural language – it is entirely up to the user what kind of words (real or made-up) to use as names of the concepts (familiar or invented) he/she defines. The semantics (explicit or not) of those user concepts is also open to arbitrary interpretation by custom application programs (of one form or another).

In this respect the iConceptStore^™ Knowledge Representation methodology appears unique in greatly surpassing (while not outright rejecting) the widely spread simplistic view of Attribute as a Name-Value (Key-Value) pair, Entity as a set of such primitive value attributes and Relation as a Subject-Predicate-Object triple, all possibly assembled as sets of Entity-Attribute-Value triples in so called triple stores. Through standardisation and other promotional measures this approach (adopted in different versions of the highly acclaimed Entity-Relationship Model, cognitive architecture SOAR, Conceptual Graphs, Knowledge Graphs and Description Logic based RDF/OWL standards with cumbersome XML syntax) has been elevated to status of ultimate, universal and only solution to Knowledge Representation. Such claim bears no justification whatsoever. As briefly indicated here, in terms of scope and expressive power the implemented iConceptStore^™ CML highly structured multi-dimensional concept of Attribute (as well as the composite Entity, Relation and other built-in CML categories) of two-way derivative nature with implicit (or explicit) single or multiple values is a much more general and sophisticated mental construct. Even the iConceptStore’s own offer of ready-to-use kind of triple store capability is just a special case – built-in CML subtype – of the general-purpose custom-definable CML entity attribute Variations construct and is logically interpreted within the broader context of related iConceptStore^™ conceptual models (incidentally, our first documented implementation of what we used to call “set of associative triples”, maintained on VAX/VMS secondary storage as an extension to Prolog, dates back to 1985).

In effect, the Conceptor^™ CML seamlessly embodies the iConceptStore^™ underlying cognitive methodology. In particular, the iConceptStore^™ Conceptor^™ CML built-in modelling categories (selected general-purpose mental models based meta-concepts with well-defined semantics, dynamically supported by its run-time engine) constitute the cornerstone of its methodological foundation, making the knowledge engineering process more systematic and efficient. By combining conditionally in CML scripts* these built-in modelling categories (ready made configurable patterns of thought) with references to his/her own custom concept definitions (or vice versa), one can form descriptions of highly structured mental models of arbitrary complexity thus reflecting dynamic compositions of nested fragments of useful expert knowledge in any specific application domain. The use (absolutely optional) of these built-in semantic categories greatly facilitates the process of appropriate conceptual model structuring and description as part of any general systems engineering activity. As such an embodiment of the iConceptStore^™ original cognitive methodology, the iConceptStore^™ CML acronym also stands for Cognitive Methodology Language. Overall, these capabilities are best utilised within Artificial Intelligence (AI) context because knowledge is the heart and soul of human intelligence – knowledge is based on understanding of (and reflects) the corresponding laws of the environment (the relevant application domain) thus providing for an adequate behaviour as a specific (not like the “new” AI's statistics based averaged) response to any specific problem situation.

* In iConceptStore^™ contexts the term ‘script’ is used as synonymous to ‘written model description’ and CML scripts have nothing in common with the widely used (usually interpreted) programming script languages such as VBScript or JavaScript.