CREATOR

Main Page 

Architecture 

Case Representation 

Memory Structure 

Problem Elaboration 

Case Retrieval
Memory Structure
 
The memory structure comprises two interconnected substructures: a tree of high level functionalities and functional specifications, and a graph of cases (see figure). The tree of HLFs and FSs has two distinct goals. One is to provide help in the problem elaboration phase. The other goal is to retrieve the starting cases. The graph of cases is a structure that, once retrieved the starting cases, is used by the exploration mechanism to retrieve cases from the case library, as needed for the adaptation process. 
 
 

The taxonomy of HLFs and FSs has five types of nodes and five types of links. The nodes can be: a HLF class, a HLF instance, a FS class, a FS instance, or a list of FSs. Links can be of type: hlf-ako, hlf-isa, fs-ako, fs-isa, or part-of. Hlf-ako links establish connections between HLF classes, while hlf-isa establish connections between a HLF class and a HLF instance. Fs-ako links connect FS classes and fs-isa links connect FS instances. Links form a HLF and FS taxonomy. These two taxonomies are previously acquired from the knowledge engineer. HLF nodes can have part-of links. Part-of edges link a HLF instance to a list of FSs. These links establish a partonomical hierarchy between HLFs and FSs. This structure is built from the case library, extracting the problem definition from cases. 

The graph of cases is composed by nodes and edges. A node represents a case, and an edge represents the differences between the cases that it links. In the figure it can be seen Case1 connected with Case2 by the differences between them. The connection between the function tree and the case graph is performed using links between a node representing a FS instance and a node representing a case. Links are automatically created, when the tree structure is built. In this way cases are indexed by the tree structure.