Ë  Ë
CPSC 679 Project - Analysis

Analysis of Repertory Grids

Development of an Expert Knowledge Base for the Control of an Intelligent Robot Sumo Wrestler

by Vladimir R Vila


Line Drawings of Various
Robots

Situation Assessment Module

Course Project for CPSC 679 -- Cognitive Processes in Artificial Intelligence


electromagnetic spectrum
line

The following are the results of the repertory grid analyses. They include links to a summary of the construct relations between each expert. These summaries are in 4 quadrant diagram form and should be read as Consensus (upper left), Correspondence (upper right), Conflict (lower left) and Contrast (lower right). Then a brief textual summary is given. Following this are some links to the data diagrams. A summary and conclusion section is at the end.


electromagnetic spectrum
line


ANTHONY


electromagnetic spectrum
line
Œ

Graph and Diagram Links


Pie Graphs of Inter-Expert Correspondence and Contrast

Relational Link Representation of Correspondence and Contrast

Pie Graphs of inter-Expert Element Consensus and Conflict

Relational Link Representation of Inter-Expert Element Consensus

Pie Graphs of Inter-Expert Construct Consensus and Conflict

Relational Link Representation of Inter-Expert Element Consensus and Conflict

electromagnetic spectrum
line

Summary and Conclusions

Overall Rob had the highest level of correspondence in his comparisons with the other experts, getting no lower that 50% and up to 55.6% correspondence in his comparison with Vlad. Vlad had similarly high results between Don and Rob however he dipped down to 30% in comparison to Anthony.

Don had contrast with all the other experts and significant amounts (50%) with Anthony. In Anthony's case he had not correspondence.

ŒIn the element consensus case, Robs psychological space seems to give Vlad more relational links than in any other. Vlad's space does the same for Anthony. This suggests that Vlad should use Rob's space to proceed in the elicitations and to simply communicate more effectively with the group, Anthony should use Vlad's space. Rob and Don never had more than two relational links in any of the spaces. Don and Vlad had no links in Anthony's space. Anthony has no links in Don's space.

In the construct consensus case, Vlad again sees the most links in Rob's space. In this case, Anthony also has the most relational links in Rob's space. Rob gets the most links in Vlad's space. Don again has no links in Anthony's space and Anthony has none in Don's.

---

There seems to be a reasonable level of correspondence between most of the group. Anthony and Don were the exception with no correspondence. This is concerning, since Anthony is the instructor for the group and is arguably the most knowledgeable on the subject. This high level of knowledge could explain the differences between Anthony and the two students Don and Vlad. But it is expected that a certain amount of correspondence be found between instructor and student or else learning would be very difficult. On the plus side, Rob seems to have good correspondence and consensus with the entire group. This could put him in an intermediary position in which he could acts as a psychological space translator.

Based on the relational links, it would seem that some combination of Rob and Vlad's space could facilitate higher overall consensus within the group. Anthony and Don do not have any more than two links in any space; element nor construct. Don does not have any links in Anthony's space and Anthony has no links in Don's space. This would indicate that using either Anthony or Don's space would reduce the amount of consensus in the group. This divergence of psychological spaces is also evident in Anthony and Don's correspondence results were both had zero correspondence and a high level of contrast.


island line