User`s guide

About SpectroRx
The Seed Case Library
What is SpectroRx? SpectroRx (SPECTRUM Resolution Expert)
1-8 User’s Guide 5.0rev1
The Seed Case Library
All case-based reasoning systems start with a seed case library. The seed case
library provides a starting point for building the case library. It contains
broadbrush solutions provided by SPECTRUM so that you can begin using
SpectroRx immediately. However, to maximize the usefulness of SpectroRx,
you should edit the seed cases, as appropriate, to fit your needs and you
should add more cases as you respond to new problems.
You can automatically generate a seed case library for SpectroRx (see
Generating a Seed Case Library Automatically in Chapter 3). Alternatively,
you can also manually develop the library (see Generating a Seed Case
Library Manually in Chapter 3).
The Case Library
The SpectroRx case library is contained in a database separate from that of
the SpectroSERVER. Notice that it is not part of the SpectroSERVER
database.
Each entry in the case library corresponds to a case and is made up of the
following nine parameters in the order listed:
Case Number
Problem Summary
Problem Detail
Parameters
Solution Summary
Solution Detail
Runable Solution
Results
Notes
The Retrieval Algorithm
The heart of any case-based reasoning system is the retrieval algorithm.
Unlike rule-based or expert systems, case-based systems must have a flexible
way to evaluate the similarities of cases in the case library.
The SpectroRx retrieval algorithm compares the information in each library
case with the information in the current case. Each case in the library is
assigned points for every item that is similar to an item in the current case.
When the points are tallied and every case is scored, SpectroRx sorts the cases
according to the number of similarities and displays the most similar case
first, the least similar case last.