Matchmaker Configuration Mechanisms

From wiki.gpii
Jump to: navigation, search

This page discusses matchmaker configuration mechanisms (see ID205.1), which are specific to a matchmaker (statistics-based, rule-based, hybrid).

Goals:

  • A rule editor for the rule/heuristics-based algorithm. (However, see A204.3.)
  • For the statistical matchmaker:
    • An editor for managing the sets of statistical training data;
    • an editor for the parameters of the statistical algorithm;
    • a visualization and solution approval tool for the classification results of the statistical algorithm
  • For the hybrid matchmaker:
    • Rule editor and tools for the statistical algorithm for the hybrid approach.
    • In addition, a conflict resolution tool is needed for competing results from the two approaches.


The Matchmaker (with uppercase M) is a component in the Architecture that can call upon different types of matchmakers, i.e.

  • the "flat" matchmaker,
  • the canopy matchmaker,
  • the rule-based matchmaker,
  • the statistics-based matchmaker,
  • the hybrid matchmaker (to be developed later).

It is possible to define configuration mechanisms at the level of the Matchmaker, but since each individual matchmaker has a different feature set, configuration mechanisms can also be defined at the level of the individual matchmakers.

General GPII Configuration

For the matchmakers developed for the first evaluation phase, only the GPII framework's configuration mechanisms were in place. These include mechanisms specific to matchmaker configuration.

Matchmaker Modules

Infer Preferences

This module alters preferences coming from the preference server or adds new preferences. It would be used, for example, for cross-platform inference or reacting to changes in context.

Infer Solutions

This module alters solutions coming from the Solutions Registry or adds new solutions. It would be used, for example, to include web-services or enforcing certain solutions. (“Solutions” is the general term for applications, access features, and assistive technologies.)

Matchmaking Strategy

This module selects solutions based on (inferred) preferences and the (inferred) solutions.

Rule-Based Matchmaker Specific Configuration

Statistical Matchmaker Specific Configuration