GOALS

GECAD Ontology ALignment System

This work is partially supported by the Portuguese MCT-FCT project COALESCE (PTDC/EIA/74417/2006).

This work is being published in conferences, workshops and other knowledge disseminating events.

Copyright remains with the authors.

This information should not be copy-pasted without acknowledging its origins.
Please contact the
authors for information on how to use or reference this information.

Data-Entities

Represent any kind of data structures that Components manipulate as input and/or output.

Currently, the system supports:

* ONTOLOGY which corresponds to one ontology specification;

* MATRIX which corresponds to a set of matches (or correspondences);

* ALIGNMENT which corresponds to a set of mappings, i.e.  a level 0, 1 or 2 Alignment.

Note: An level 0 Alignment might be represented as a MATRIX.

Components

Also called actors, are objects that play one or more roles in the ontology matching process. According to that roles, most common components might be classified as:

* Loaders — are able to load a Data-Entity from a given location;

* Matchers — when implementing a given matching algorithm (e.g. String-Distance );

* Operators —  when implementing a specific Data-Entity operation (e.g. filtering one Matrix/Alignment using a threshold value or aggregate several Matrix/Alignment into just one through a given function);

* Alignment Extractors — are those that applying some technique are able to transform 1  or more Matrix into one Alignment;

* Alignment Evaluators — are able to compare one Alignment (generated by the system) with a provided reference Alignment  applying evaluation metrics (e.g. Precison, Recall);

* Adapters — are responsible to transform a Data-Entity in the common representation format to another (different) format that is expected by an (external) Component and vice-versa.

Each component explicitly defines what are the required input Data-Entities and which output Data-Entities are provided. Particularities of each component are configurable through a set of parameters.

Workflow specification

Is about choosing, from all available components, which ones will take part of the matching process, its parameters and how Data-Entities flow between components. Note that, output Data-Entities of one component might be used as input Data-Entities for others components. The result of a workflow specification is a new complex matcher; referred to as meta-matcher.

A workflow specification can be done (i) programmatically, (ii) by script through an XML file or (iii) using a GUI (which is not available yet).

Execution Engine

Is able to tackle an workflow specification (i.e. meta-matcher) and automatically execute/run it. Following steps are done:

* reads the list of components that are present in the workflow and creates one instance for each component  (respectively);

* each component instance is informed of what are the components responsible for providing required input data entities;

* particular parameters of each component instance are set up with the defined values during workflow specification phase;

* terminator  components are automatically identified and a ‘run’ command is sent to each one that must adopt a pull-strategy behavior to get their required inputs and further process those inputs in order to generate the intended outputs.

Core concepts