Universität Paderborn » SFB 901 » Projects » Tools & Demonstration Systems » MatchBox

MatchBox

MatchBox is a framework for comprehensive Matching Processes matching functional and non-functional properties of service specifications.

MatchBox also supports Fuzzy Matching.

MatchBox Main Features

MatchBox provides a comprehensive framework to match service specifications describing functional and non-functional service properties. All in all, MatchBox provides the following features:

  • Matching of comprehensive service specifications
    • Integration of service matchers considering functional and non-functional service properties
    • Creation, validation, and execution of configurable matching processes
    • Extensive matching results view
  • Fuzzy Matching in the presence of incomplete or imprecise specifications
  • Matching Result Validation
    • Comparison of a set of service specification pairs and their expected matching result with the actual matching result (computing statistics containing metrics like precision and recall or runtime)

These tasks are organized in three phases. In Phase 1, MatchBox needs to be set up, i.e., matchers need to be integrated. In Phase 2, matching processes can be modeled and configured. In Phase 3, matching processes can be executed fully automatically and their results can be inspected and validated.

The current version of MatchBox comes with a couple of already integrated matchers for different functional and non-functional properties:

  • Ontological Signature Matcher: This matcher analyzes the parameter types of the provided and requested service's operations for co- and contravariance leveraging domain knowledge from an underlying ontology. In addition, also operation names and exception types can be considered. All in all, our ontological signature matcher is configurable in many ways. 
  • Validated Ontological Signature Matcher: This matcher is an extended version of the ontological signature matcher. It implements concepts based on fuzzy sets to cope with fuzziness.
  • Fuzzy Condition Matcher: Our condition matcher matches pre- and postconditions in a FOL-based language.
  • Trace-Inclusion-based Protocol Matcher: This matcher matches protocol specifications based on finite automata by calculating how many paths in the required protocol are covered by the provided protocol.
  • Privacy Matcher: The privacy matcher considers a list of privacy-related properties, e.g., delegation depth, retention period, and the location where data is allowed to be used or stored. 
  • Reputation Matcher: This matcher takes ratings stored in a reputation system, calculates reputation values, and matches these values.
  • Simple Price Matcher: This simple matcher matches simple prices, i.e., single integer numbers.
  • Price Model Matcher: This matcher considers more complex price models. These price models can include a collection of price-related concepts, like accounts and flatrates.
  • Keyword Matcher: Our keyword matcher takes lists of keywords and matches them ontologically.

Screenshots, Screencast, and Posters

(click to enlarge)

Posters:

   

Screencast:

A MatchBox Screencast from June 2015 is available here: https://www.youtube.com/watch?v=Jarvb7nTqHY

MatchBox and other tools

Technical Information & Installation

  • MatchBox is a set of Eclipse Plug-Ins.
  • Matching Processes created with MatchBox are instances of our Matching Metamodel: see PDF
  • The current state of our implementation (nightly build) can be installed within the scope of SeSAME via our Eclipse Update Site

Publications related to MatchBox

Contact

If you have any questions regarding MatchBox, please contact research staff from Subproject B1.