jCOLIBRI 2 is a framework in Java for building Case-based Reasoning (CBR) systems. It includes mechanisms to Retrieve, Reuse, Revise and Retain cases and is designed to be easily extended with new components.
jCOLIBRI 2 is a major release where most of the jCOLIBRI 1.x framework has been reimplemented. The aim of this new version is to improve the architecture of the framework making it more robust and easier to extend.
However, jCOLIBRI 2 does not include the authoring tools that in version 1.x allow to build a CBR system without writing a single line of code. Authoring tools are under development for jCOLIBRI 2.x but, in the meantime, version 1.1 can be a better solution for those users that don't want to write Java code. * Text retrieval based on TFDF
Implemented through Apache Lucene.
- Text clustering
Implemente through Carrot2.
- Classification and maintenance methods
Developed by Derek Bridge and Lisa Cummins from University College Cork, Ireland.
- Case base visualization methods
Developed by Josep LLuís Arcos from Artificial Intelligence Research Institute, Spanish Scientific Research Council.
- Improved methods for textual CBR
Using OpenNLP and GATE new and more sophisticated similarity measures have been included for textual CBR (like the compression based similarities developed by Derek Bridge).
- Improved connectors to store case bases
Connector to in databases through Hibernate. Hibernate is a powerful, high performance object/relational persistence and query service. This project and a critical component of broadly used JBoss J2EE server supporting persistence in Data Bases or XML files among many other features like its own object-oriented query language.
Connector to ontologies through OntoBridge, a simple wrapper for Jena/DIG reasoner.
Connector to textual files.
For the connector to work cases have to be implemented as JavaBeans.
- Improved ontology-based similarity measures
New ontology-based similarity functions are included. Ontologies are accessed through OntoBridge what makes the implementation more robust and efficient.
- Improved access to Wordnet
Wordnet is now easier to use than in version 1.1, and it can be loaded into memory through our WordNetBridge library, so there is no need to install it separately.
- Improved Tomcat integration
jCOLIBRI 2.0 jar file can be registered in a servlet engine and call from servlets. Running jCOLIBRI 1.1 as a servlet required some bizarre hacking.
jCOLIBRI includes 16 completely documented examples and a stand-alone CBR application that uses the library to easily learn to use the framework. Version 2.1 includes methods for developing recommendation systems. Some of the implemented methods can be used in general CBR applications and other are specific for recommender systems.
General CBR methods:
- Filtering Retrieval method.
- XML utils to serialize cases and queries.
- Methods to obtain the query graphically (using forms).
- Methods to display cases.
- Cases retrieval using diversity.
- Cases selection using diversity.
Specific methods for recommendation systems:
- Methods to implement the Expert Clerk recommender.
- Methods to implement collaborative recommenders.
- Methods to obtain the query from user profiles.
-Methods to implement Navigation by Asking recommenders.
- Methods to implement Navigation by Proposing recommenders.
- Local similarity measures for recommender systems.