ECOBWEB is a concept formation program for the creation of hierarchical classification trees. It implements several extensions to Fisher's COBWEB program. In particular, it can work well with numeric attributes, it can perform simple constructive induction, it has a procedure for mitigating order effects, it has an experimentation procedure, and it has several methods for classification that make it suitable for design domains. ECOBWEB employs multistrategy learning; it is a concept formation program that includes case-based reasoning capabilities.
ECOBWEB was developed as part of Bridger--- an experimental learning system developed to explore the possibility of using machine learning techniques for the acquisition and improvement of design knowledge. Two instantiations of ECOBWEB are used in Bridger, one for learning knowledge for candidate designs generation and one for learning to adapt candidate designs for given specifications.
ECOBWEB was implemented in Common Lisp. I expect it to run on most implementations of the language. (If you don't have Lisp, you can get a free copy of CMU Common Lisp from the AI repository at CMU. I have recently (March, 97) downloaded another free common lisp implementation from this repository and run ECOBWEB successfully. This version is gcl).
Click here to obtain ECOBWEB's code (compressed tar file, 112K, to get the files do: tar xvf file-name). Note that there is only minimal documentation/explanations coming with it on how to use it. Have fun and tell me about your experience with the code. You can also get an old and very incomplete manual (postscript 300KB; pdf 2.8MB) that contains a description of the program and its basic functions.
The database of bridges used in this study --- the Pittsburgh Bridge Database --- was donated to the ML Database Archive at UC Irvine, and can be retrieved from the subdirectory bridges.