Summarizing Spoken Developer Conversations to Automatically Generate User Stories
Paige Rodeghero, Siyuan Jiang, Ameer Armaly, Collin McMillan
Please use the resources here to help you verify our results and recreate our approach. Several README files are included to explain specific scripts. If you need additional help or information, feel free to contact Paige Rodeghero.
Here is a virtual machine containing the resources above that can be used to test the approach. The download contains an Oracle VirtualBox Archive file.
- Size: 4.4GB
- OS: Debian (64bits)
- Memory: 2048MB
- User: everyman
- Password: password
- userstories: contains all the resources for the study
- db: contains all the data used for both the AMI meeting experiment and real company meeting experiment. (However, for privacy, much of the real company information has been removed from this package.)
- ami: contains the AMI meeting corpus, as well as a SQL dump of our experimental data. Also contains the results used in the paper.
- data: contains the partial real company meeting corpus, as well as a partial SQL dump of our experimental data. Also contains the results used in the paper.
- dbscripts: contains the scripts used for both experiments.
- setup: contains scripts used to setup the database before experimental running.
- attributes: contains scripts used to put conversation attributes into the database.
- mlout: contains scripts for performing machine learning on the database and produce results. If you have loaded our databases from the dumps, these are all you need to test our results.
Dr. Collin McMillan