ReFeed is a tool for automatically extracting relevant information from end user feedback and computing priorities for the associated set of requirements. ReFeed is able to:
- create a meaningful association between requirements and user feedback
- extract relevant properties from user feedback (sentiment, intention, severity)
- synthesize prioritization for the given set of requirements based on the properties of the associated user feedback
The project is developed using Java 8 and the build is managed via [Maven] (https://maven.apache.org/), so you need to have a working Maven installation and Java 8 runtime to build and execute ReFeed.
To build the project:
- clone the repo
$ git clone https://github.com/se-fbk/ReFeed.git
- set some local paths in
config.properties
: path to GATE files and path to WordNet DB files; download the files from the respective websites and place them locally, then update the paths in the config file - build the project
$ mvn package
make sure that there are not tests failing.
To run the overall tool, it suffices to run the class userfeedbacknlp.util.App.java
To reproduce the results from the experiments reported in the paper, it suffices to run the class userfeedbacknlp.util.Stat.java
The necessary data required are included in the project directory.
Apache V2.0 (https://www.apache.org/licenses/LICENSE-2.0)