This repository is organized as follows:
-
datasets/
- contains all of the results of the annotations procedures outlined in the "ANNOTATIONS" section of "Automatically Extracting Subroutine Summary Descriptions from Unstructured Comments" as well as the predicitons automatically generated by models described in the "AUTOMATED ANNOTATION PREDICTION" section. Raw data from the works are included, as well as processed pickle files with data appropriate filtered and organized, and some side by side comparisons of various results.
-
metrics/
- contains python implementations of the various metrics used to measure similarity between summary annotations (see the "EXPERIMENTS" section of the paper).
-
scripts/
- contains a number of shell scripts that can be run to quickly address each of the 7 research questions outlined in the paper.
-
src/
- contains model and inference code used for the automatic summary prediction