This is the replication package of the work:
valid_instances.xlsx
: contains the valid instances inspected during the manual analysis and the respective labels.discarded_instances.xlsx
: contains the instances discared during the manual analysis and the respective reason for the discard.chatgpt_analysis.xlsx
: contains the instances used to evaluate the performances of ChatGpt-3 on the two considered code review automation tasks and its predictions.statistical_analysis_instance_complexity.xlsx
: contains the results of the statistical analysis on the instances complexity.code_comment_to_code_labels.xlsx
,code_to_comment_labels.xlsx
: contain information, such as number of total istances, number of correct/wrong predictions, percentage of correct predictions, for each label used, for code&comment-to-code and code-to-comment task respectively.chatgpt_analysis_code-to-comment_labels.xlsx
,chatgpt_analysis_code&comment-to-code_labels.xlsx
: contain information showing how the success and failure instances of ChatGPT are spread in our taxonomies, for code&comment-to-code and code-to-comment task respectively.code_complexity
: contains all the necessary to compute the code generation complexity based on AST-changes between the input code (code submitted for review) and the code to generate in order to implement the reviewer comment (code revised).