Information Needs for Identifying Bug-introducing Changes
Paper Study 2 - Research Questions
Collect the frequency of each comment ended with the ?. Data:
- the id of issue;
- the id of reviewers, devs, and reporters (all people involved in the process);
- the body of comments with the questions and aswers
- the size of comments
- the labels of issue/PR
- the code source related to the change. (later)
- the creation and update date of each comment, respecting the respective time zones.
RQ01- With information needs can be captured from viewpoints?[Done].
Study 1 collected from literature, social media, personal interview, and experience.
RQ02 - How often each information need occur (frequency)?
Metric: Compute the number of replies of each comment.
Example: 
RQ03 - Which information needs attract more discussion (density)?
Metric: Compute the density of comments among the replies/time. The number of replies of each information needs view.
Example: 
RQ04 - How long does it take to get a response to each information (response time)?
Metric: Compute the density of comments among the replies/time (in terms of minutes) needed to get a response in differents perspectives such as: category, addressee (dev or reviewer), topic (fix, enhancement, project). Example: 
RQ05 - Which information needs have been applied?
Collect the information with at least one answer. How to automate this?.
RQ06 - How long does it take to apply an information need?
Collect the information with at least one answer and compute the time of interaction. How to automate this?.
RQ07 - How do the information needs change over the development?
Metric: Compute the number of times a certain information need appears in each iteration of a code review.
Example: 
RQ08 - How do the information needs impact in the software quality?
If the presence of each information needs has a statistical difference on issues performing changes with bugs and without bugs.
Data: Collect the same amount of issues without bugs.
Metric: Compute a Spearman’s test.
RQ09 - Is there a relationship between information needs and the number of issues/bugs/enchancements/fixes in a project?
RQ09.1. Use a statistical modeling technique called Analysis of Variance (ANOVA) to measure which factors and interactions of factors influence the dependent variable. The independent variable is the category, while the dependent variable is the issues.
RQ9.2, ANOVA models (ARIMA) will be built with the bugs as the dependent variable.
RQ9.3, ANOVA models will be built with the enchancements as the dependent variable.
RQ9.4, ANOVA models will be built with the fixes as the dependent variable.
RQ10 - Are there temporal trends on information needs?
Metric: Construct time series models with ARIMA to forecast future information needs.
RQ11 - Are these correlations between informations needs?
Metric: Compute the Spearman’s correlation tests between categories of information needs.
Example: 
RQ12 - Can information needs forecast for future bugs, enhancement, and fixes?
Metric: Construct ARIMA models on time series data of categories for each project. Then, transfer ARIMA models between:
- information needs -> bugs
- information needs -> enhancements
- information needs -> fixes
Example: 