The sentiment analysis is fundamental and not specified towards the language used in GitHub issues. Most of the time it will not consider the context of a word which results in inaccurate scores. Additionally, it only takes the issue title as input and does not look at comments or descriptions which can say a lot about the sentiment around an issue and thus about the individual sustainability of the developers working on the issue.
It would be an idea to get the data of the comments and descriptions as well as create a better ML model that gives more accurate sentiment scores.
Due to the above reasons, sentiment analysis has currently been marked as an experimental metric (see #50 and #51). If this issue is addressed properly, it can be moved to the initial dashboard.
Description
The sentiment analysis is fundamental and not specified towards the language used in GitHub issues. Most of the time it will not consider the context of a word which results in inaccurate scores. Additionally, it only takes the issue title as input and does not look at comments or descriptions which can say a lot about the sentiment around an issue and thus about the individual sustainability of the developers working on the issue.
It would be an idea to get the data of the comments and descriptions as well as create a better ML model that gives more accurate sentiment scores.
Reference Issues
40
Implementation PR
No response