Research group

Machine Learning Methods in Software Engineering

Large-Scale Anomaly Detection for Kotlin

Project supervisor: Timofey Bryksin
Status: Active

We apply anomaly detection algorithms on all publicly avaliable Kotlin code to identify anomalous code fragments that can be of interest for the Kotlin compiler team.
Basically, we download all Kotlin code from GitHub, vectorize it, and feed it into an anomaly detection algorithm. We do the same for bytecode, where we can.We found a bunch of highly unconventional code snippets that are of interest for the Kotlin compiler team, and a few examples of suboptimal compiler behaviour.


The paper has been rejected at ICSE 2020. We are submitting it to another venue in early 2020.

Participants