DISCO: A Dataset of Discord Chat Conversations for Software Engineering Research

Keerthana Muthu Subash, Lakshmi Prasanna Kumar, Sri Lakshmi Vadlamani, Preetha Chatterjee, Olga Baysal

Type: dataset
Venue: IEEE / ACM (MSR 2022)
Year: 2022
Source: ieee
Access: free
Language: English
Added: 2026-07-17T20:18:03.571473+00:00
Verified: 2026-07-17T20:18:03.571473+00:00

Summary

DISCO is a dataset of Discord chat conversations collected from four software development community Discord servers, consisting of 28,712 disentangled conversations and 1,508,093 messages posted by 323,562 users. The researchers applied a disentanglement technique to extract individual conversations from chat transcripts and manually validated a random sample of 500 conversations. The dataset enables research on developer communication tools such as automated virtual assistants, chat bots, chat summarization, and Q&A systems. Published at the 2022 IEEE/ACM 19th International Conference on Mining Software Repositories (MSR).

Keywords

discord software-engineering chat developer-communication msr conversations disentanglement

Topics

Software Engineering / NLP

Research notes

IEEE Xplore page was accessible via search. The paper is also available via DOI 10.1145/3524842.3528018 and on the SOAR Lab website. Authors are from Carleton University and Drexel University.