KakologArchives
VerifiedArchive of 11 years of Nico Nico Jikkyo live commentary logs.
What is KakologArchives?
The collection preserves complete historical logs from the former Nico Nico Jikkyo platform after its 2020 renewal ended access to the old API and data.
Researchers working on Japanese social-media text, live-comment analysis, or text-classification models can use the logs for training and evaluation.
What you can build with KakologArchives
Train Japanese comment sentiment models
Use the 11-year corpus of live viewer comments to fine-tune sentiment classifiers that detect reactions to TV broadcasts or events in real time.
Build timeline-based event detectors
Analyze spikes in comment volume and keywords across channels to automatically identify major moments in sports, news, or anime episodes.
Create domain-specific language models
Pre-train or continue-train Japanese LLMs on authentic live-chat text to improve handling of slang, abbreviations, and real-time conversational style.
Load KakologArchives
from datasets import load_dataset
ds = load_dataset("KakologArchives/KakologArchives")- 1pip install datasets
- 2from datasets import load_dataset
- 3dataset = load_dataset('KakologArchives/KakologArchives')
- 4Access splits and columns to inspect raw Japanese comments
- 5Filter or tokenize the text for your downstream NLP pipeline
KakologArchives: pros & cons
Pros
- +11-year complete archive of Japanese live comments
- +Covers all channels including TV, radio and BS
- +Preserves data lost after the 2020 service change
- +Large authentic user-generated Japanese text corpus
Cons
- –Japanese language only
- –No updates after 2020 service shutdown
- –Raw comments contain noise and spam
Frequently asked questions
A community-collected archive of all past comments from the original Nico Nico Jikkyo service spanning 2009-2020.
User reviews
Verified reviews from the community shape this listing's rating.
Loading reviews…
Similar datasets
Other text & nlp options worth comparing.