Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
json
Sub-tasks:
sentiment-classification
Languages:
Russian
Size:
10K - 100K
License:
metadata
languages:
- ru
multilinguality:
- monolingual
pretty_name: Kinopoisk
size_categories:
- 10K<n<100K
task_categories:
- sentiment-classification
Kinopoisk movie reviews dataset (TOP250 & BOTTOM100 rank lists).
In total it contains 36,591 reviews from July 2004 to November 2012.
With following distribution along the 3-point sentiment scale:
- Good 27,264
- Bad 4,751
- Neutral 4,576
Load with python like:
import pandas as pd
df = pd.read_json('kinopoisk.jsonl', lines=True)
df.sample(5)