metadata

annotations_creators:
  - found
language:
  - cs
language_creators:
  - found
license:
  - cc-by-nc-sa-3.0
multilinguality:
  - monolingual
pretty_name: Czech Facebook comments
size_categories:
  - 10K<n<100K
source_datasets:
  - original
tags: []
task_categories:
  - text-classification
task_ids:
  - sentiment-classification

Dataset Card for Czech Facebook comments

Dataset Description

The dataset contains user comments from Facebook. Each comment contains text, sentiment (positive/negative/neutral). The dataset has in total (train+validation+test) 6,600 reviews. The data is balanced.

Dataset Features

Each sample contains:

comment_id: unique string identifier of the comment.
sentiment_str: string representation of the rating - "pozitivní" / "neutrální" / "negativní"
sentiment_int: integer representation of the rating (1=positive, 0=neutral, -1=negative)
comment: the string of the comment

Dataset Source

The data is a processed adaptation of Facebook CZ Corpus. This adaptation is label-balanced.