metadata
dataset_info:
features:
- name: sentence
dtype: string
- name: aspect
dtype: string
- name: sentiment
dtype: string
- name: text
dtype: string
splits:
- name: train
num_bytes: 1681487
num_examples: 3759
- name: test
num_bytes: 477702
num_examples: 1056
download_size: 4972663
dataset_size: 2159189
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
semeval2016-full-absa-reviews-english-translated-resampled Dataset
Description
The semeval2016-full-absa-reviews-english-translated-resampled dataset is an annotated dataset for aspect-based sentiment analysis. Each example in the dataset consists of a sentence, its corresponding aspect, sentiment, and additional text for context. The dataset is divided into train and test sets.
Credits
Used eastwind/semeval-2016-absa-reviews-english-translated-resampled, which is derived from eastwind/semeval-2016-absa-reviews-english-translated-stanford-alpaca, by upsampling the neutral class and then resampling 3k examples from each class.
The here represented dataset is a result of summarization of the aspects and sentiments per equal sentence. In addition a textual representation is added for training purpose.
Features
- sentence: The main text of the sentence.
- aspect: The aspect targeted by the sentiment.
- sentiment: The sentiment expressed towards the aspect.
- text: A textual representation of the dataset for training.
Splits
- Train: Contains 3,759 training examples, totaling 1,673,969 bytes.
- Test: Contains 1,056 testing examples, totaling 475,590 bytes.
Download
- Download Size: 990,515 bytes
- Dataset Size: 2,149,559 bytes