carblacac/twitter-sentiment-analysis

Dataset Card for TSATC: Twitter Sentiment Analysis Training Corpus

Table of Contents

• —Dataset Description
• —Dataset Summary
• —Supported Tasks and Leaderboards
• —Languages
• —Dataset Structure
• —Data Instances
• —Data Fields
• —Data Splits
• —Dataset Creation
• —Curation Rationale
• —Source Data
• —Annotations
• —Personal and Sensitive Information
• —Considerations for Using the Data
• —Social Impact of Dataset
• —Discussion of Biases
• —Other Known Limitations
• —Additional Information
• —Dataset Curators
• —Licensing Information
• —Citation Information
• —Contributions

Dataset Description

• —Homepage: TSATC
• —Repository: TSATC
• —Paper: TSATC: Twitter Sentiment Analysis Training Corpus
• —Point of Contact: Carlos Blanco

Dataset Summary

TSATC: Twitter Sentiment Analysis Training Corpus The original Twitter Sentiment Analysis Dataset contains 1,578,627 classified tweets, each row is marked as 1 for positive sentiment and 0 for negative sentiment. It can be downloaded from http://thinknook.com/wp-content/uploads/2012/09/Sentiment-Analysis-Dataset.zip. The dataset is based on data from the following two sources:

University of Michigan Sentiment Analysis competition on Kaggle Twitter Sentiment Corpus by Niek Sanders

This dataset has been transformed, selecting in a random way a subset of them, applying a cleaning process, and dividing them between the test and train subsets, keeping a balance between the number of positive and negative tweets within each of these subsets. These two files can be founded on https://github.com/cblancac/SentimentAnalysisBert/blob/main/data.

Finally, the train subset has been divided in two smallest datasets, train (80%) and validation (20%). The final dataset has been created with these two new subdatasets plus the previous test dataset.

Supported Tasks and Leaderboards

[More Information Needed]

Languages

The text in the dataset is in English.

Dataset Structure

Data Instances

Below are two examples from the dataset:

Data Fields

In the final dataset, all files are in the JSON format with f columns:

Each feeling has two possible values: 0 indicates the sentence has a negative sentiment, while 1 indicates a positive feeling.

Data Splits

The number of examples and the proportion sentiments are shown below:

Dataset Creation

Curation Rationale

Existing paraphrase identification datasets lack sentence pairs that have high lexical overlap without being paraphrases. Models trained on such data fail to distinguish pairs like flights from New York to Florida and flights from Florida to New York.

Source Data

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

Mentioned above.

Annotations

Annotation process

[More Information Needed]

Who are the annotators?

[More Information Needed]

Personal and Sensitive Information

[More Information Needed]

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

Dataset Curators

[More Information Needed]

Citation Information

@InProceedings{paws2019naacl,
  title = {{TSATC: Twitter Sentiment Analysis Training Corpus}},
  author = {Ibrahim Naji},
  booktitle = {thinknook},
  year = {2012}
}

Contributions

Thanks to myself @carblacac for adding this transformed dataset from the original one.