Twitter Indonesia Sarcastic: Indonesian Tweets Labeled for Sarcasm Detection

Twitter Indonesia Sarcastic is a dataset for sarcasm detection in the Indonesian language, introduced by Khotijah et al. in 2020. The dataset contains Indonesian tweets labeled as either sarcastic or non-sarcastic. Author w11wo performed cleaning procedures including sentence order re-reversal, deduplication, and PII masking before uploading it to Hugging Face in December 2023.

Use Cases

• Train sarcasm detection models based on labeled Indonesian tweets.
• Benchmark NLP models for Indonesian language understanding based on the binary sarcasm labels.
• Study linguistic features of sarcasm in informal Indonesian text based on the social media source.
• Perform cross-lingual analysis of sarcasm by comparing this dataset with others in different languages.

Strengths

• Data underwent cleaning procedures including deduplication with minHash LSH.
• Personal identifiable information was masked, removing usernames, hashtags, emails, and URLs.
• Dataset is based on a published academic source (Khotijah et al., 2020).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect temporal and platform bias inherent to Twitter as a source.

Provenance

Source

Hugging Face user w11wo, based on data from Khotijah et al. (2020).

Collection Method

Indonesian tweets were collected and labeled as sarcastic or non-sarcastic.

Freshness

Last updated 2023-12-21 09:01:38; freshness should be verified.

Geography

Indonesia (inferred from language and dataset title).