Manually Segmented Hashtags from Movie and TV Show Tweets

Loaded with 1,000 manually segmented hashtags, split into 500 for development and 500 for testing. The hashtags were selected from tweets about movies, TV shows, popular people, and sports teams. It was created by author 'ruanchaves' and last updated in October 2022.

Use Cases

• Train a word segmentation model on 1,000 manually segmented hashtags to split compound hashtags.
• Evaluate the performance of segmentation algorithms on a test set of 500 hashtags from social media topics.
• Analyze segmentation patterns in hashtags related to movies, TV shows, and popular people.
• Build a benchmark for hashtag normalization using expert-generated annotations.

Strengths

• 1,000 total manually segmented hashtags provide a defined corpus for model development and testing.
• Annotations are expert-generated, indicating a controlled labeling process.
• Clear split of 500 samples for development and 500 for testing supports standard evaluation workflows.

Limitations

• The dataset is small with only 1,000 samples, which may limit the training of complex models.
• The scope is limited to hashtags from specific topics like entertainment and sports, which may not generalize to other domains.
• Sample data and column structure are unavailable, hindering immediate understanding of the data format.

Provenance

Source

huggingface

Collection Method

Manually segmented hashtags selected from tweets.

Freshness

Last updated in October 2022.

Geography

Region tag indicates 'us', suggesting a potential US focus.