tags:
- emoji_extraction
- bluesky_emoji_extraction
- emojis
- emoji_analysis
- emoji
Dataset Card for Emoji Extraction from Bluesky Posts
Dataset Description
Overview
This dataset contains extracted emojis from Bluesky social media posts, totaling 11.2 GB per file, with a total of 10 JSONL files. Each file is structured to include emojis along with their frequency, name, and Unicode code. The dataset is a valuable resource for researchers and developers working on natural language processing (NLP), large language models (LLMs), generative AI, and data analysis.
Dataset Size
Total Size: 112 GB (11.2 GB per file × 10 files) Format: JSONL (JSON Lines)
Dataset Structure
Each line in the JSONL files contains the following fields:
emoji: The actual emoji character.
frequency: The frequency of the emoji in the dataset.
name: The name or description of the emoji.
code: The Unicode code point for the emoji
Dataset Source
The data is sourced from Bluesky social media posts, capturing the usage of emojis across various posts.
Dataset Usage
Potential Applications
This dataset can be utilized for a variety of purposes, including but not limited to:
Large Language Models (LLMs):
Training :
The dataset can be used to train LLMs to understand and generate text that includes emojis.
Fine-tuning:
Fine-tune existing LLMs to better interpret and produce emojis in context.
Generative AI:
Text Generation:
Enhance text generation models to include emojis that are contextually relevant.
Chatbots:
Improve chatbot responses by incorporating emojis that match the tone and context of the conversation.
Data Analysis:
Trend Analysis:
Analyze the popularity and usage trends of emojis over time.
Sentiment Analysis:
Use emojis as a proxy for sentiment in social media posts.
User Behavior:
Study how users interact with emojis in different contexts.
AI Research:
Emoji Prediction:
Develop models that predict the most appropriate emoji to accompany a given text.
Multimodal Learning:
Combine text and emoji data to improve multimodal models that understand both textual and visual (emoji) content.
Social Media Analysis:
Content Moderation:
Identify and moderate content based on the presence of specific emojis.
Engagement Analysis:
Analyze how the use of emojis affects user engagement on social media platforms.
Access and Usage
The dataset is available for research and commercial use. Users are encouraged to cite the source when using the dataset in their work.
Ethical Considerations
Privacy:
Ensure that the dataset does not contain any personally identifiable information (PII).
Bias:
Be aware of potential biases in emoji usage and consider how these might affect model training and analysis.
License
The dataset is released under the MIT License, allowing for free use, modification, and distribution, provided that the original license is included.
Dataset Details
File Naming Convention
File Names:
emoji_df_.jsonl where ranges from 1 to 10.
Data Fields
emoji: The emoji character itself.
frequency: The number of times the emoji appears in the dataset.
name: A descriptive name for the emoji.
code: The Unicode representation of the emoji.
Dataset Creation
Data Collection
The data was collected from Bluesky social media posts, focusing on the extraction of emojis and their metadata.
Data Processing
Emoji Extraction:
Emojis were extracted from the text of each post.
Frequency Counting:
The frequency of each emoji was calculated.
Metadata:
Additional metadata such as the name and Unicode code point were added for each emoji.
Data Storage
The processed data was stored in JSONL format, with each emoji entry on a new line.
Citation
If you use this dataset in your research, please cite it as follows:
Martin Rivera [DeepSeek-V2.5-1210]. (2024). Emoji Extraction from Bluesky Posts [TroglodyteDerivations/Bluesky_Emoji_Extraction].
Available at https://huggingface.co/datasets/TroglodyteDerivations/Bluesky_Emoji_Extraction
Acknowledgments
We would like to thank Bluesky for providing the platform and data, and all contributors who helped in the collection and processing of this dataset.