Update README.md

README.md

@@ -1,3 +1,50 @@
----
-license: cc-by-nc-4.0
----
+---
+license: cc-by-nc-4.0
+language:
+- en
+tags:
+- twitter
+- nlp
+- social-media
+- english
+- sentiment
+pretty_name: Twitter Sentiment Threads
+---
+
+# Twitter Sentiment Dataset
+
+Sample English-only tweet sentiment dataset. Each row represents a single tweet with anonymized text and conversation structure.
+
+## What’s included
+- Anonymized tweet text
+- Conversation linkage via `root_id` and `parent_id`
+- 3-class sentiment label (positive, negative, neutral); for replies, sentiment is defined using both the tweet and its thread context.
+
+## Columns
+| Column | Type | Description |
+|--------|------|-------------|
+| `tweet_id` | string | Unique tweet identifier |
+| `parent_id` | string (nullable) | Parent tweet ID in the thread |
+| `root_id` | string | Root tweet ID of the conversation |
+| `text` | string | Anonymized tweet text |
+| `label` | string | Sentiment: `negative`, `neutral`, or `positive` |
+
+### Conversation structure
+- `root_id`: ID of the first tweet in the conversation (the root). For a root tweet, `root_id == tweet_id`.
+- `parent_id`: ID of the tweet this one replies to. `null` when the tweet is not a reply.
+- Threads form trees; following `parent_id` repeatedly walks from a reply up to the root.
+
+## Dataset Stats
+- **Total tweets**: 6,195
+- **Unique threads**: 2,660
+- **Language**: English only
+
+## Anonymization Example
+- Original: `Hey @jack check https://example.com at 2:30pm! 😊`
+- Masked: `Hey <USER> check <URL> at <TIME>!`
+
+## Source Data
+Data was collected from publicly accessible Twitter content that did not require paywalls or proprietary access. Only content available in the public domain was used, and all data is provided for research and trial purposes under CC BY-NC 4.0 licensing.
+
+## Contact
+For questions or takedown requests: contact@datahive.ai