BlueTrack SQL Database Tables Guide

This document provides a comprehensive overview of the SQL database tables used in the BlueTrack project, including their structure, fields, and purposes.

Tables Overview

Table Name	Row Count	Purpose
user_actions	27,713,990	Complete record of all user actions on the platform
user_stats	8,294	Aggregated statistics for each user
followers_list	728,506	User follow relationships mapping
user_actions_sampled	1,695,892	Sampled subset of user actions

---

Detailed Table Structures

1. user_actions

Purpose: Stores all user actions on the platform, including posts, replies, reposts, and likes.

Field Name	Type	Description
user_id	integer	Unique user identifier
action_type	text	Type of action: post, reply, repost, like, etc.
post_id	integer	ID of the post/action created by this operation
target_post_id	integer	ID of the target post (for replies or reposts)
target_author	integer	User ID of the target post's author
action_text	text	Text content entered by the user in this action
original_text	text	Original text of the target post (preserved for replies/reposts)
date	text	Timestamp of the action (format: YYYYMMDDHHmm)
labels	text	Action labels or categories
topic_name	text	Associated topic name

Example: A user posting a comment, replying to another post, or reposting content.

---

2. user_stats

Purpose: Stores aggregated statistics for each user, enabling quick queries of user activity and influence metrics.

Field Name	Type	Description
user_id	integer	Unique user identifier (primary key)
following_count	integer	Number of users this user follows
follower_count	integer	Number of followers this user has
mutual_follow_count	integer	Number of mutual followers
sampled_count	integer	Number of sampled actions for this user
post_count	integer	Total number of posts created
reply_count	integer	Total number of replies made
repost_count	integer	Total number of reposts made
like_count	integer	Total number of likes given
community_id	integer	Community ID the user belongs to

Example: User 100032 has 205 following, 83 followers, and created 60 posts.

---

3. followers_list

Purpose: Stores follow relationships between users, used to construct the social network graph.

Field Name	Type	Description
user_id	integer	ID of the user being followed
follower_id	integer	ID of the user who is following

Note: Each row represents that follower_id follows user_id. This is a directed relationship.

Example: If user_id=100032 and follower_id=100447, it means user 100447 follows user 100032.

---

4. user_actions_sampled

Purpose: Stores a sampled subset of user actions to reduce data volume while maintaining representativeness.

Field Name	Type	Description
user_id	bigint	Unique user identifier
action_type	text	Type of action: post, reply, repost, like, etc.
post_id	bigint	ID of the post/action created
target_post_id	bigint	ID of the target post
target_author	bigint	User ID of the target post's author
action_text	text	Text content entered by the user
original_text	text	Original text of the target post
date	text	Timestamp of the action
labels	text	Action labels
topic_name	text	Associated topic name
source_type	text	Data source type

Note: This table is a sampled version of user_actions, retaining all fields but with significantly reduced data volume (~1.7M rows vs 27.7B rows), enabling faster analysis and testing.

---

Usage Recommendations

1. Quick User Queries: Use user_stats to retrieve aggregated user data
2. Behavior Analysis: Use user_actions or user_actions_sampled for detailed action analysis
3. Social Network Construction: Use followers_list to build follow relationship graphs
4. Performance Optimization: Prioritize user_actions_sampled over full user_actions for large-scale analysis
5. Community Analysis: Use community_id in user_stats to segment users by community

---

Data Statistics

• Total unique users: 8,294 (from final_users)
• Total actions recorded: 27.7 billion (from user_actions)
• Total follow relationships: 728,506 (from followers_list)
• Sampled actions: 1.7 million (from user_actions_sampled)

Dataset Source: Based on the Bluesky dataset from Failla & Rossetti (2024)