README.md

---
title: "WatchTower"
emoji: "🗼"
colorFrom: "black"
colorTo: "blue"
sdk: "gradio"
sdk_version: 3.9
app_file: app.py
pinned: true
---

**Note: Use WatchTower via the embedded interface at [HF.space](https://hf.space/embed/User1342/WatchTower/?)!**

## Files
- app.py - Gradio web interface for WatchTower.
- radical_tweet_aggregator.py - Pinpoint Wrapper and Tweepy integration for aggregating and assigning extremism scores to user posts.
- predictor.py - The Pinpoint wrapper script. Also see /Pinpoint and the respective Github repo for the Pinpoint codebase.

## Tweet Attribution and Bias
Please Note: If an account is blocked by Pinpoint, that is not to say that the author(s) of this tooling deem such a user to be violent or extremist, instead it is simply a result of the training the machine learning model has undergone and a result of the analysis on the last 10 tweets of the user in question. WatchTower is undergoing tuning to ensure it provides a fair and an as un-bias as possible assessment of Tweets, feedback is always welcome.

## Model Statistics
Pinpoint has several configurations, the configuration used by WatchTower has the following stats: 0.827 (accuracy), 0.676 (recall), 0.739 (precision), 0.706 (f-measure).

## Constraints
The Twitter APIv2 rate limits blocking requests to a 'User rate limit (User context): 15 requests per 15-minute window per each authenticated user'. This will pose issues to WatchTower when blocking larger numbers of users and will cause the tooling to hang or crash completely. A fix would be to change hosting and have a running service in the background that performs the blocking tasks adhering to the rate limiting.