Discovering the True Identity of @Horse_js Using Machine Learning
summary
Project collects ~7,000 Horse_js tweets via the Twitter API, backfills with horseplane data, and stores them in an API called app horse hunt using Azure Functions and Cosmos DB. Analysis uses time‑series and frequency aggregation queries plus text analytics (cognitive services); machine learning plays a small role while statistics drive most results. Findings: clusters of quoted originals within 60 seconds, a suspect set of six in Eastern/Central time, and an Android source unique to one account. Rate limits slow collection but the pipeline works.
AI-written summary · about
chapters (23)
- 0:00Intro
- 1:10Meet Burke Jasmine
- 3:26How the idea came about
- 4:40What is Horse Genius
- 5:58The Process
- 7:56The Questions
- 10:04Assumptions
- 14:30How will we do this
- 15:28Meet Vadim
- 16:22What is machine learning
- 17:24When do you need machine learning
- 18:54Getting started
- 22:22Azure Functions
- 24:49Cosmos DB
- 25:49Cognitive Services API
- 27:44Sentiment Analysis
- 29:04Time Series Analysis
- 29:54Graph Analysis
- 32:35Comparing the tweets
- 35:02Comparing the followers
- 37:02Frequency analysis
- 40:26Aggregate query
- 42:33Sources
related talks
26mLooking at 4 years of web honeypot attacks: tactics, ...
34mBig and Small Data at @Twitter
47mExtendable by Design: Building Generative AI Apps with Postgres and Vector Storage and Azure AI
44mPostgres and OpenAI - a match made in Azure
26mBot Of The United States
38mData and Business
41mInside Twitter's Realtime Delivery of Tweets for Enterprise Customers
28mAdvancing Drug Search with PostgreSQL and Azure AI