Abstract: Autonomous data curation through web scraping has been way more robust than manual data entry by way of tedious research. But what if there was a way to automate webscrapping while also increasing the inflow of attack data from social media sites such as Twitter? Twitter is the number one host of attack alerts for crypto users.
⭐ In this sentiment analysis repo we explore Autoscraper, a Python package to handle the autonomous data entry task.
We also introduce the use of natural language processing models to sift through tons of cryto/Web3 related attack text data propagated on Twitter. The natural language processing approachs are the following:
- attack type detection
- price detection
- name/nomenclature
- chain
- 128-bit hash detection (contract or address)