GitHub Repositories

​This repository contains the Twitter scraper I used for my dissertation, "Information Sharing for Validation: Evidence from Social Networks."

The scraper includes four main functions:

• 'scrape_request': Extracts the user tag, ID, main post, and the number of replies, retweets, and likes based on a search query.
• 'scrape_replies': Extracts the user tag, ID, main post, and the number of replies, retweets, and likes of the replies, using the ID of the original post.
• 'scrape_retweeters': Extracts a list of retweeters of a post, using the ID of the original post.
• 'scrape_likers': Extracts a list of likers of a post, using the ID of the original post.​

Code Repository for the Twitter Scraper

​This code repository supports the paper "How Much Should We Trust LLM-Based Measures for Accounting and Finance Research?" It contains code scripts for the experiments, input data, and results. The scripts are easily customizable to accommodate other LLMs and datasets for your use.

Code repository for the Experiments

---

Code Scripts for Finetuning LLMs

1. Sentiment Analysis Task with Financial PhraseBank Dataset

1. Finetuning ChatGPT (and Llama2 with QLoRA)
2. Finetuning RoBERTa and FinBERT

2. Linguistic Complexity Analysis Task [Coming Soon]

---

Fine-tuned LLMs

1. Sentiment Analysis Task with Financial PhraseBank Dataset

1. Fine-tuned RoBERTa
2. Fine-tuned FinBERT

2. Linguistic Complexity Analysis Task [Coming Soon]