Methodology

Objectives

This study aims to conduct real-time analysis of scam-related discussions on Hong Kong social media platforms. By updating the database daily with relevant posts from the BabyKingdom and LIHKG forums, it analyzes information from the past week to address the following research questions:

1. What are the most popular scam techniques?
2. How do scammers exploit different technologies?
3. How do scammers use new communication technologies to reach their targets?
4. What are the linguistic characteristics of scam content?
5. What are the most common scam appeals?
6. What measures can we take to help the public prevent scams?

Data Collection

• Daily Update:
  • Collect scam-related posts from the BabyKingdom and LIHKG forums continuously.
  • Retain only the title, content, replies, posting time, and URL information of the forum posts.

Data Processing

• Batch Processing:
  • Read data from the database for the past week and split it into batches of 10 rows each.
  • Each batch contains a string of 10 comments to ensure analytical consistency.

Analysis Process

• Conduct analysis using GPT-4o:
  • Pass each batch of data to the GPT-4o model for detailed analysis.
  • Obtain analysis results for each batch.

Result Integration

• Review existing analysis results:
  • If no prior analysis results exist, store the current batch’s results directly as the primary analysis outcome.
  • If prior analysis results exist, merge the current batch’s results with the existing ones, then pass them to GPT for deduplication and comprehensive analysis.

Comprehensive Analysis

• Result synthesis:
  • GPT will generate a comprehensive analysis result based on the current batch and previously merged data.
  • Store the deduplicated and merged results as the updated analysis outcome.

Iterative Processing

• Repeat the process:
  • Apply the above steps to each batch of data until all batches have been processed.

Final Output

• Final results:
  • Save the final comprehensive analysis results, ensuring they are complete, deduplicated, and thorough.

By adhering to this methodology, this study aims to provide real-time, accurate, and comprehensive insights into scam-related discussions on Hong Kong social media platforms, effectively addressing the specified research questions.