Key Points
- Used Google Trends and natural language processing (CKIP Transformers, TextRank, TF-IDF, BERT) to find the trending topic for the week and designed the approach for compiling news articles
- Developed the system infrastructure and designed a pipeline with Docker and Flask to integrate five open-source repos, keeping the system stable and data flowing between them
Contribution
Researched and designed the approach for compiling articles, implemented Docker for rapid deployment, reducing 2 hours of manual deployment to 20 minutes automatically, and designed a pipeline for integrating five open-sourced repositories with a pipeline architecture for project integrity, data transferring, and future development possibilities.
Honors
- Won 1st prize in Project Competition at the College of Information and Distribution Science, NTCUST
- Won Emerging Technology Application Award in Fi-Award 2023 by the 13th International Conference on Frontier Computing, Tokyo, Japan
Details
- The “AI Editor-in-Chief” collects news from different websites, compiles different viewpoints, and reorganizes them into new news articles, accomplished with web crawling, natural language processing (CKIP Transformers, TextRank, TF-IDF, BERT), and Google Trends
- The “Virtual News Presenter” generates videos of cute anime characters reporting news to make news-consuming entertaining, utilizes speech generating, lip syncing, facial landmark estimating, and Live2d with Unity
- Skills: Project Management · Natural Language Processing (NLP) · Docker Products · Python (Programming Language)