Research Projects

Interpretable Machine Learning for Information Retrieval

Graduate Research Project; Advisor: Dr. Yue Wang, Dr. Jaime Arguello Aug 2019 - Feb 2020
-Research on explaining search result relevance by modeling human experts' internal decision-making processes on the biomedical literature searching task
-Proposed a variant decision tree model with better interpretability and comparable prediction performance to sophisticated learning-to-rank models
-Developed intermediate classifiers serving as the decision tree's nodes with manually designed features

Relevant News Recommendation

TREC 2019 News Track; Advisor: Dr. Yue Wang Jul 2019 - Aug 2019
-Research on finding relevant news accurately from a huge corpus given a target news
-Implemented the initial retrieval in Lucene with query terms generated from the news text
-Trained a learning-to-rank model to re-rank the initial result with text features and context features
-Best run achieved a NDCG@5 of 0.59 (2018 best run achieved 0.46)

Semanticlly Enhanced Information Retrieval

Master Thesis; Advisor: Dr. Yue Wang Sep 2018 - Apr 2019
-Research on incorporating domain knowledge to enhance PubMed medical literature retrieval
-Proposed a framework of resource-based query expansion and a learning-to-rank approach
-Comparable to top teams on the TREC 2018 PM track leaderboard
-Recipient of Dean’s Achievement Award (2 out of 95)

Yelp Businesses Success Prediction

Graduate Research Project; Advisor: Dr. Jaime Arguello Jan 2018 - Jun 2018
-Research on predicting future success of restaurants on Yelp after a one-year period
-Worked on multi-level feature generation and selected informative ones from text and numeric data
-Conducted sentiment analysis of 4,736,897 pieces of Yelp reviews with NLTK and Word2vec toolkits
-Corresponding paper was published in PEARC 18 and was invited for a presentation

Professional Experiences

School of Information and Library Science, Chapel Hill, NC, US

Graduate Research Assistant Aug 2019 - Present
-Carry out research projects with Dr. Yue Wang and Dr. Jaime Arguello on Machine Learning, Natural Language Processing and Data Mining in various domains
-Propose & implement state-of-the-art algorithms, design & conduct experiments and write research papers

Credit Suisse, Raleigh, NC, US

NLP specialist Intern Jan 2019 - May 2019
-Implemented web crawlers to automatically scrape data from stock exchange websites like Nasdaq, NYSE and HKEX to detect suspeded stocks
-Retrieved relevant news for each suspended stock from various financial news APIs with summarization and cleaned the text for human reading
-Build sentiment analysis models for financial news with both bag-of-words representations and word vectors
-Build models to analyze how sentiments in a stock's news affected its price trend