Naihao(Neo) Deng

I am a PhD student at University of Michigan working on Natural Language Processing (NLP) at LIT lab.

Previously, I was a research assistant at Westlake University, where I worked with Yue Zhang. I did my Master's at University of Michigan. I did my undergrad at Shanghai Jiao Tong University and University of Michigan.

Email  /  CV  /  Google Scholar  /  Twitter  /  Github

I am interested in the three aspects of NLP.

First, using NLP tools to understand the way human beings perceive the world, and build NLP models that can perceive the world better by studying how human beings perceive the world.

Second, building NLP applications that can help research and real-life applications.

Third, using NLP tools to study social problems systematically.

We propose WildQA, a video understanding dataset of videos recorded in outside settings. In addition to video question answering (Video QA), we also introduce the new task of identifying visual support for a given question and answer (Video Evidence Selection).

We review recent progress on text-to-SQL for datasets, methods, and evaluation and provide this systematic survey, addressing the aforementioned challenges and discussing potential future directions.

We propose the shared task of crosslingual conversation summarization, ConvSumX Challenge.

We report the results of DialogSum Challenge. We find that there is a salient gap between model generated outputs and human annotated summaries by human evaluation from multiple aspects.

We study annotator gender bias in NLP tasks of word similarity, sentiment analysis and NLI in 4 NLP datasets. We develop a robust framework to test for differences in annotation across genders for four benchmark datasets.

We propose to generate SQL queries before users type in the complete text query. We build datasets to study this problem.