SNU Ambient NLP Lab

The Ambient NLP Lab of SNU GSDS aims to investigate BERT-based models that can be pre-trained from vast amounts of language data, and explore their possible applications. Additionally, we are developing a pre-trained model capable of analyzing the eligibility of a given legal/contract document. Resources specific for Korean language processing are also available: KR-BERT, KOSAC(KOrean Sentiment Analysis Corpus)

Main Topic

Natural Language Processing for Ambient Intelligence

We are examining potential areas where concepts related to NLP could be integrated in Ambient Intelligence(AmI). Combining NLP and AmI can open up doors for novel research topics with the possibilities of challenging state-of-the-art methods in both fields. Service oriented computing, context aware systems, and natural human computer interfaces are just a few examples of areas that are currently being studied. We believe that AmI could be the catalyst for NLP researchers producing ground-breaking discoveries.

Deep Learning-based Language Processing

We are heavily working on word/contextual embeddings such as ELMO and BERT. We are developing a sub-character BERT representation for Korean noisy user-generated data.

Sentiment / Opinion Analysis

We have been working on (Korean) Sentiment/Opinion Analysis. We have completed Korean Sentiment Analysis Corpus (KOSAC)