포트폴리오

Profile

Jaedong, Oh

♤ Research Interest: Natural Language Processing (NLP), Data Analysis

♤ Email: ojd9512@gmail.com

♤ Github: https://github.com/Jaedong95 

Education

⊙ 성균관대학교 인공지능융합학과 석사과정 재학 (2022.02.21 ~ ) 

⊙ 한국공학대학교 컴퓨터공학부 소프트웨어공학과 졸업 (2015.03 ~ 2021.02) 

Publication

국제

⊙ Oh, Jaedong, et al. "Are You Depressed? Analyze User Utterances to Detect Depressive Emotions Using DistilBERT." Applied Sciences 13.10 (2023): 6223.

 

국내

⊙ Jaedong Oh, Hayoung Oh. (2022). Detects depression-related emotions in user input sentences. Journal of the Korea Institute of Information and Communication Engineering,26(12),1759-1768.

⊙ Jaedong Oh, Ha-young Oh. (2022).A Comparative Analysis of Personalized Recommended Model Performance Using Online Shopping Mall Data.Journal of the Korea Institute of Information and Communication Engineering,26(9),1293-1304.

⊙ So-Hee Choi, Ju-Ha Kim, Jae-Dong Oh, Ki-Sok Kong. (2020). A Smart Closet Using Deep Learning and Image Recognition for the Blind. The Journal of the Institute of Internet, Broadcasting and Communication, 20(6), 51-58.

Career

⊙ (주)대양이엔지  - 학부 현장실습, 2020.07 ~ 2020.08 

- 파이썬 웹 크롤링 프로그램 개발 

- 카카오 오픈빌더 서비스를 이용한 챗봇 구축 

- 입찰공고 데이터를 활용한 낙찰하한가 예측 모델 개발 

⊙ 데이터마케팅코리아  - 인턴, 2021.04.12 ~ 2021.07.01

- 딥러닝 서버 관리

- 언어 모델 관련 Multi-GPU 연구 

Certification

⊙ ADSP (2021.04)

⊙ 정보처리기사 (2020.08)

⊙ 컴퓨터활용능력 1급 (K) (2018.01)

⊙ SQLD (2016.10)
⊙ DASP (2016.11)

Awards

⊙ Silver award for Smart Closet with deep-learning in a contest for the socially disadvantaged, 2020

⊙ Winning the Excellence Award at the Service Planning Contest Using Abstract Image Data in the NIA Vision Area, 2021

⊙ Winning the grand prize at the Hackathon Competition using artificial intelligence learning data estimation, 2021

Patent

⊙ 텍스트 기반 우울 감정 탐지 모델, 10-2023-0091935 (2023.07.14)

Projects

⊙ Develop DACS(Depression Adaptive Conversation System), 2023.05 ~ 

We propose a chatbot named DACS(Depression Adaptive Conversational System) comprising three key modules to address this social issue. First, the model predicts the user's depression intensity based on users' utterances. Next, it summarizes the user-chatbot conversations to maintain coherence in long-turn conversations. Lastly, leveraging the predicted depression intensity and the summarized context, the response model retrieves the most appropriate and supportive answers. Through this innovative process, our model aims to effectively reduce users' depression levels and provide mental health support.

논문 수정사항 

- 초안 수정 (chatbot -> chatbot named DACS (Depression-Adaptive Conversational System)
- 표 제목 수정 

 

⊙ Building Datasets for Depression Emotional Intensity, 2023.01 ~ 2023.09

⊙ Develop Best-Worst Scaling Annotation Tool, 2022.12 ~ 2023.01 

⊙ Detection of abnormal behavior using CCTV, 2021.12 ~ 2022.02

A project that sends a notification by text and e-mail when abnormal behavior is detected on CCTV. The C3D model was used to detect abnormal behavior, and the UCF Crime Dataset was used for model learning

⊙ ViewCloset Project, 2020.05 ~ 2020.11

A project to create a closet that takes pictures of costumes through voice recognition and delivers them to CNN models to recognize the types, patterns, and attributes of the costumes and print them out as voice. Through the project, our team won the silver prize at the Probono Contest, a contest for the socially disadvantaged.