Past Projects

Designed a website providing generative AI guidelines, educational resources, and trend analysis.
Delivered special lectures in a hybrid format, including “How to Use Generative AI Chatbot”, “How to Write Academic Papers in Science and Engineering”, and “Academic Database Utilization”.
Led the “CAU ChatGPT Prompt Engineering Competition” (2024, 2023) and AI-related seminars.
Organized research performance data, recommended tailored information through the Research Management Information System (RIMS), and analyzed AI research trends for the College of Engineering and Software faculty.

Research on University Library Digital Literacy Curriculum Development (Apr. 2024 - Oct. 2024)

Developed AI literacy curricula, mainly responsible for “How to Utilize Generative AI”, “Introduction to Data Analysis using Python”, and “How to Write a Thesis in Science and Engineering”.
Co-authored a research report and a Korean journal paper with team members.
Received a research project award from the Korean Council for University Education.

Marriott Hotel Concierge TV Voice Assistant (Mar. 2019 - Jul. 2019)

Developed NLU (Natural Language Understanding) NER (Named Entity Recognition) in English for the Marriott hotel concierge TV.
Constructed NER models of hotel service guides, etc.
Registered patents in South Korea, the U.S., and PCT for unifying and improving NLP models.

LG WebOS Smart TV Voice Assistant ‘DeepThinQ’ (Jun. 2018 - Jul. 2019)

Developed NLU NER in Japanese for WebOS TV.
Constructed NER models of TV program guides, etc.
Formulated NER rules using regular expressions and trained NER models with the Support Vector Machine (SVM).
Achieved ~99% accuracy in closed tests and ~85% in open domain tests.
Conducted Quality Assurance (QA) testing before product launches and resolved post-launch issues.

Incheon Airport & PyeongChang Olympic Guide Robot Voice Assistant (Jan. 2018 - Sep. 2018)

Improved NLU NER models in Japanese for the robot.
Constructed NER models of road navigation, etc.
Formulated NER rules using regular expressions and trained NER models with the SVM.
Achieved ~99% accuracy in closed tests and ~85% in open domain tests.
Maintained the models based on users' logs.

LG Mobile Voice Assistant ‘Q-Voice’ (Jan. 2018 - May. 2018)

Supported NLU NER models in Korean for LG smartphones.
Constructed NER models of schedule alert functions, etc.
Formulated NER rules using regular expressions and trained NER models with the SVM.
Maintained the models based on users' logs.

Developed a pilot hotel concierge chatbot, using IBM Watson APIs, and completed the IBM Watson Hackathon.
Designed an AI-based trend prediction system architecture, ‘Lotte Confectionery Intelligence Advisor (LCIA)’, for new product development using IBM Watson Explorer, and systemized requirements by communicating with LC marketers.
Applied social media text mining/analysis models to predict future food trends when categorizing product properties.
Contributed to transforming the new product development process from a top-down to a big data-driven approach, improving decision-making efficiency.
Co-developed an AI-based chatbot, ‘Lotte Shopping Advisor (LOSA)’, for Lotte Department Store using IBM Watson’s deep learning recommendation engine, conversation, and visual recognition APIs.

K-Culture Time Machine (advisor: Jungwha Kim) (Dec. 2014 - Feb. 2016)

Revised and validated an ontology-based Korean Culture Heritage Data Model (KCHDM) to manage cultural heritage databases.
Integrated and enhanced metadata of the Cultural Heritage Administration, Encyclopedia of Korean Culture, and the National Museum of Korea’s ‘E-museum’ into the KCHDM.
Applied KCHDM to an AR-based mobile app describing historical locations through temporal and spatial contexts; registered KCHDM as a standard with the Telecommunications Technology Association (TTAK.KO-10.0850).
Authored a master’s short thesis detailing KCHDM model development.

Exobrain – Korean NLP2RDF (advisor: Key-Sun Choi) (Mar. 2014 - Feb. 2015)

Conducted research on ontology modeling for Korean NLP, Question & Answering (QA) System.
Designed an ontology-based Korean NLP2RDF framework to convert Wikipedia info-box data into Resource Description Framework (RDF) triple format.
Combined morphological analysis, phrase syntax information, NER, and dependency structures into the framework, aligned with the international standard NLP Interchange Format (NIF).
Resolved functional issues in the Korean NLP2RDF framework, specifically addressing the need for word segments due to the unique morphological characteristics of the Korean language.
Published conference papers and a patent in Korea and the PCT.

LOD (Linked Open Data) 2 – Korean DBpedia (advisor: Key-Sun Choi) (Mar. 2014 - Sep. 2014)

Conducted research on multilingual LOD fusion, ontology mapping
Researched QA and Knowledge-based systems, such as ontology mapping to integrate DBpedia with FrameNet, considering DBpedia’s hierarchical structures and FrameNet’s semantic frame structures.
Managed the LOD2 Korean DBpedia project
Published conference papers on establishing a standard format to convert unstructured text into RDF triples.

Page updated

Google Sites

Report abuse