Jaekeol Choi

I have been a programmer for 19 years at NAVER in Korea.

I received my Ph.D. from Seoul National University, under the supervision of Wonjong Rhee.

I am currently a postdoctoral researcher at the University of New Hampshire, collaborating with Professor Laura Dietz.

jkchoi.naver@navercorp.com / CV / Google Scholar

Research

I am heavily interested in Information Retrieval and Natural Language Processing.

Recently, I have been working on relevance evaluation using GPTs.

Publications

Influence of Prompt Terms on Relevance Evaluation with GPTs
Jaekeol Choi
arXiv 2024
[paper / code]

Isotropic Representation Can Improve Dense Retrieval
Euna Jung, Jungwon Park, Jaekeol Choi, Sungyoon Kim, and Wonjong Rhee
PAKDD 2023
[paper / code]

Finding Inverse Document Frequency in BERT
Jakeol Choi, Euna Jung, Sungjun Lim, and Wonjong Rhee
arXiv 2022
[paper / code]

Semi-Siamse Bi-enocder Neural Ranking Models using Lightweight fine-tuning
Jakeol Choi, Euna Jung, and Wonjong Rhee
WWW 2022
[paper / code]

Improving Bi-encoder Neural Ranking Models with Two Rankers and Multi-teacher Distillation
Jakeol Choi, Euna Jung, Jangwon Suh, and Wonjong Rhee
SIGIR 2021
[paper / code]

Interpreting Neural Ranking Models using Grad-CAM
Jakeol Choi, Jungin Choi, and Wonjong Rhee
arXiv 2020
[paper / code]

Improving FastText with Inverse Document Frequency of subwords
Jakeol Choi, and Sangwoong Lee
Pattern Recognition Letters 2020
[paper / code]