Tag Recommendation System for Data Catalog Site of Japanese Government

Yasuhiro Yamada

2023

Abstract

This paper proposes a tag recommendation system for a data catalog site of the Japanese government. The site publishes datasets that include files containing statistical data, government documents, and other files of the Japanese government. These datasets also each include the title, description, publication date, and tags, where a tag is a single-word or compound term which represents the content of a dataset. The system uses multi-label classification in machine learning to recommend tags for the datasets; multi-label classification is a method that outputs multiple tags for each input dataset. There are many tags already in datasets hosted on the site that appear infrequently. It is difficult to predict such infrequent tags from the datasets by multi-label classification. To deal with this problem, we use an existing oversampling approach which increases the data of infrequent tags in a training dataset for the learning process of the multi-label classification.

Paper Citation

in Harvard Style

Yamada Y. (2023). Tag Recommendation System for Data Catalog Site of Japanese Government. In Proceedings of the 15th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management - Volume 3: KMIS; ISBN 978-989-758-671-2, SciTePress, pages 325-331. DOI: 10.5220/0012260000003598

in Bibtex Style

@conference{kmis23,
author={Yasuhiro Yamada},
title={Tag Recommendation System for Data Catalog Site of Japanese Government},
booktitle={Proceedings of the 15th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management - Volume 3: KMIS},
year={2023},
pages={325-331},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012260000003598},
isbn={978-989-758-671-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 15th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management - Volume 3: KMIS
TI - Tag Recommendation System for Data Catalog Site of Japanese Government
SN - 978-989-758-671-2
AU - Yamada Y.
PY - 2023
SP - 325
EP - 331
DO - 10.5220/0012260000003598
PB - SciTePress