Presentation 2019-02-22
Active Inference of Variable Labels Using Annotation to Some Variable Labels in Data Jackets
Hiroo Iwanaga, Teruaki Hayashi, Yukio Ohsawa,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Data Jacket (DJ) is a framework for sharing summaries of the data and encouraging collaboration and utilization of data in the market, with the data itself hidden. Variable labels (VLs) are described in DJs as the name or meaning of variables in the data. While VLs are the key information for planning data utilization or data collection, DJs have missing VLs because the registration rule of DJs does not force data holders to enter all the information about variables in the data. To infer likely VLs in DJs which are not explicitly in the outline of data, matrix-based method (MBM) are proposed. However, MBM is difficult for DJ users to reflect their prior knowledge or requirement about variables of the data (e.g. “some VLs should be included in the DJ”). In this paper, we propose a method for inferring VLs using ODs and annotation to some VLs, extending MBM to be applicable for active learning situations. Our method enables DJ users to update the inference results as needed by adding the annotation to VLs.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Data Jacket / Variable Label / active learning / partial annotation
Paper # AI2018-44
Date of Issue 2019-02-15 (AI)

Conference Information
Committee AI
Conference Date 2019/2/22(2days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Tsunenori Mine(Kyushu Univ.)
Vice Chair Daisuke Katagami(Tokyo Polytechnic Univ.) / Naoki Fukuta(Shizuoka Univ.)
Secretary Daisuke Katagami(Ritsumeikan Univ.) / Naoki Fukuta(Univ. of Electro-Comm.)
Assistant Yuko Sakurai(AIST)

Paper Information
Registration To Technical Committee on Artificial Intelligence and Knowledge-Based Processing
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Active Inference of Variable Labels Using Annotation to Some Variable Labels in Data Jackets
Sub Title (in English)
Keyword(1) Data Jacket
Keyword(2) Variable Label
Keyword(3) active learning
Keyword(4) partial annotation
1st Author's Name Hiroo Iwanaga
1st Author's Affiliation The University of Tokyo/NTT DATA Mathematical Systems Inc.(UTokyo/NTT DATA MSI)
2nd Author's Name Teruaki Hayashi
2nd Author's Affiliation The University of Tokyo(UTokyo)
3rd Author's Name Yukio Ohsawa
3rd Author's Affiliation The University of Tokyo(UTokyo)
Date 2019-02-22
Paper # AI2018-44
Volume (vol) vol.118
Number (no) AI-453
Page pp.pp.39-44(AI),
#Pages 6
Date of Issue 2019-02-15 (AI)