| 研究課題/領域番号 |
17K12786
|
|---|
| 研究機関 | 国立研究開発法人産業技術総合研究所 |
|---|
研究代表者 |
LEBLAY JULIEN 国立研究開発法人産業技術総合研究所, 情報・人間工学領域, 研究員 (70757377)
|
|---|
| 研究期間 (年度) |
2017-04-01 – 2019-03-31
|
| キーワード | Knowledge Management / Machine Learning |
|---|
| 研究実績の概要 |
This first of this research has been devoted to two line of works. First, we extended the work done prior to the project. During this year, we build a proof-of-concept platform showcasing the system on small datasets, which was presented as a demonstration and poster CIKM 2017.
Second, we explored novel techniques to infer temporal context information from knowledge graph where such information is typically missing. This led to a Workshop paper presented in TempWeb 2018 (collocated with the Web Conference in Lyon, France), in which we compare various Machine Learning approaches to the problem.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
3: やや遅れている
理由
So far, the work has led to internal as well as external collaborations (namely Nanyang Technical University and Mannheim University). The original expectations for the first line of work have generally been met, however the second line of work have progressed slightly slower than originally expected, due in part to coordination overhead with overseas collaborator, and other ongoing projects. We still think the general goals can be met by the end of the fiscal year.
|
| 今後の研究の推進方策 |
For the coming months, the plan consists in focusing on the second line of work, pursing my collaboration with Mannheim University as well as internal collaboration. We plan to consider deep learning approaches to the problem of (i) inferring both time and spatial context information from knowledge graphs and, (ii) inferring implicit knowledge which would otherwise typically have to be derived through deductive approaches.
|
| 次年度使用額が生じた理由 |
The incurring amount to be used next year will be, as planned, devoted to attended conferences and covering costs in cloud platform usage and development.
|
