| 研究実績の概要 |
Large-scale text archive data are prevalent in many areas including informatics, computational social science and finance. In this research project, it is aimed at establishing fundamental methodologies for supporting object search as well as relationship search in text archive data. The achievements of each research problem are listed as below:
1. Detecting Semantically Similar Terms across Different Domains. We addressed the problem of terminology gap in object search and proposed a series of techniques and published our work in Journal TKDE2016, WWW2017, Deim2017 and the upcoming Tutorial at PAKDD2017.
2. Explaining Similarities between Terms in Semantic Vector Spaces. We proposed several unsupervised methods to approach this problem and the results are published in IEEE BigData2016.
3. Detecting Cause-Effect Relationships in Text Archive. We investigate how the objects change and what are the effects of these changes. The achievements are published in WWW2016 and Journal TOIS2016.
To sum up, we have conquered several search problems and lowered the accessibility barriers for average users to perform search and understand text archive data. Our methodologies can aid with education objectives to let young generations learn more about knowledge unknown to them such as knowledge about the past. Furthermore, our research works provide computational support for sociologist and historians enabling to construct machines which can automatically examine historical data and generate easy-to-understand results.
|
