2018 Fiscal Year Annual Research Report
A Novel Interactive Information Retrieval System Using Deep Neural Network
| Project/Area Number |
17K12784
|
|---|
| Research Institution | University of Tsukuba |
|---|
Principal Investigator |
于 海涛 筑波大学, 図書館情報メディア系, 助教 (30751052)
|
|---|
| Project Period (FY) |
2017-04-01 – 2020-03-31
|
| Keywords | click model / user modelling / learning-to-rank / optimal transport |
|---|
| Outline of Annual Research Achievements |
This year I focused on developing novel user models and ranking models based on deep neural networks. To better understand users’ web search behaviour, I proposed a novel neural click model, which is more effective in coping with the rank bias issue. To provide high-quality search results, I proposed a novel learning-to-rank model based on the Optimal Transport theory. The above two models have been extensively evaluated using benchmark collections and published in top international conferences (CHIIR2019 and WSDM2019), respectively. Moreover, the source codes are also disseminated via the project website, encouraging the entire community to improve this research towards new stages. Overall, the project has been smoothly conducted step by step according to the research proposal.
|
Research Products
(6 results)
