A Novel Interactive Information Retrieval System Using Deep Neural Network
| Project/Area Number |
17K12784
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (B)
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Web informatics, Service informatics
|
|---|
| Research Institution | University of Tsukuba |
|---|
Principal Investigator |
Yu Haitao 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
|
| Project Status |
Discontinued (Fiscal Year 2019)
|
| Budget Amount *help |
¥3,900,000 (Direct Cost: ¥3,000,000、Indirect Cost: ¥900,000)
Fiscal Year 2019: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2018: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2017: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
|
|---|
| Keywords | learning-to-rank / click modelling / optimal transport / click model / user modelling / User modelling, / Embedding, / Result diversification / ウェブシステム / 機械学習 |
|---|
| Outline of Final Research Achievements |
This year I developed two models based on deep neural networks. The first one is a novel learning-to-rank model based on the theory of optimal transport, which is published at the 12th international conference on web search and data mining. The second one is a new click model for decoding users' search behaviour, which is published at the 2019 conference on human information interaction & retrieval. Based on a series of experiments using benchmark datasets, the experiments have demonstrated their effectiveness for information retrieval. Moreover, I released the open-source project titled as PT-Ranking. PT-Ranking is highly complementary to the previous packages for learning-to-rank. I envision that PT-Ranking will lower the technical barrier and provide a convenient open-source platform for evaluating and developing learning-to-rank models in different fields, and thus facilitate researchers from various backgrounds.
|
| Academic Significance and Societal Importance of the Research Achievements |
The proposed learning-to-rank models shed new light on how to solving the ranking problem. The released open-source project makes it reasonable to envision that PT-Ranking will lower the technical barrier and provide a convenient open-source platform for examining ranking models in different fields.
|
Report
(3 results)
Research Products
(8 results)
