| Presentation | 2015-03-06 Quality Control in Human-Machine Hybrid Crowdsourcing Toshihiro WATANABE, Toshinari ITOKO, Shin SAITO, Masatomo KOBAYASHI, Hironobu TAKAGI, Hisashi KASHIMA, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | The power of crowdsourcing has dramatically shortened the required time to create accessible content for disabled people with high quality. A number of existing crowdsourcing projects for accessibility use human-machine hybrid systems, where a computer processes the input and human workers correct the output if it is wrong. However, some workers overlook the incorrect output or corrupt the correct one for various reasons, which lower the quality of the final corrected results. Therefore, a quality control method taking account of these workers' behaviors is needed for higher-quality content. This paper proposes a novel statistical model that describes the influence of the machine output on how the workers react, along with an algorithm that infers the correct results from the efforts of the workers. Our experimental evaluations with synthetic and real datasets demonstrate that the proposed method can decrease the number of correction errors compared to the existing methods. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | human-machine hybrid crowdsourcing / label aggregation / generative models |
| Paper # | IBISML2014-91 |
| Date of Issue |
| Conference Information | |
| Committee | IBISML |
|---|---|
| Conference Date | 2015/2/26(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Information-Based Induction Sciences and Machine Learning (IBISML) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Quality Control in Human-Machine Hybrid Crowdsourcing |
| Sub Title (in English) | |
| Keyword(1) | human-machine hybrid crowdsourcing |
| Keyword(2) | label aggregation |
| Keyword(3) | generative models |
| 1st Author's Name | Toshihiro WATANABE |
| 1st Author's Affiliation | Graduate School of Information Science and Technology, The University of Tokyo() |
| 2nd Author's Name | Toshinari ITOKO |
| 2nd Author's Affiliation | IBM Research-Tokyo |
| 3rd Author's Name | Shin SAITO |
| 3rd Author's Affiliation | IBM Research-Tokyo |
| 4th Author's Name | Masatomo KOBAYASHI |
| 4th Author's Affiliation | IBM Research-Tokyo |
| 5th Author's Name | Hironobu TAKAGI |
| 5th Author's Affiliation | IBM Research-Tokyo |
| 6th Author's Name | Hisashi KASHIMA |
| 6th Author's Affiliation | Graduate School of Informatics, Kyoto University |
| Date | 2015-03-06 |
| Paper # | IBISML2014-91 |
| Volume (vol) | vol.114 |
| Number (no) | 502 |
| Page | pp.pp.- |
| #Pages | 8 |
| Date of Issue | |