%A WANG Gang, WANG Han-ru, HU Ke ,HE Xi-ran %T Improved OCCF Method Considering Task Relevance and Time for Task Recommendation %0 Journal Article %D 2018 %J Computer Science %R 10.11896/j.issn.1002-137X.2018.07.030 %P 172-177 %V 45 %N 7 %U {https://www.jsjkx.com/CN/abstract/article_196.shtml} %8 2018-07-30 %X With the development of crowdsourcing system,researchers pay more attention to the crowdsourcing system.Based on the task recommendation of crowdsourcing,most of research scholars convert the behavior data into rate data,without considering the relationship between tasksor the influence caused by the change of user interest on the recommendation results.Therefore,this paper proposed an improved OCCF method considering the task relevance and the time factor to recommend task.On the one hand,this paper introduced a forgetting function when extracting the negative cases,and extracted a certain number of negative cases according to users’ activity.On the other hand,it merged the similarity information of tasks in the probability matrix factorization phase.The proposed method was further applied to recommend tasks in the crowdsourcing system.This paper used the data set of Taskcn to conduct experiments.The experimental results show that the proposed method achieves better results,and effectively improves the quality of recommendation compared with the mainstream methods.