Machine Learning Approach to Task Ranking

There are variety of methods and algorithms that can be used to overcome the ranking problem. Task ranking is one of the problems that can be solved by using a machine learning algorithm ranking problem. This work focuses on finding the right approach and corresponding algorithms in the process of ranking to be able to help people in determining which jobs have a higher priority than others. Our approach is to compare several algorithms performed in the process of ranking that are Bipartite Ranking, k-partite Ranking, and Ranking by pairwise comparison. We're used questionnaires and deployment of prototype of Intelligent Personal Assistant Agent to apply the appropriate algorithm in intelligence agent in arranging task priority in daily activity that must be done by the users. After training dataset and evaluate the validation dataset using NDCG, it is found that the collaborative ranking used have a more accurate value / lower variance test evaluation because it uses a large dataset and smaller training dataset. We found that labeling for more than 2 values it is not recommended to use a bipartite ranking if there are many repetitive data, both k-partite ranking and rank by pairwise comparison are able to be used for multi-dimensional data labeling.