Paper Title
Cultivation Support With Extraction of Farming Work Rules

Abstract
An experienced Japanese farmer expects to cultivate high quality crop for market. However, he often performs a farm task mainly according to his own intuitive observation of the vegetable field. Since the observation is qualitative, not quantitative, he gives an inadequate amount of farm task to the field. That causes him fail to get high quality vegetable. The paper proposes a system to assist the farmer doing cultivation. Sensors are plunged into the field soil to take quantitative information of soil humidity. At the same time, quantitative information of weather is given to the system. The farmer inputs the evaluation of the product when he harvests the vegetable. The system extracts good cultivation rules with a decision tree algorithm. During the farmer cultivates a new crop of the same vegetable, the system recommends a specific farm task with a specific amount to the farmer, enables him to achieve high quality vegetable. The preliminary experiment is conducted on 3 soil ridges where a farmer plants mustard spinach. The results show it is able to extract good cultivation rules. Keywords� Agriculture, Machine learning Decision tree algorithm.