On Rejecting Outliers by Smirnov Test for Massive Data Input System in Vegetable Geographical Origin Identification System

Abstract

As co-work with agricultural chemists, we developed a vegetable geographical origin identification system using trace element composition to solve cultivated place forging problems on vegetables in Japan. Vegetables cultivated in farm fields absorb trace metal elements from soil. Here, trace element means that very small quantity elements in chemistry. Since trace element compositions in soil differ from districts, trace element composition of cultivated vegetables also differs from districts, even farms. Our system can identify geographical origin of vegetables in short time by comparing trace element composition of doubtful vegetable and genuine vegetables. Therefore, trace element compositions of vegetables must be measured when they are shipped from farms. Normally, since fresh vegetables are sold in the same day of shipping, trace element compositions must be measured and recorded into databases in few hours, before they will arrive at market, sold to general consumers. In this paper, we describe a massive data gather system for our system which rejects outliers. In addition, we describe an interface between our system and measurement instruments that are PC/WS based but closed system.