A concept extraction method is presented based on tea dictionary and statistics. The method takes tea dictionary as basis. Firstly, unstructured data source is in Chinese word segment processing, and then, two statistics algorithms are applied to extracted tea concept from Chinese segment results. The approach improves the precision and efficiency of Chinese segment and concept extraction by reducing the time complexity of statistical algorithms with the rich tea dictionary. The experimental results show that the degree of dictionary richness determines the efficiency of tea concept extraction, and can be improved by updating tea dictionary.