The aiming of this paper is to automatically extract the causality knowledge from documents for the contribution knowledge sources of the question answering system. This paper is only concern of extracting the causality knowledge from a single sentence or EDU (Elementary Discourse Units) with two problems of the causality identification and the zero anaphora. Then, we propose the usage of causal verb rules mined from the specified sentence pattern by ID3 to extract the causality knowledge of the single EDU. However, our intra-casual EDU extraction model shows the 0.87 precision and the 0.73 recall.