The construction of knowledge graph of dangerous goods (KGDG) is with great significance of inferring relative information of dangerous goods, developing corresponding policy for its storage and transport, preventing disaster caused by dangerous goods(DG), and providing emergency plan when the disaster happens. Since distributed representation of natural language is an effective method for knowledge representation, we proposed a distributed method of relation extraction for constructing KGDG. We firstly automatically crawled the description of various DG from web to obtain a large corpus. Secondly, we cut the words and represented them by training an embedding vector matrix. Thirdly, we extracted the relation among entities of DG based on similarity of any two words. At last, we compared the performance of relation extraction between co-occurrence and embedding vector. The results showed that our method works well for constructing KGDG.