Abstract: Based on the most core HDFS and MapReduce in Hadoop distributed system architecture，a rapid extraction method of table information for massive geological documents is proposed.In order to improve the extraction speed of geological information document form，first of all，using the key WORDs to find documents stored in the HDFS root directory，then，using the Hadoop distributed cluster Map function and a Reduce function reduction cell information extraction and information，according to the mineral resources potential evaluation result data in Chongqing in WORD document form rapid extraction experiments.It is proved that the method of rapid extraction of geological document table information in this paper can greatly reduce the time needed to extract the information of the traditional single-machine serial geological document form.