【Purpose/significance】The era of big data is coming, the amount of unstructured data such as text, image, videoand audio in library increases very quickly, with the new characteristics such as huge volume, diversified data structuresand mobility of library big data, traditional rational databases in library can no longer meet the requirements of big datamanagement and analyses.【Method/process】In this paper, we propose an unstructured big data analysis and decision mak-ing system for library based on Hadoop.【Result/conclusion】This system can analyze large amounts of unstructured data inlibrary quickly, process large amounts of unstructured data to discover the hidden value, discover knowledge from unstruc-tured big data mining, and can provide decision making and service support for readers in library.