In this paper we investigate the usage of machine learning and text cavity. Growing database of almost every area of human activity has led to powerful new tools use for changing the data to the useful knowledge to increase. To satisfy this need, researchers in various areas such as machine learning, pattern recognition, statistical data analysis, visualization of data, neural networks, econometrics, data recovery, explore methods and ideas. Text cavity uses unstructured text information and study it to explore the hidden structure and connotations in the context. This article is about the extraction of useful text from word for word and irregular and unstructured text descriptions and correct the errors caused by these descriptions. This research emphasizes the development of machine learning technologies for knowledge discovery obtained from text description of the error detection of a specific environment, in particular, we focus on machine learning algorithms to detect documents that contain descriptions of systematic failure and root cause the errors and Finally, we offer some suggestions and recommendations.