Operating system experienced a rise in number of incidents in recent years. Analysis and reemployment of past solution therefore may make a contribution in reducing service interrupt time and minimizing business losses. The training and retaining of human resources is another primary disbursement source for enterprise. Thus, it is of great significance for enterprises to find reasonable solutions automatically. Combined with keyword tokenization, data mining, numerical optimization and neural network, this paper presents a system that compares and finds the most similar incident solution in the past, based on the description provided by customers in natural language. We try to improve the automated process by increasing the efficiency and accuracy through machine learning methodology and also devote to presenting a practical decision support method.
展开▼
