| ABSTRACT:In recent years, rail transit has leapt forward in China, while taking up the transport task in several major economic zone and cities in China, rail transport face greater challenges in both the growth of passenger and security. Studies conducted to assess the state of rail safety have important theoretical and practical significance on the protection of personal and property safety of the public, promoting urban development, enhancing level of service in rail traffic and reduce accidents events.In rail transportation safety information, such as text, images and video, text data is extensively used which has high mining value. However, the current manual handling of text data is quite slow and has low amount for publishing, also not working well with structured data. As one significant point, through the critical study of techniques of text data mining, the analysis result and the safty indicators of word frequency is obtained and working with safety indicators based on traditional structured data when using machine learning techniques, a new perspective on the comprehensive assessment of the state of traffic safety is made to provide a theoretical basis for rail traffic management departments. The main work is as follows:1) Study on the relevant departments of the rail transit for the safty information, and analysis the character of safty information of rail transit。2) Study the techniques of Chinese word segmentation, pick up the apposite algorithm and dictionary for it. Construct the safty indicators of word frequency based on text data.3) Proposed fusion algorithm based on association rules for the two kinds of indicators, build the rail safety assessment indicators.4) Based on rail safety assessment indicators, proposed the algorithm of assessment of rail safety based on SVM and built the model of it.5)) Based on the above results, make use of Python for each algorithm implementation, and build the platform of comprehensive rail transit safety assessment. |