The project aims at building a generic, consolidated and comprehensive railways incident database which contains data from the historical incidents from a railway company.

The database shall contain the keywords and attributes of the incidents through the data tagging process, which will then be utilized to train and build an AI system with Natural Language Processing (NLP) technology to understand semantic meaning of the keywords.

Users can use the AI system to search historical incidents by a flexible and unstructured way, such as an equipment name or a high-level summary. The AI can extract similar cases that have close proximity instantly. It can also identify "hidden" symptoms of the incidents based on the correlations of the attributes, perform trend analysis and generate advance visualization reports.

The project shall also derive a long-term template for the railway company to submit incident data.

Project Challenges:

• Identify the key words and attributes from various sources such as Excel files and detailed PDF reports, and input data to construct the database.
• The AI shall be able to find out "hidden" symptoms of the incidents which are not apparent from the superficies.
• Generate insightful visualization reports to show the correlations and trend analysis.