Envisioned Solution

The final product will be an event-drive machine-learning, intelligent assessor that will be a retainable solution based on our clients existing architecture. EMELIA will correctly categorize system failure reports via textual analysis into predefined categories, with a minimum accuracy of 90% for future reports.

Ultimately, EMELIA will be a reliable and long term solution that will allow our client to maintain their complex communication system.

High-level Requirements

These are the initial high level requirements for EMELIA based on client specifications:

• Command line interface
• Ability to pull from a growing database of previously classified failure cases
• Train a neural network machine-learning classifier
• Effectively classify new failure cases with a 90% success rate
• Test harness feature to explore system performance:
• Ability to change various parameters
• Run on various target sets

The client has stated that technologies such as:

• Python as a programming language
• Anaconda developement environment to support third party dependencies
• TensorFlow library to support graph-based computation

will be useful for creating EMELIA. This system will make use of package management systems and external libraries to provide the functionality that will make this project possible to implement. More information and research on these technologies will be documented in our Technological Feasibility Document listed in our Deliverables page.