CSE498, Collaborative Design, Fall 2020
Computer Science and Engineering
Michigan State University

Since its founding in 1999, Vectorform has provided a platform to a wide variety of organizations wishing to solve big problems with inventions emerging at the forefront of technology. This approach has grown Vectorform into a truly worldwide company, with offices in Seattle, Detroit, Mumbai, and Hyderabad. Vectorform combines expertise in hardware engineering, internet of things, augmented and virtual reality, and user-experience focused design to their solutions.

Our Self-Improving Assistant platform acts as a framework to allow Vectorform and its clients to easily create custom support chatbots for any purpose. The system is explicitly designed to be domain-independent, meaning it can be configured to provide automatic customer support in any field, if provided with relevant training data. Each chatbot resides in a Microsoft Teams channel, allowing easy access to anyone in an organization.

To achieve these goals, the chatbot framework takes loosely structured text (FAQ pages, call logs, product manuals, etc.) and uses natural language processing to extract the underlying answers to any questions from a customer seeking support.

Our system is designed to learn and improve as it experiences more customer interaction. As customers ask questions and get answers, the quality of the feedback is collected and analyzed. If a correct solution is achieved, the bot learns that a particular solution is valuable to a certain type of question.

If the user’s question is not answered, the user is directed to human support, and the steps the bot took in its unsuccessful resolution of a question will not be repeated for similar questions.

The Self-Improving Assistant uses a C# .NET Core application which queries Microsoft Azure Cloud Services. MS Cognitive Services processes user queries. The MS Graph API handles permissions and support channels.