Project Description
The capstone project assigned to our team is project 63: SmartChat - a virtual cloud bot, which is offered by Tata Communications Limited (TCL). Together with another capstone project (project 62), the company wants a new framework that manages integrated application performances, as the task is getting more and more complicated due to increasing proliferation of cloud computing services they provide for clients. The TCL proposed framework can be decomposed to three modules. The first 2 modules are the focus of project 62, basically application relationship and performance management through data analytics. The third module, which is our project, is to develop a NLP-implemented chatbot to translate natural languages into actionable commands and run corresponding bots to execute the commands. The process flow is as shown in the diagram below.
Background of Team and Industry Partner
There are 6 members in my capstone team, and four of us are from ISTD pillar, and the other two are from ESD pillar, including myself. As this project is IT-based and data driven, thus both ISTD students and ESD students play an important role. Techniques such as data analytics, project management, software development are needed and to be utilised.
The industry partner, TCL, is known as an India-based global telecommunications service provider, and its customers include more than 2000 service providers and enterprises from over 240 countries. More importantly, from October 2014 onwards, Tata started to provide cloud services for their clients, and launched IZO platform, quoted, "a global network platform for enhanced hybrid cloud enablement". However, the increasing proliferation of virtualisation and cloud computing services has added another layer of complexity to application management. Therefore, the company proposed a an integrated application performance management framework, which includes a smart chat front end module to translate NLP to actionable commands and run corresponding bots. And that is basically our project.
What We Have Done
1. Kickstart meeting with industry partner
On 24th Jan, 2018, we had our first meeting with industry partners Ramesh and Prateek (manager) via Skype call, because the industry team was based in India. In this meeting, we got to know more about each other, the project background and TCL's areas of business. We also resolved our questions regarding this project and anything that's not specified or obscure to us. For example, what kind of training data they can provide, or what platforms exactly they wanted this chat bot to be built on, etc. After Q&A session, we also selected the person of contact(Amish for this month) and fixed the time of next meeting, which was around 20 Feb, 2018.
A week after the meeting, the industry mentors sent us a questionnaire to find out how much we know about project-related topics, and they sent back a bunch of reading materials to us on 2nd, Feb.
2. Project scoping
In the meeting with Ramesh and Prateek, we narrowed down the type of chatbot, the focus of this project, and a few open source NLP libraries that we can use, like StanfordCore NLP. Besides, we also proposed a possible logical flow of the chatbot as shown below.