Project Submission for UMKC Hackathon Spring 2017
- Raj Anantharaman (rah59@mail.umkc.edu)
- Bill Capps (bjctgc@mail.umkc.edu)
- Adam Carter (arcbpc@mail.umkc.edu)
- Amy Lin (clthf@mail.umkc.edu)
For our project, we chose the CoxAutomotive/VinSolutions Problem, which presented the following use case:
- Natural Language Processing (NLP) has many uses in both academia and the business world. Parsing the user messages/mails could be one of the applications. We would like you to develop an appointment detection system just like the one you see in smartphones from GMAIL and Office 365. The goal is to allow the user to type a message to another user. When that user opens the message content via a browser, the system should detect texts that infer a meeting invitation.
This problem statement defined the following tasks:
- The application should allow users to type a message to a different user.
- The application should allow users to read the messages they have received from other users.
- The application should highlight, via hyperlink, any text you have detected as a possible appointment.
- When users click on that hyperlink, it should display a popup screen that pre-populates with the detected meeting details (When, Who). Do not worry about location (Where).
Our solution comprises the following elements:
- MongoDB for data storage
- REST Service written in Java using the Spring Boot Framework.
- NLP Date Parsing in Java using the Natty Engine ( Alternative approach : Python Parser. Didn't end up using it but is still a good way to handle NLP Data Parsing.)
- Web Client using HTML5, CSS, Javascript and asp.net MVC.
The REST API is documented on the project wiki site here.
For alternative approach using Python Parser, a brief documentation on the project wiki site is here.