Venter
Venter is an Intelligent Complaint Management and Response Categorization System
Many civic problems fail to reach resolution because of barriers to submitting a complaint or knowing the right person to contact. As a solution to such problems, Venter is intelligent complaint management and response categorization system being developed at IIT Bombay, in collaboration with Microsoft Research.
It aims to provide machine learning solutions and an interactive dashboard interface that empowers our business users – to analyze and visualize any textual data in order to make real-time, data-driven decisions that would enhance the opportunity for meaningful dialogue between citizens and Government.
The main objective of this application is to reduce the human intervention in categorizing the citizen feedback for several domains (road, water, environment etc), thereby reducing the time spent on processing and understanding large volumes of citizen’s inputs. The Venter application is expected to:
- Provide a Web interface and RESTful APIs to its client partners to upload huge volumes of citizen’s text feedback to be classified and understood through the use of a machine learning model.
- Function in a simple, user-intuitive manner while providing secure authentication functionalities, user profile and response category management tools.
- Generate a well formatted document based on the output of the Machine Learning model and fetch it for the user on the interface.
- Implementation of visualizations in the web portal by including a single page dashboard with numerous visualization parameters to analyze the responses.
- Use of Venter portal to extend the service for other ML models and leverage the same to other clients that fall in the same domain of machine learning classification (based on taxonomy heads)
General Design Constraints
Venter will include a web application that includes an interactive web interface as well separately offering ML solutions through a dedicated quasi-RESTful API endpoints based on its client requirements. This application will interface with a web server such as Ngnix suitable to our design and a Python WSGI-compliant HTTP server such as Gunicorn for deployment of Django-based Venter application.
Hardware Requirements
RAM: 8GB Graphics Processing Unit: NVIDIA GeForce GTX 1050 4 GB Device: Standard Laptop 8GB storage
Mandated Constraints
The application will run on Windows/Mac/Linux operating system as a Website application. The technology stack described in section 2.5, was chosen based on the experience with Django (Python) framework and team consensus.
User Requirements
Interactive dashboard for downloading output xlsx/csv files generated by the machine learning prediction model. Response list for a particular category to be represented in the form of stacked cards that are clickable from the bar graph. Interactive multi-variety chart editor to visualize category and response relationship of a particular domain that is printable. Implementation of word cloud component that displays a cloud of top 20-30 important keywords belonging to a group of responses per category per domain.
Tech Stack
| Technology | Tools |
|---|---|
| Web Framework | Django |
| Backend | Python |
| Frontend | HTML, CSS, JavaScript, JQuery, Bootstrap |
| Database (Development phase) | SQLite |
| Database (Production phase) | PostgreSQL |
| Cloud-hosting platform | Microsoft Azure, Google Cloud Platform |
| DevOps | Git |
| IDE | Visual Studio Code, Sublime Text 3 |
Clients:
- Civis
- ICMyC
- SpeakUp