COVID-19-Optimal-Resource-Allocation_and_Request-Classification

A solution developed to map essential COVID-19 Relief resources to the needy across a city in the most cost-optimal way, and also to classify incoming SOS messages from those in need of help, for organizational and lesser response times.

Capabilities

1. Optimal Resource allocation : -- This functionality was designed to ingest dataset provided by the government containing the following data :

   • Available COVID-19 sanitary resources like - Hand sanitizers, Face masks, Gloves, Face Shields etc.
   • Emergency medical resources like COVID-19 hospital beds, Oxygen Tanks, etc.
   • Available donations of dry ration items for COVID-relief like - Rice, lentils, vegetables, spices etc.
   • Quantity of Supply and Demand of the above resources across the city
   • The name of the locality/business/firm/entity where the above supply/demand is found, locatable on Google Maps.

   The tool then attaches a geographical tag (latitude and longitude) to each location. Then it builds a graph network with each location as a node and a supply/demand value associated with each. The cost of each edge is obtained from a configurable distance matrix as required. After the previous steps, the tool suggests a list of optimal resource transfers (according to their specific item category) to minimize the gap between demand and supply with the the following fields :

   • From Location
   • To Location
   • Quantity of Transfer
   • Cost of Transfer

   This boils down to a LP (Linear Programming) problem and can be posed in the standard form.

2. Automatic SOS Text Classification

During the COVID-19 pandemic, the end-users are given a free-text field to write and submit their grievances, medical emergencies and relief requests to the state government. This data is collected, pre-processed and each request is classified to one or more of the following configurable categories :

• Travel
• Food
• Medical
• Donations
• Others etc.

This classified list of citizen SOS requests lets the government authorities re-route the requests to the relevant departments to address them with minimal response time.

Here, we use state-of-the-art NLP, Sequence encoding and Deep Learning Techniques to achieve the fucntionality

This Solution was developed and demonstrated to the Dept. of Rural Department and Panchayat Raj, Government of Karnataka, India to be implemented in and around the city of Bengaluru, India.