Today 69

Yesterday 95

All 39210991

Friday, 5.03.2021
eGovernment Forschung seit 2001 | eGovernment Research since 2001

Government departments have huge collections of data. In a few of these departments, data had been accumulated over tens of decades. Most of this data which has accumulated is lying in paper records scattered across different offices. However, in the last decade both at central and state governments, several e-Governance projects have been undertaken. Very few IT companies can successfully deliver eGovernance projects and one of them is Suparna Systems. The company has blended big-data and Artificial Intelligence (AI) technologies in its delivery.

Suparna Systems is a bootstrapped software company founded by Satish Kumar who left the high-paying job of Silicon Valley California to pursue the dream of entrepreneurship. This entrepreneur has an exceptional track record of delivering highly successful eGovernance projects. The work of transparent toggling of mobile applications between online and offline mode has been widely adopted in the delivery of services in rural areas where typically telecom signals are poor. His work on online citizen services delivery has inspired many cities and government departments to go online and provided a template for successful implementations.

The entrepreneur played a very important and key role in developing technologies for COVID-19 State War Room in Bangalore, Karnataka. In the initial days of State War Room Satish has involved in evaluations and selections for different technologies offered by Startups and big companies primarily in AI/ML for suitability of their use in fighting COVID-19 and later on architecting different solutions.

In one to one conversation Satish, tells us about successful implementations of AI in Public Sector Organisations and E-Governance and how AI can be used to fight COVID-19.

Question: Last year you had a very successful Enterprise Resource Planning Implementation in Hubli Electricity Supply Company, an electricity distribution company. Can you tell us where you used AI/ML ?

Answer: One of the module implementations of Transformer Management in our ERP is a lot like Robotic Process Implementation (RPA). We took the concepts of RPA and then selected bits and pieces and implemented them at various workflow stages. In fact, our whole workflow implementation process itself is RPA.

The other place where I used AI is in building the capacity of my team for capacity building. We developed a Natural Language Processing (NLP) based chatbot. The chatbot was able to answer the questions of our support people on various business processes and how to implement them. It also provided answers to which senior engineer to talk in case of doubt.

Question: AI/ML is a lot about learning algorithms. Can you tell us about their use and real-life implementation ?

Answer: We used two sets of learning algorithms in Complaint Management Systems Implementation. The first set of implementation was an ML algorithm where we fed the complaint titles to train the classification of complaints.

The second one which we implemented was not a textbook example of Learning Algorithms. Most of the textbook examples are based on training the algorithm first with a huge amount of data. What we faced problem was of the correct routing of complaints. In Government departments and Public Sector Organisations (PSUs), there are many many supervision and management layers however the complaint needs to be routed to executing staff. Each of the executing staff has a jurisdictional area. Complaints are usually received at levels several up. Here the learning algorithm was used in creating and self-correcting the mapping table. Once a complaint is routed to an executing office, any complaint coming from that jurisdiction gets automatically routed to that office irrespective of at what level the complaint was received. In the case of bifurcation of executing areas, the system automatically learned the new jurisdiction based on mis-routings.

Question: Tell us something about the use of AI/ML in your COVID-19 work.

Answer: COVID-19 was itself a very evolving and changing situation every day. In the beginning, there were very few known things and the focus was on knowing the infections and tracing the contacts. There are many many data sets in the COVID-19 State War Room. For geographical jurisdictions again there are different data sets by different departments. We implemented record matching and standardized on the code for jurisdictions. That helped us in architecting and evolving our Contact Tracing Systems both Mobile Apps as well as Web-Based Systems as well as other developing IT Systems.

We took inspiration from many outstanding work done at different places like John Hopkins Dashboard and New Work Times Dashboard.

Question: Where you see AI/ML can help most?

Answer: If we see where it can impact most is COVID-19 Vaccination right now. Vaccination is the program on which everybody is keeping a lot of hope. Big-Data and AI technologies if used properly can make a huge difference. One of the effective ways is to create a Data Lake and dumping various datasets about COVID-19 infection rates, Population Datasets, Critical Vulnerable group datasets (EHR Records if possible), Vaccination Providers, etc. Making this information available to local health authorities through Data Catalog and Self Service Reports and Visualisations is going to add real value. The curated and cleaned data can be used for the selection of populations that can be vaccinated in the first group, second group, etc. Real-time data of travel, hospitals can be combined to get the prioritized vaccination jurisdictions. This data set contains private individual data, so all security and data governance should be properly implemented in the Data Lake.

AI-based prediction models can be used to set the target and then real time monitoring of vaccinations. AI can help in knowing the root cause of the mis-match between the targeted vaccinations and real vaccinations done. Machine Learning can be used for target setting and achievement and the model will get more and more perfect as we will have more vaccination data.

In case of adverse reporting of vaccine dose, the correlation of data can be used across population groups to identify individuals with similar conditions (In terms of age, health record, etc) and then alerting and monitoring them. Chatbots can also help in answering questions about locations of vaccine providers, doses, and eligibility etc.


Quelle/Source: The Week, 23.12.2020

Bitte besuchen Sie/Please visit:

Go to top