Mumbai-based data analytics and customer engagement platform Locobuzz launched a COVID-19 resource website on May 5 to aggregate crowd-sourced leads on Twitter. “ While we talked to NGOs, we realised they did not know who was seeking help. They needed a repository of information,” said Co-founder Vishal Agarwal.
Locobuzz’s COVID relief platform uses artificial intelligence, machine learning and big data to collate relevant real-time data from Twitter and presents it in a compact, user-friendly format. “We want to ensure that the gap between those looking for assistance and those who can supply it is bridged,” added Vishal.
The platform covers a wide range of geographies, providing the latest confirmed leads for oxygen, beds, plasma, medicines and critical supplies for COVID-19 patients, based on the device’s location or manually entered location.
Earlier, the team had built an automation solution with an AI-based bot. Every time a user posts a tweet seeking help for COVID-19 resources, a Locobuzz bot will automatically respond with the most recent, relevant information and sources under the same tweet. However, the service was withdrawn after Twitter intervened citing it didn’t comply with their policies.
Co-founder and CTO Nitin Agarwal, along with his tech team, used ASP.NET Core to build the COVID Relief platform. For the database operations, the platform uses Elasticsearch. “We use it (Elasticsearch) because our main aim has been to serve information in a real-time basis. With the normal RDBMS approach, the search becomes sluggish and slow. Elasticsearch allows collection of millions of data points without affecting the performance,” Nitin said.
The team has created a dashboard using Angular. When a user searches for resources in a particular city, the platform uses Twitter streaming APIs to collect information related to COVID-19 sources; sorts them based on cities; and finally, processes and consolidates them in real-time. It then categorises and separates tweets (demand from the supply) based on specific keywords.
The tech team has built an internal mechanism to rate authenticity of users who post help on Twitter, and to tackle frauds. First, it gives more importance to verified handles. Secondly, it leverages Twitter engagement statistics to understand which tweets are more reliable and authentic. Thirdly, it prioritises tweets based on ratings.
“It is similar to how Google Maps function,” Nitin explained. “When a street is closed, users travelling through that street report about the closure. Google does not send people to survey it. Similarly, we have built ‘Helpful Sources’ and ‘Not Helpful Sources’ buttons to rank tweets,” he said. Therefore, tweets with the maximum number of ‘Helpful’ votes appear at the beginning and are arranged in descending order.
The platform uses Regex to extract contact details from tweets. Users can click on the contact number provided on tweets and will be redirected to the device’s dial-up.
What’s unique?
Locobuzz’s platform provides comparisons on the supply and demand of resources, helping NGOs and governments track trends. It compares the demand and supply of resources for two days to provide on-ground insights. “It helps see if the situation on the ground is changing for the better or worse,” said Co-founder Shubhi Agarwal.
Additionally, the platform supports six languages — English, Hindi, Kannada, Gujrati, Tamil and Marathi.
While the COVID-19 relief platform is functional in Delhi, Pune, Bengaluru, Mumbai, Nagpur, Chennai, Nashik and Lucknow, it also pulls relevant content from Twitter and provides it to users looking for resources in other cities.
According to Shubhi, the platform organically grew from zero users to 8,000 users within the first ten days of operation. Since May 5, it has recorded 70,000 visitors, each spending an average time of 1.20 minute on the website.
You can access the website here.
