Multi-lingual social media platform – Koo – has rolled out an exciting in-app feature ‘Topics’ across 10 languages. Topics offers a highly personalized experience to multi-lingual users. Koo is the first and only platform to enable this feature in 10 Indian languages – Hindi, Bangla, Marathi, Gujarati, Kannada, Tamil, Telugu, Assamese, Punjabi and English.
Being an inclusive platform built around a language-first approach, Koo has a diverse demographic of users, including millions of first-time creators who actively express themselves through poetry, literature, art & culture, sports, movies, spirituality, among 100s of other themes. Through Topics, users get to view only the kind of the content which is most relevant to them, thus making their journey on Koo far more meaningful and enriching.
Amidst numerous conversations that take place on Koo, Topics makes it way easier for users to pick and choose content as per their interest and preferences, instead of scrolling through the feed on the platform. A user seeking news and information related to ‘health’ (for example) can click the ‘health’ section under the Topics tab to consume all relevant Koos pertaining to vaccination, lifestyle diseases, healthcare tips from medical experts, etc.
Says Mayank Bidawatka, Co-founder, Koo, “We are proud to be the first social media platform to launch topics across 10 Indian languages. This feature helps users find content they are interested in and helps many creators get discovered by relevant users. We have over 20 millions topic follows every month, showing the relevance of this feature to users. We achieve topic classification through complex machine learning models that have a very high level of precision. We are proud to have mastered such complexity in a short span of our existence. I foresee over 100 million topic follows every month by the end of this year.”
Harsh Singhal, Head of Machine Learning, Koo said, “Topics in multiple languages is a combination of many state-of-the-art Machine Learning and Natural Language Processing (NLP) techniques. NLP technologies for Indian languages do not enjoy the extensive ecosystem that is available for English. Koo innovated in a variety of areas to implement Indian language NLP tasks to build Topics across Indian languages. The Machine Learning team at Koo trained LLMs (Large Language Models) and some of the most complex neural network architectures to extract important entities being discussed in a Koo. Koo probably has one of the widest variety of subjects being discussed everyday in India. Given this reality, achieving what we have is a huge deal for India. The exciting thing is that this is just the beginning for us!”
Koo recently registered 45 million downloads, marking a period of hyper growth, having clocked 10 million only a year ago. “Koo aspires to attain 100 million downloads in the future, and build technology which can empower native speakers everywhere in the world. Like India, about 80% of the world speaks a native language. Being a platform from India, Koo understands the nuances and ethos of multi-lingual societies, and our technology can do India proud at a global stage,” adds Bidawatka
Topics reflects the kind of conversations users across the 10 languages are having on Koo at any given point in time, with the most popular topics making their way under various categories (like health, education, environment, movies, sports), eminent personalities, organizations (like Isro, IMF, etc), places, (states, cities, countries that are in news) and a host of other trending topics.