Introducing Right Relevance Beta

Today we are unveiling publicly a brand new service, Right Relevance Beta, whose mission is to provide
our customers with timely and highly relevant information in any topic of interest. It marks an exciting milestone for our users and the team behind our innovations. Seeing how any of our consumers can discover information and influencers in areas of deep topical interest, in ways that make a difference in their personal and professional lives, helps us reflect on the journey that led us here..

Sumit and I have known each other for 20 years and been planning to build something we truly care about for quite a while. Nothing compelling really tugged at our hearts till early 2013 when we started discussing a problem that has been plaguing us for years with no obvious solution in sight.

Both of us are voracious readers, with different tastes, not just news but fine grained topics like behavioral finance, information retrieval, cognitive neuroscience, monthly newsletters from prominent hedge fund managers, latest articles from people whose opinions we truly care about and the ilk. We were using a variety of ways to find the latest and most relevant information (articles, videos, opinions etc.) on the topics of our interest from bookmarking URLs, to Wikipedia, to googling for articles from people we cared to read, to feedly like RSS curation, to flipboard and more recently curating twitter lists. None of these truly catered to our needs and even with the time and energy we spent, relevance, dynamism and missing out on something valuable were still serious concerns.

 Sumit, with his search background, had been experimenting with a private Hadoop project involving indexing books to provide recommendations using dbpedia data (freely provided Wikipedia extract) in his spare time. As we discussed using authors and their expertise in a genre to find recommendations, it dawned on us that we can scale the paradigm to any domain and not just books. The most basic premise is that whenever we’re really looking for very specific information, the first thought that comes naturally is to seek out the source (person or website), that has the most knowledge, expertise and experience on the topic in question. We decided to index all living experts aka influencers and partition them based on fine grained topics of interests. We started with Wikipedia but Twitter proved to be the most effective source and by Fall 2013 we had a prototype ready which was compelling enough for us to leave our jobs and bring the vision to reality.

Over the last year, we’ve come a long way from conception to delivering our public Beta on the web and the iPhone. It provides the ability for end users to search and discover highly relevant information, which includes articles, conversations and influencers, in over 35K areas of deep topical interest, with context and opinion from over 2M influencers curating that information. We’ve made major inroads into the critical discovery and relevance problems along with providing a real-time push experience. The user interface and experience are constantly evolving as we deal with the challenge of expressing our technology to users in a simple way yet making the underlying power obvious. But we’re making progress in explaining what defines “relevance” in our platform with very positive feedback. We provide automated personalized recommendations based on current interactions esp. on Twitter as we realize how difficult it is for someone to list their interests especially on request. We’ve also thrown in some basic insights we’re working on and would be adding much richer analysis as we go along.

The initial idea seemed conceptually elegant and natural, and, dare we say a little too obvious. The engineering proved to be not that obvious and the gamut of technologies leveraged to make this happen boggles our mind. The ability to parse through several terabytes of data and ~50M tweets a day would not have been possible if not for all the brilliant and amazing people out there putting out open source and freely available technologies like hadoop, storm, lucene/solr, mongo, play framework, bootstrap and many others. Personally, we love the technology, and the continuously evolving end product. The ability to share it with everyone is one of the biggest pleasures of this journey for us. Please give it a try and if you find value, share it with others who you feel would be interested. We hope our users will enjoy it just as much as we do and help define how the end product should look like as we march on from Beta to a full release that we can truly be proud of.

Write a Reply or Comment

Your email address will not be published. Required fields are marked *