Music accounts a significantly large part of online activity these days which can be attributed to the availability of various online music stores, streaming services, news and podcast services, social networks, and even cloud-based personal music collection. With increasing digitization of music and the industry itself, it opens up immense opportunity to determine inside analytics like which are the emerging trends, discover new talents, and be able to market your music more efficiently (based on the likes and dislikes of demographics).
Shubhanshu Gupta, who has just completed his Under Graduate studies in Information and Communication Technology from Dhirubhai Ambani Institute of Information and Communication Technology. He has been involved in an array of things, from web and app development, to research projects which have provided unprecedented solutions, and then he is also the co-founder of The College Store. He also presented his paper in Big Data Conference at Harvard University.
Let’s know more about his research !
Big Data and Machine Learning
What is your research about ? Explain with some analogies and/or examples
- In days of hard record sales (disc, cassette tapes, records) it was easy to keep track of music sales whereas difficult or impossible to track number of times they were played by the listeners in their music systems. As music became increasingly distributed, played and discussed online, it has become possible to individually keep track of various aspects by analyzing the data in near real-time. However, with millions of songs, artists, events touched by potentially billions of listeners online; the resulting online activity opens up vast avenues of research for data science community like predicting and suggesting algorithms for music style prediction, music genre, music mood, music recommendation, or even predicting which song is going to be the top hit even before it releases into the market.
- Hit Song Prediction: There are machine learning algorithms which can be employed for predicting the success of songs even before they are released in the market, referred to as the Hit song science. In this, accurate models are built to predict if a song would be a top 10 dance hit or not. Same goes for music emotion, music recommendation, music mood, music genre prediction. The difference in each one of them being that the dataset used to change and the algorithms and techniques used to change.
How is your research unique ?
- The kind of analysis which involves proposing an in-depth survey of existing algorithms along with the panoply of wide algorithms and the short comings of each one them have been proposed in my research, some of which work best for various music use cases. The various algorithms revolve around machine learning, information theory, social network analysis, semantic web, and linked open data which have been implemented for various music cases over humongous datasets and have been represented in the form of taxonomy. All this has not been done before.
How does your research help the people in industry or in academia ?
- Basically people in the music data analytics industry (Some very famous companies include next Big Sound, echonest, spotify etc) and academia have long been waiting for such a current state-of-the-art in music related analysis. 5 exabyte (5000,000,000 GB) of data was generated by 2003, and these days, this much amount of data is generated in just 2 days. The digitization of music industry is one of the biggest contributor to this compounding growth of data. The music industry needs the data science industry to make sense of this data so as to help drive sales, pay proper royalties to artists, and most importantly know how people engage, consume, purchase music and their listening habits and their tastes. And there comes the contribution of my research. It proposes various algorithms which makes sense of music related data and give meaningful insights.
Where did you get the inspiration about the research (A professor who guided you, or maybe you got inspiration from a peer, or a senior. Describe a little) ?
- I was always inclined towards a career in research and development. But at initial stages of undergraduate studies, it’s a tad bit difficult to formulate your own research statement and get excited towards working in that direction for coming years (yes, research takes enormous patience and perseverance. You lose your will power or interest and then it’s very difficult to come back). But there comes the role of your guide or advisor. My advisor is the R&D head of promusicDB and when I came in touch with him, he explained me the problem statement and that, it has not been solved uptill now. That was sufficient for me to take off for a long-haul flight.
Any advice for newbies in this field of research ? Any good competitions/conferences the students should apply for ? How to take a start with research ?
- I would suggest all my juniors, if they are already involved in some kind of research project, to be very patient and continue doing the good work. It’s very easy to lose interest very quickly but if you are really passionate about the problem you are solving, then that shouldn’t be an issue.
- For all those who are yet to start on their research career or a project, I would suggest that the first task should be to find a really good mentor or a guide. A guide is the one, who is going to hand-hold you, keep you motivated during the entire course of your research. He/she would also be the one, who is going to give you really good problem statement to pursue or else would help you solve your own problem statement. Second of all, is you are really passionate, stick to what you are doing (though, this remains valid as long as you feel that the project you are working on is worth the time you are spending on it). The reason why I said this is, people tend to switch projects very quickly because of many factors like boredom, lack of interest and what not. Whereas once you get into research, it requires patience and diligence without which no outcome should be expected.
- Don’t worry about conferences. There are plenty of them taking place every now and then. Concentrate on getting a quality output (not necessarily a publication, though).
Any outstanding achievement or recognition for the project/research?
- Was the only Indian, also the only Under Graduate (this made me the youngest presenter as well) in Big Data Conference at Harvard University which had an acceptance rate of a meager 8.5%. The paper that I presented can be seen here: http://www.ase360.org/handle/123456789/163
- Also, I was covered by a national daily: Gujarat Samachar. http://www.gujaratsamachar.com/index.php/articles/display_article/future-of-the-music-industry-data-analysis-to-prepare
How did you get selected for your paper at Harvard?
- They basically release a “Call for Paper”, popularly known as CFP in research community which is like a window to submit your research work. They first asked me for an abstract of my research. After it’s screening and short listing, they called for the full paper. There is a program committee which consists of a pool of the most elite professors from universities in all parts of the planet (and that depends how much prestigious is the conference). The paper is sent anonymously to a few selected of those professors who reviews your paper based on various factors like how much original is your research work, how much is your research contributing to the society, the technicality of the paper, paper’s novelty etc and then the conference has its own acceptance rate (mine had one of the lowest acceptance rate in the field, making it even more difficult to get into). Once your paper stands upto the mark in all of the above criteria, your paper gets selected to be published.
How was the response of the jury?
- All the 3 days of the conference are divided into various sessions. Each session is headed by session chair(s) who are eminent professors. My paper presentation, and the depth of the research impressed everyone and even my session chair approached me upfront outside the presentation auditorium. She appreciated me a lot and even offered words of praises to me in writing (getting a letter appreciating your research work is a big thing, because it can open n number of doors for pursuing further research).
Contact details for the people to reach out to you.
- Email : [email protected]
- Facebook : https://www.facebook.com/shubhanshu.gupta.982
- Google+ : https://plus.google.com/u/0/+ShubhanshuGupta93/posts
- Twitter : https://twitter.com/Shubhanshugupta
We are very sure that Shubhanshu’s research work would have inspired or motivated many of you. We would be constantly unearthing the hidden talents all across the country. To read more of our interviews, visit our research blog series. You can also comment down below, anything that you have to say about the article above. If you want your research work to be covered by The College Store, contact us here. And of course, stay tuned for more
To know more about us, visit About Us.
Team College Store