Text Analysis approach to Predict Recommendation system




Shaik, Roshan

Journal Title

Journal ISSN

Volume Title



We plan a novel technique for consequently creating a playlist of suggested tunes in the famous social music sharing application Spotify that are preferred with high likelihood by a client. Our strategy utilizes various seed specialists as an info that are acquired through any semblance of craftsmen and the listening history of melodies of a Spotify client. In the first place, we develop an information vector involving every one of the craftsmen that the client likes and tunes in to in Spotify. At that point, we look for different craftsmen and groups identified with them . We appoint a score to each craftsman in the along these lines got assortment, in light of the recurrence of his/her appearance. At last, we develop a playlist including arbitrarily chosen well known tunes related with the most as often as possible refered to craftsmen. We analyze the suggestion execution of our calculation by registering its WTF score (part of loathed melodies) and curiosity factor (part of new preferred tunes) on playlists produced for various seed input sizes.



Machine learning, Natural language processing, Spotify