Is creating a personalized media recommender against the TOS

[Sam-sultana]

Country

United Kingdom

My Question or Issue

I am creating a media recommender, specifically a white box algorithm, I was reading the TOS specifically section IV 2.a.i. which states:
"using the Spotify Platform or any Spotify Content to train a machine learning or AI model or otherwise ingesting Spotify Content into a machine learning or AI model;"

is not allowed. I will be training my model on this dataset: 

https://www.aicrowd.com/challenges/spotify-million-playlist-dataset-challenge

I was planning on importing recent listening data for a user to find recommendations, I the model wont be using this data to learn and it will not be stored merely used to get recommendations based on listen history. 

To me this seemed like a grey area I am not using the data pulled to train a model and the data isnt really being ingested just put in to prompt a list of recommendation output. But before I finalize my plan and start development I wanted to double check this would be ok.

Thanks in advance 🙂

[Ximzend]

It's at the page you have linked:

"The dataset and challenge are available strictly for research and non-commercial use. You may not redistribute or make available any part or whole of this dataset."

[Sam-sultana]

Yes, the dataset is OK but in addition I am planning for a user to be able to use the Spotify API to get there recent listening history and use this to generate recommendations instead of using playlists, Because I'm using the actual spotify API the rules in the dataset would not apply. 

I guess the gist of the question is although I am not using the data gathered to train my model, does it count as "ingesting" a users data into the model if it takes the listen history, runs through the algorithm, and then outputs recommended songs with no memory of original input afterwards

[Ximzend]

Using the API and the dataset to train your AI is not allowed.