I started developing a commercial web application that uses spaCy – a MIT-licensed library for natural language processing. In order to use this library, one needs to initialize a processing pipeline by pointing to one of the language models – data files distributed under the CC BY-SA-3.0 license. The model, in an unaltered form, gets loaded and then the application interacts with the library through its API, not directly with the model.
I’ve read the license (CC BY-SA-3.0) but it’s hard for me to determine whether my application should be considered a derivative work or not. On the one hand the application only indirectly uses the model (doesn’t improve/modify it in any way), but on the other hand, such a model is indispensable for the application to work.
But what’s crucial for me is this. Let’s assume that my application constitutes a derivative work. Does this mean that I have to release the application’s source code under the CC BY-SA-3.0 license, even though the application runs only on my servers and users can only interact with it through a computer network? Does it work similarly to the AGPL?
I really appreciate any help you can provide.