Missing Details related to Signatures #6
Description
Hi, thanks a lot for your valuable contribution.
I have a question regarding the definition of "signature" in your paper. You mentioned that “We therefore define the signature of PTM usage in an application as the combination of (1) library import and (2) calls into that library to load a PTM.” However, many of the signatures listed seem to be just model names (e.g., "en_textcat_emotion_umberto" or "grc_ud_perseus_lg") paired with generic function calls like "load", rather than the full combination of import + function call. Do you also use these names to identify the mapping between repo and model?
Could you clarify what you meant by this? For example, "en_textcat_emotion_umberto" looks like a specific pre-trained model name used via spaCy. How exactly is this considered a “signature,” and how should it be used when searching in tools like Sourcegraph? and How you manually define or identify these signatures? What sources did you used?
If possible, could you give a concrete example of how to construct a Sourcegraph query using this kind of signature?