OOP implementation of metrics.py about decepticonlp

[someshsingh22]

I think .distance() would be more specific and elaborate
Jaccard Distance, Edit distance, Euclidean distance are the metrics till now.
You don't want the user to confuse between the terms, calculate could refer to a lot more functions

[rohts-patil]

I think the calculate() method will be appropriate.

For example, class name is EuclideanDistanceMetric and use it as EuclideanDistanceMetric.calculate(vector from, vector to)
@someshsingh22

Also, is there any possibility that there would be metrics added in the future that are not distance-based?

[someshsingh22]

This is basically how different two strings are so it will remain distance for most cases, however, when we will transition to vectors terms like cosine similarity will be used. .calculate() alone looks ambiguous however EucledianDistance.calculate(x,y) looks exactly what we'd want. What's your call @rajaswa

[rajaswa]

Lets stick to .calclate() for the time being

[parantak]

I think as we implement more metrics, we could perhaps split them into subcategories like edit-based(Levenshtein, etc.), token-based (Jaccard, etc.), vector-based, etc.

[parantak]

OOP implementation. If you want a different structure, let me know.