Add new metric DIEM (Dimension Insensitive Euclidean Metric)

[jakubxy08]

Checked

• I searched existing ideas and did not find a similar one
• I added a very descriptive title
• I've clearly described the feature request and motivation for it

Feature request

I propose the addition of the new DIEM metric described here: https://arxiv.org/pdf/2407.08623
Functions such as cosine similarity e.g. to get relevant data from text database.

Motivation

Cosine similarity metric for extracting relevant texts from documents is sometimes insufficient (size and dimensionality of data and interpretability).

Proposal (If applicable)

If someone find this can be useful and give green light I can implement that in this repository.

[BodduSriPavan-111]

Hi @jakubxy08, we have recently published a Python library "diemsim", which implements "Diemsion Insensitive Euclidean Metric" by utilizing NumPy and vectorization concepts. Hope, it will help in this regard.

Feel free to explore the repository here: https://github.com/BodduSriPavan-111/diemsim