Hi Sriram,

I believe not. Like the name of the approach says it looks for semantic similarity between sentences or pieces of text (at a sentence level).

For the variables you mentioned, I believe you would ideally need a direct matching. You could also look for edit distance to handle 'near similar’ cases.

Also, if you implement storage of these using the Trie Data Structure, you can, in a way, mimic similarity search (even if you don’t find what you are searching for) by returning everything under the lowest common matched entity.