Do unavailable judgements imply irrelevance?

[vishal-git]

I have a question about the following sentence in README:

For each query, the dataset provides a list of up to 40 potentially relevant results, together with ESCI relevance judgements (Exact, Substitute, Complement, Irrelevant) indicating the relevance of the product to the query.

For <query, item> pairs where a judgement is not available, would it make sense to infer that that particular item is not relevant to that query? Or should we assume that the ground truth for that <query, item> pair is unknown.