Questions about the meta information

[x_i_n_g]

Each example in the validation sets comes with some meta information, i.e., image_quality, domain, query_category, and dynamism.

1. Are they also available in the test/blind sets?
2. Are they used to scope the resources in search APIs? For example, simple-recognition examples will not have relevant web pages/images; fast-changing examples will always have relevant information that can be retrieved?

[yilun_jin8]

1. They will be available in the private test sets.
2. I don’t think there are such assumptions.