Blend image suggestions based on section alignment with those based on visual topics to form an initial proof-of-concept dataset for manual evaluation.
- ask Research to share their datasets - on HDFS at /user/mnz/imagerecs/recs-2022-06-07
- join it with Structured Data's one, see section 4
- filter out obvious sections like references, see also T311730: [L] Exclude certain sections from having generated image suggestions
- compute suggestion counts per language