Extracting mentions (i.e. sentences that contain the exact name of wikipedia concepts) takes a lot of CPU on the frontend. Depending on the amount of text, it can noticeably block the UI. E.g. on my modern MacBook Pro, loading the entire Andrew Ng course takes more than 20 seconds. That's already hard to accept - now imagine on a slow computer...
Solution
It's better to do these computations once in the backend, rather than multiple times on the frontend.
Tasks
[x] Implement the mentions extraction on the backend and include it wisely in the enrichment pipeline. Decide whether to extend the current enrichment table (I think this is best) or add a new table.
[x] Load the mentions on the frontend along with the other enrichments.
Problem
Extracting mentions (i.e. sentences that contain the exact name of wikipedia concepts) takes a lot of CPU on the frontend. Depending on the amount of text, it can noticeably block the UI. E.g. on my modern MacBook Pro, loading the entire Andrew Ng course takes more than 20 seconds. That's already hard to accept - now imagine on a slow computer...
Solution
It's better to do these computations once in the backend, rather than multiple times on the frontend.
Tasks