Linking backend implemented using disjointed sets

[loolmeh]

sketch:

• schema:
  • components table:
    • sentence_id: the exact sent, fk
  • component_id: representative sentence_id of the disjointed set, chosen by min(sentence_id) in the disjointed set
• operations:
  • link:
  • find the two disjointed sets using sent's component_id
  • concat both sets into one set
  • find minimum id in set, update both disjointed sets to have this id
  • unlink:
  • find the disjointed set
  • remove this id from the set
  • find the minimum id in set, update disjointed set, removed sentence becomes its own set, with component_id same as its own id
  • find:
  • simple select on component_id column
  • get_subgraph:
    • find which set a sent_id is in and get it
  • feed the nodes to the graphing code and calculate all distances