Makes the changes a reviewer decides on: Add or drop item links,
prerequisite edges or related edges. Use it after mv_kg_fit() or
mv_kg_compare() for the changes that mv_kg_refine() does not make
on its own, such as dropping a prerequisite edge the data do not
support.
Arguments
- kg
a knowledge graph from
mv_kg_build()oras_mv_kg().- add, drop
optional edges: Data frames with columns
fromandtoand an optionaltype. An edge from an item is an item link ("measures"); other edges are"prerequisite"unlesstypesays"related". Edges are dropped before they are added.- name
optional new name; by default the name is kept.
Value
An mv_kg object, with the item metadata kept. An item whose
last link is dropped leaves the graph, with a warning.
Examples
# the two changes the data suggest for the published fractions graph
kg <- mv_kg_edit(fractions_kg,
add = data.frame(from = c("equivalent", "F14"), to = c("compare", "common_denominator")),
drop = data.frame(from = "F14", to = "compare"))
mv_kg_compare(kg, lab_truth("fractions")$graph)
#> <mv_kg_compare> Grade 5 fractions vs Grade 5 fractions (true graph)
#> Item links: 27 of the 27 reference links are in the graph (recall
#> 1.00), and 27 of the graph's 27 links are in the reference (precision
#> 1.00).
#> 24 of 24 items are linked to exactly the same concepts.
#> Prerequisites: 6 of 6 reference edges found (recall 1.00).