35 lines
968 B
Markdown
35 lines
968 B
Markdown
You are casting the deciding judgment on disagreements about
|
|
a consolidated vocabulary of thematic keywords from a song
|
|
corpus: for each pair of keyword blocks in question, earlier
|
|
independent judgments disagreed on whether the two blocks
|
|
express the same theme.
|
|
|
|
Input: a JSON object. "blocks" maps a block id to the
|
|
keywords of that block; "pairs" lists the block-id pairs in
|
|
question:
|
|
|
|
{
|
|
"blocks": {
|
|
"b1": ["keyword", "another-keyword"],
|
|
"b2": ["keyword"],
|
|
"b3": ["keyword"]
|
|
},
|
|
"pairs": [["b1", "b2"], ["b1", "b3"]]
|
|
}
|
|
|
|
Task: for each listed pair, decide whether the two blocks
|
|
express the same theme, by your own understanding of what the
|
|
keywords mean.
|
|
|
|
Rules:
|
|
|
|
- Judge every listed pair, each on its own merits from the
|
|
block contents alone.
|
|
- Judge only the listed pairs.
|
|
|
|
Output a single JSON array holding the pairs whose two blocks
|
|
express the same theme, written exactly as given in "pairs";
|
|
an empty array when none do:
|
|
|
|
[["b1", "b2"]]
|