task_charts__multiseries__ranked_change_extremum_label¶
Contract¶
- Domain:
charts - Scene id:
multiseries - Query ids:
largest_increase_label,largest_decrease_label,largest_absolute_gap_label,smallest_absolute_gap_label - Answer schema:
string_label - Annotation schema:
point_map - Program contract:
arg_extreme(categories, change(value(series_a), value(series_b), change_measure), direction)
Program Contract¶
Program: select_label(arg_extreme(categories, change(value(category, series_a), value(category, series_b), measure), direction)); output=string_label; annotation=point_map(mark_center(answer_category, {series_a,series_b})); scene=multiseries; scope=ranked_change_extremum_label
Candidate set: the visible series marks across shared x/category labels inside the ranked_change_extremum_label objective scope.
Operands: prompt-bound labels, categories, series names, thresholds, intervals, references, and encoded chart values, plus the active query id's comparator, direction, target role, or extremum focus when present.
Operation: evaluate select_label over the candidate set using the filters, comparisons, aggregations, rankings, projections, or counterfactual edits named in the program expression; generation enforces a unique final answer.
Output binding: answer is the unspecified value bound by string_label.
Annotation witnesses: unspecified witnesses bound by point_map(mark_center(answer_category, {series_a,series_b})). The Annotation Contract below defines the prompt-facing witnesses.
Query ids: largest_increase_label, largest_decrease_label, largest_absolute_gap_label, smallest_absolute_gap_label.
Reasoning Operations¶
Families: ranking, formula_evaluation
Implementation¶
- Source:
src/trace_tasks/tasks/charts/multiseries/ranked_change_extremum_label.py - Class:
ChartsMultiseriesRankedChangeExtremumTask - Prompt bundle:
src/trace_tasks/resources/prompts/charts/multiseries/charts_multiseries_v1.json
Annotation¶
Annotate the two queried series marks in the answer category using keys of the form <category>:<series>.