Skip to main content

pub.layers.judgment

Linguistic judgment records for annotation experiments, crowdsourced judgments, and inter-annotator agreement. Inspired by bead's framework for constructing, deploying, and analyzing large-scale linguistic judgment experiments.

Types

experimentDef

Type: Record

Definition of an annotation or judgment experiment.

FieldTypeDescription
namestringExperiment name.
descriptionstringDetailed description.
taskTypeUriat-uriAT-URI of the task type definition node. Community-expandable via knowledge graph.
taskTypestringTask type slug (fallback). Known values: CATEGORICAL, ORDINAL, SCALAR, RANKING, SPAN_SELECTION, FREETEXT, PAIRWISE_COMPARISON, BEST_WORST_SCALING, ACCEPTABILITY, CUSTOM
guidelinesstringTask guidelines and instructions.
ontologyRefat-uriReference to the ontology used.
personaRefat-uriReference to the persona defining the annotation framework.
corpusRefat-uriReference to the corpus.
templateRefsarrayReferences to pub.layers.resource#template records used to generate stimuli. Array of at-uri
collectionRefsarrayReferences to pub.layers.resource#collection records providing filler pools. Array of at-uri
designrefExperiment design specification (list constraints, distribution, presentation). Ref: #experimentDesign
scaleMinintegerMinimum scale value for scalar/ordinal judgments.
scaleMaxintegerMaximum scale value.
labelsarrayAvailable labels for categorical judgments. Array of strings
knowledgeRefsarrayKnowledge graph references. Array of ref: pub.layers.defs#knowledgeRef
featuresrefRef: pub.layers.defs#featureMap
createdAtdatetimeRecord creation timestamp.

judgmentSet

Type: Record

A set of judgments from a single annotator for an experiment.

FieldTypeDescription
experimentRefat-uriReference to the experiment.
agentrefThe agent who produced this judgment set. Ref: pub.layers.defs#agentRef
judgmentsarrayThe judgments. Array of ref: #judgment
metadatarefRef: pub.layers.defs#annotationMetadata
knowledgeRefsarrayKnowledge graph references. Array of ref: pub.layers.defs#knowledgeRef
featuresrefOpen-ended features (e.g., annotator demographics, session metadata, completion time). Ref: pub.layers.defs#featureMap
createdAtdatetimeRecord creation timestamp.

judgment

Type: Object

A single judgment about a linguistic item.

FieldTypeDescription
itemrefReference to the item being judged. Ref: pub.layers.defs#objectRef
fillingRefat-uriReference to the pub.layers.resource#filling that generated the item being judged.
categoricalValuestringCategorical judgment label.
scalarValueintegerScalar/ordinal judgment value.
rankValueintegerRank position.
textSpanrefSelected text span for span-selection tasks. Ref: pub.layers.defs#span
freeTextstringFree-text response.
responseTimeMsintegerResponse time in milliseconds.
confidenceintegerConfidence score 0-10000.
behavioralDatarefBehavioral analytics (e.g., mouse movements, keystroke patterns, eye tracking). Ref: pub.layers.defs#featureMap
featuresrefRef: pub.layers.defs#featureMap

experimentDesign

Type: Object

Experiment design specification controlling how stimuli are constructed, distributed, and presented to annotators.

FieldTypeDescription
listConstraintsarrayConstraints on how items are distributed into lists. Array of ref: #listConstraint
distributionStrategyUriat-uriAT-URI of the distribution strategy definition node. Community-expandable via knowledge graph.
distributionStrategystringDistribution strategy slug (fallback). Known values: latin-square, random, blocked, stratified, custom
presentationModeUriat-uriAT-URI of the presentation mode definition node. Community-expandable via knowledge graph.
presentationModestringPresentation mode slug (fallback). Known values: random-order, fixed-order, blocked, adaptive, custom
timingMsintegerTime limit per item in milliseconds, if applicable.
featuresrefRef: pub.layers.defs#featureMap

listConstraint

Type: Object

A constraint on how experimental items are distributed into lists (e.g., Latin square balancing, no two items of the same condition adjacent).

FieldTypeDescription
kindUriat-uriAT-URI of the constraint kind definition node. Community-expandable via knowledge graph.
kindstringConstraint kind slug (fallback). Known values: latin-square, no-adjacent-same-condition, balanced-frequency, min-distance, custom
targetPropertystringThe property being constrained (e.g., 'condition', 'verb-type', 'length').
parametersrefRef: pub.layers.defs#featureMap
constraintrefFormal constraint expression. Ref: pub.layers.defs#constraint

agreementReport

Type: Record

An inter-annotator agreement report summarizing agreement metrics across judgment sets.

FieldTypeDescription
experimentRefat-uriReference to the experiment.
judgmentSetRefsarrayThe judgment sets compared. Array of at-uri
metricUriat-uriAT-URI of the metric definition node. Community-expandable via knowledge graph.
metricstringMetric slug (fallback). Known values: cohens-kappa, fleiss-kappa, krippendorff-alpha, percent-agreement, correlation, f1, custom
valueintegerMetric value scaled 0-10000.
numAnnotatorsintegerNumber of annotators.
numItemsintegerNumber of items judged.
featuresrefRef: pub.layers.defs#featureMap
createdAtdatetimeRecord creation timestamp.