NewBestSub/it.uniud.newbestsub.dataset.model/PrecomputedData

PrecomputedData

data class PrecomputedData(val averagePrecisionBySystem: Array<DoubleArray>, val topicColumnViewByTopic: Array<DoubleArray>, val fullSetMeanAPBySystem: DoubleArray, val numberOfSystems: Int, val numberOfTopics: Int, val systemIdsInOrder: List<String>, val systemIndexById: Map<String, Int>)

Immutable, shared, hot-path numeric bundle.

Built once at dataset load time and injected wherever evaluations happen (Best, Worst, and Average problems, as well as seeders).

Layout

Row-major matrix of AP scores per system and topic.
Column views to support delta evaluations.
Precomputed full-set means per system.

Indices & IDs

Systems are referenced by 0..numberOfSystems-1 in the order of systemIdsInOrder.
systemIndexById maps a system identifier to its compact row index.

Safety

All arrays are owned by this object.
Do not mutate them after construction.

Constructors

PrecomputedData

constructor(averagePrecisionBySystem: Array<DoubleArray>, topicColumnViewByTopic: Array<DoubleArray>, fullSetMeanAPBySystem: DoubleArray, numberOfSystems: Int, numberOfTopics: Int, systemIdsInOrder: List<String>, systemIndexById: Map<String, Int>)

Properties

averagePrecisionBySystem

val averagePrecisionBySystem: Array<DoubleArray>

Row-major AP matrix: averagePrecisionBySystem[s][t] gives the AP of system s on topic t.

fullSetMeanAPBySystem

val fullSetMeanAPBySystem: DoubleArray

Mean AP of each system over all topics.

numberOfSystems

val numberOfSystems: Int

Total number of systems in the dataset.

val numberOfTopics: Int

Total number of topics in the dataset.

systemIdsInOrder

val systemIdsInOrder: List<String>

Deterministic list of system IDs, defining row order.

systemIndexById

val systemIndexById: Map<String, Int>

Map from system ID to its row index.

topicColumnViewByTopic

val topicColumnViewByTopic: Array<DoubleArray>

Column-major AP matrix: topicColumnViewByTopic[t][s] gives the AP of system s on topic t. Used for fast ± updates.