org.hd.d.pg2k.ai.scorer
Class ScoreAndConf

java.lang.Object
  org.hd.d.pg2k.ai.scorer.ScoreAndConf

All Implemented Interfaces:

MemoryTools.Internable

public final class ScoreAndConf
extends java.lang.Object
implements MemoryTools.Internable

Immutable (and Internable) return type from scorer method. Allowed values run from -MAX (bad) through 0 (neutral) to +MAX (good) for score, and from 0 (no confidence) to +MAX (full) for confidence.

Note that the score values are symmetrical about zero.

The low-ish precision of the elements is deliberate to reflect the fact that scoring is bound to be fairly imprecise and to save a little space.

MAX

public static final short MAX

Maximum positive or negative value of any element; strictly positive.

See Also:

Constant Field Values

score

public final short score

Score: -MAX (bad) through 0 (neutral) to +MAX (good).

confidence

public final short confidence

Confidence: 0 (no confidence) to +MAX (full) for confidence.

NO_OPINION

public static final ScoreAndConf NO_OPINION

All-zeros value used to mean "no idea" / "no opinion" when returned by a Scorer.

MAX_GOODNESS

public static final int MAX_GOODNESS

Max (+ve) "goodness" value returned by computeScorerGoodness(); strictly positive.

See Also:

Constant Field Values

ByGoodness

public static final java.util.Comparator<ScoreAndConf> ByGoodness

A static Comparator that sorts by goodness.

ByConfidence

public static final java.util.Comparator<ScoreAndConf> ByConfidence

A static Comparator that sorts by confidence.

ScoreAndConf

public ScoreAndConf(short score,
                    short confidence)

Construct an instance.

ScoreAndConf

public ScoreAndConf(int score,
                    int confidence)

Construct an instance using int values; requires extra checking.

hashCode

public int hashCode()

The hash depends on all elements.

Overrides:

hashCode in class java.lang.Object

equals

public boolean equals(java.lang.Object obj)

Equality depends on all elements.

Overrides:

equals in class java.lang.Object

toString

public java.lang.String toString()

Compute human-readable summary; never null. This is also designed to be parsable to (re)construct an instance as an alternative to serialisation.

This includes the scale (maximum) for non-zero values to allow for robust reconstruction in future versions.

Overrides:

toString in class java.lang.Object

fromString

public static ScoreAndConf fromString(java.lang.String s)

Parse toString() representation to return ScoreAndConf value; never null.

Throws:

java.lang.IllegalArgumentException - if the input is unparsable

isGood

public java.lang.Boolean isGood()

Find out if this is notably good or bad using the same thresholds as for voting (etc) in the rest of the Gallery. Returns TRUE if rated good, FALSE if bad, null if not significantly either (ie too close to neutral) or if unknown.

This is based on the product of score and confidence.

computeScorerGoodness

public static int computeScorerGoodness(ScoreAndConf sac)

Compute Scorer "goodness" from its ScoreAndConfidence value. We want Scorers with a good correlation (+ve) and good confidence.

This is a unitless relative value used to sort Scorers in the range [-MAX^2,+MAX^2] ie [-MAX_GOODNESS,+MAX_GOODNESS where -ve is "wrongheaded", 0 is "useless", and +ve is "good".