Package de.fu_berlin.ties.eval

This packages provides functionality for evaluating results of classifiers and extractors.

See:
          Description

Interface Summary
AccuracyView Provides a read-only view on Accuracy statistics and the underlying raw counts.
EvalInput Classes implementing this interface provide input for calculating evaluation metrics: true positives, false negatives and false positives.
FeatureCountView Provides a read-only view on the statistics calculated by the FeatureCount class and the underlying raw counts.
FMetricsSummary Implementations of this interface can show statistical summaries of precision, recall, and F1 metrics updated in several operations.
FMetricsView A read-only view of the evaluation results calculated by the FMetrics class and the underlying raw counts.
MultiFMetricsView A read-only view of multiple FMetrics and the sums and averages calculated over them.
 

Class Summary
Accuracy Counts true and false items and measures the accuracy: A = true / (true + false).
AverageLength A simple goal that reads a list of EvaluatedExtractionContainers and calculates the average length (in characters and tokens) for extractions of of all types (e.g. speaker, location etc.) and all evaluation statuses (e.g. correct, missing etc.)
EvalStatus Type-safe enumeration of possible evaluation states for predictions (unknown, correct, spurious etc.) and answer keys (missing etc.).
FeatureCount Keeps track of the average number of features and of unique features in context representations and of the average number of contexts in documents.
FMetrics This class manages and updates evaluation results, calculating precision (P), recall (R) and F-measure (F).
LineShuffleGenerator Randomly reshuffles the lines in a file (except for the first n lines, if configured).
Mistake Each instance of this class describe a mistake.
MistakeAnalyzer Reads an EvaluatedExtractionContainer (in DSV format) and analyses the types of prediction errors that occurred.
MistakeMatrix Stores the results of a mistake analysis performed by MistakeAnalyzer.
MultiFMetrics Instances of this class manage multiple FMetrics for different types.
PredictionEvaluator Reads a set of files that must contain predictions and evaluates them against the corresponding answer keys (*.ans files).
ReEvaluator A processor that can be used to re-evaluate the contents of an EvaluatedExtractionContainer.
ShuffleGenerator Arranges all input arguments (for example, files or URLs) in random "shuffles", so they can subsequently processed in random (but fixed) order.
SummaryFMetrics FMetrics extension that additionally calculates a StatisticalSummary of the intermediate precision, recall, and F1 metrics resulting from different update operations.
ValueSummary Creates StatisticalSummary for any number of items ("keys") that occur zero or more times in any number of runs ("identifiers").
 

Enum Summary
Mistake.MistakeTypes The types of mistakes that can occur.
 

Package de.fu_berlin.ties.eval Description

This packages provides functionality for evaluating results of classifiers and extractors.



Copyright © 2003-2007 Christian Siefkes. All Rights Reserved.