ActiveCrowdToolkit  0.1
CrowdsourcingModels Namespace Reference

Classes

class  ActiveCrowdToolkitExperiment
 
class  ActiveLearning
 Class of active learning functions More...
 
class  ActiveLearningResult
 
class  BCC
 The BCC model class. More...
 
class  BCCPosteriors
 The BCC posteriors class. More...
 
class  CBCC
 The CBCC model class. More...
 
class  CBCCPosteriors
 CBCC posterior object. More...
 
class  DataMapping
 Data mapping class. This class manages the mapping between the data (which is in the form of task, worker ids, and labels) and the model data (which is in term of indices). More...
 
class  Datum
 This class represents a single datum, and has methods to read in data. More...
 
class  Program
 The class for the main program. More...
 
class  Results
 

Enumerations

enum  RunType {
  RunType.VoteDistribution = 0, RunType.MajorityVote = 1, RunType.DawidSkene = 2, RunType.BCC = 3,
  RunType.CBCC = 4
}
 Options for which model to run. More...
 
enum  TaskSelectionMethod { RandomTask, EntropyTask, UniformTask }
 Metrics for selecting tasks. More...
 
enum  WorkerSelectionMethod { RandomWorker, BestWorker }
 Metrics for selecting workers More...
 
enum  RunMode {
  RunMode.ClearResults, RunMode.BatchTraining, RunMode.OnlineTraining, RunMode.LookAheadExperiment,
  RunMode.LoadAndUseCommunityPriors, RunMode.Prediction
}
 The different modes in which the model can be run. More...
 

Enumeration Type Documentation

The different modes in which the model can be run.

Enumerator
ClearResults 

Clears all posteriors

BatchTraining 

Training from a batch of data - uses initial priors.

OnlineTraining 

Online training from a batch of data - uses previous posteriors as priors.

LookAheadExperiment 

Online training where we don't update the posteriors

LoadAndUseCommunityPriors 

Use communities as workers in a BCC

Prediction 

Prediction of worker labels

Options for which model to run.

Enumerator
VoteDistribution 

The true label distribution as given by the normalised workers' label counts.

MajorityVote 

The true label is the majority label.

DawidSkene 

The Dawid-Skene model.

BCC 

The BCC model.

CBCC 

The CBCC model.

Metrics for selecting tasks.

Metrics for selecting workers