Package edu.odu.cs.cs350

Class ACMTrainingSetBuilder

java.lang.Object

edu.odu.cs.cs350.ACMTrainingSetBuilder

public class ACMTrainingSetBuilder
extends Object

Build a training set based on TrainingData PDFs and train a FilteredClassifier.

Constructor Summary

Constructors

Constructor	Description
ACMTrainingSetBuilder()

Method Summary

Modifier and Type	Method	Description
static void	addStringInstancesFromRepository(weka.core.Instances raw, List<String> possibleCategories, List<String> docPaths)	Gets PDFs from Maven repo, extracts text, and adds Instances.
static weka.filters.unsupervised.attribute.StringToWordVector	buildTrainingFilter(weka.core.Instances raw)	Build a StringToWordVector filter for training and classification.
static String	determineCategoryFromPath(String resourceRelative, List<String> possibleCategories)	Determine ACM category from document path within repo.
static List<String>	getACMClasses(List<String> docPaths)	Get list of ACM categories from document paths (using the subfolder names).
static List<String>	getDocumentListing()	Get list of document paths from TrainingData repository.
static String	getTextFromDocument(String resourceRelative)	Extract text from a PDF document in the TrainingData repository.
static void	main(String[] inputArguments)	Main method to build training set and train classifier.
static void	writeModels(weka.core.Instances raw, weka.core.Instances trainingSet, weka.classifiers.meta.FilteredClassifier trainedClassifier)	Write trained models to disk.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

ACMTrainingSetBuilder

public ACMTrainingSetBuilder()

Method Details

main

public static void main(String[] inputArguments)
                 throws Exception

Main method to build training set and train classifier.

Parameters:

inputArguments - program arguments (unused)

Throws:

Exception - if files cannot be read or written

writeModels

public static void writeModels(weka.core.Instances raw,
 weka.core.Instances trainingSet,
 weka.classifiers.meta.FilteredClassifier trainedClassifier)
                        throws Exception

Write trained models to disk.

Parameters:

raw - raw Instances header

trainingSet - vectorized training Instances

trainedClassifier - trained FilteredClassifier

Throws:

Exception - if files cannot be written

addStringInstancesFromRepository

public static void addStringInstancesFromRepository(weka.core.Instances raw,
 List<String> possibleCategories,
 List<String> docPaths)
                                             throws IOException

Gets PDFs from Maven repo, extracts text, and adds Instances.

Parameters:

raw - empty Instances object to fill

possibleCategories - list of valid ACM categories

docPaths - list of document paths within the Maven repo

Throws:

IOException - if a file can't be read

determineCategoryFromPath

public static String determineCategoryFromPath(String resourceRelative,
 List<String> possibleCategories)

Determine ACM category from document path within repo.

Parameters:

resourceRelative - path to the document relative to the TrainingData resource path

possibleCategories - list of valid ACM categories

Returns:

determined category or null if not recognized

getTextFromDocument

public static String getTextFromDocument(String resourceRelative)
                                  throws IOException

Extract text from a PDF document in the TrainingData repository.

Parameters:

resourceRelative - path to the document relative to the TrainingData resource path

Returns:

extracted text, or null if extraction failed

Throws:

IOException - if the document cannot be read

buildTrainingFilter

public static weka.filters.unsupervised.attribute.StringToWordVector buildTrainingFilter(weka.core.Instances raw)
                                                                                  throws Exception

Build a StringToWordVector filter for training and classification.

Parameters:

raw - raw Instances to set input format

Returns:

configured StringToWordVector filter

Throws:

Exception - if filter cannot be initialized

getDocumentListing

public static List<String> getDocumentListing()
                                       throws IOException

Get list of document paths from TrainingData repository.

Returns:

list of document paths including category

Throws:

IOException - if the listing cannot be read

getACMClasses

public static List<String> getACMClasses(List<String> docPaths)

Get list of ACM categories from document paths (using the subfolder names).

Parameters:

docPaths - list of document paths from repository

Returns:

list of unique ACM categories