/*
* Definition: parse the xml files in the Knowtator annotation output folder to build
* CRF training file.
*/
public void buildCRFi2b2MedicationTrain(
String crfTrainingFile,
boolean bMakeLowerCase,
boolean bLabelProblemList,
boolean bAddReasonSingleton)
throws Exception {
String baseDirectory = KnowtatorXMLDirectory;
RawInput rinput = new RawInput();
ArrayList<String> kxmlfileList = new ArrayList<String>();
rinput.getDirectoryFile(baseDirectory, kxmlfileList);
BufferedWriter fTrain;
try {
fTrain = new BufferedWriter(new FileWriter(crfTrainingFile));
for (String knowtatorXmlFile : kxmlfileList) {
System.out.println(knowtatorXmlFile);
Parser p = new Parser(knowtatorXmlFile);
String taggedArticle =
p.getTaggedArticleForCRFModel(bMakeLowerCase, bLabelProblemList, bAddReasonSingleton);
taggedArticle = taggedArticle.replaceAll("\\|\\|O", "");
// System.out.print(taggedArticle);
fTrain.write(taggedArticle + "\n");
}
fTrain.close();
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
System.out.println(kxmlfileList.size());
}
private void InitialBuilder() {
// root_id = 990000;
tempFilepath = RawInput.getTemporaryFilePath("KnowtatorXmlBuilder", "xml");
try {
fos = new FileOutputStream(tempFilepath);
} catch (FileNotFoundException e) {
e.printStackTrace();
}
OutputFormat of = new OutputFormat("XML", "UTF-8", true);
of.setIndent(1);
of.setIndenting(true);
serializer = new XMLSerializer(fos, of);
try {
hd = serializer.asContentHandler();
} catch (IOException e) {
e.printStackTrace();
}
try {
hd.startDocument();
} catch (SAXException e) {
e.printStackTrace();
}
}
public static void GenerateListNarrativeSentenceTrainingData(
String listFile, String narrativeFile, String negativeFile) throws Exception {
// TODO Auto-generated method stub
String baseDirectory = KnowtatorXMLDirectory;
RawInput rinput = new RawInput();
ArrayList<String> kxmlfileList = new ArrayList<String>();
rinput.getDirectoryFile(baseDirectory, kxmlfileList);
BufferedWriter fList = null;
BufferedWriter fNarrative = null;
BufferedWriter fNegative = null;
try {
fList = new BufferedWriter(new FileWriter(listFile));
fNarrative = new BufferedWriter(new FileWriter(narrativeFile));
fNegative = new BufferedWriter(new FileWriter(negativeFile));
for (String knowtatorXmlFile : kxmlfileList) {
System.out.println(knowtatorXmlFile);
Parser p = new Parser(knowtatorXmlFile);
ArrayList<String> narratives = p.GetListNarrativeSentences("narrative");
ArrayList<String> lists = p.GetListNarrativeSentences("list");
ArrayList<String> negatives = p.GetListNarrativeSentences("negative");
for (String sent : narratives) {
fNarrative.write(knowtatorXmlFile + "\t" + sent + "\n");
}
for (String sent : lists) {
fList.write(knowtatorXmlFile + "\t" + sent + "\n");
}
for (String sent : negatives) {
fNegative.write(knowtatorXmlFile + "\t" + sent + "\n");
}
}
fList.close();
fNarrative.close();
fNegative.close();
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
System.out.println(kxmlfileList.size());
}
/**
* @param trainFile
* @param makeLowerCase
* @param labelProblemList
* @param addReasonSingleton
* @throws Exception
*/
private void buildCRFTrainBySentence(
String trainFile, boolean makeLowerCase, boolean labelProblemList, boolean addReasonSingleton)
throws Exception {
String baseDirectory = KnowtatorXMLDirectory;
RawInput rinput = new RawInput();
ArrayList<String> kxmlfileList = new ArrayList<String>();
rinput.getDirectoryFile(baseDirectory, kxmlfileList);
BufferedWriter fTrain;
try {
fTrain = new BufferedWriter(new FileWriter(trainFile));
for (String knowtatorXmlFile : kxmlfileList) {
System.out.println(knowtatorXmlFile);
// if(!knowtatorXmlFile.contains("11995"))
// continue;
Parser p = new Parser(knowtatorXmlFile);
// String taggedArticle = p.getTaggedArticleForCRFModel(makeLowerCase,
// labelProblemList, addReasonSingleton);
ArrayList<String> taggedSentences = p.getTaggedSentences();
for (String sentence : taggedSentences) {
sentence = sentence.replaceAll("\\|\\|O", "");
// System.out.print(taggedArticle);
fTrain.write(sentence + "\n\n");
}
}
fTrain.close();
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
System.out.println(kxmlfileList.size());
}
/*
* Definition: used to build ARFF training file for medication relationship
* model.
*/
static void buildMedicationRelationshipTrainingData(
String relationshipTrainfile, boolean addReasonSingleton) throws Exception {
String baseDirectory = KnowtatorXMLDirectory;
RawInput rinput = new RawInput();
ArrayList<String> kxmlfileList = new ArrayList<String>();
rinput.getDirectoryFile(baseDirectory, kxmlfileList);
String head = "";
String dataSet = "";
BufferedWriter fTrain;
JMerki jm = new JMerki();
jm.initializeParser();
for (String knowtatorXmlFile : kxmlfileList) {
Parser p = new Parser(knowtatorXmlFile);
String rawContent = p.getRawContent();
ArrayList<String> listedReason = new ArrayList<String>();
listedReason = jm.getListedReason(rawContent);
HashMap<String, String> data = p.getARFFDataset(listedReason);
if (head.isEmpty()) head = data.get("head");
System.out.println(data.get("data"));
dataSet += data.get("data");
}
try {
fTrain = new BufferedWriter(new FileWriter(relationshipTrainfile));
fTrain.write(head + "@data\n" + dataSet);
fTrain.close();
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
System.out.println(kxmlfileList.size());
}
private void EndBuilder() {
try {
hd.endElement("", "", "annotations");
hd.endDocument();
} catch (SAXException e) {
e.printStackTrace();
}
try {
fos.close();
} catch (IOException e) {
e.printStackTrace();
}
System.out.println(tempFilepath);
XMLContent = RawInput.getFullText(tempFilepath);
}
private static void GenerateLNTrainingFileForMalletDocumentClassification() throws Exception {
// TODO Auto-generated method stub
String baseDirectory = KnowtatorXMLDirectory;
RawInput rinput = new RawInput();
ArrayList<String> kxmlfileList = new ArrayList<String>();
RawInput.getDirectoryFile(baseDirectory, kxmlfileList);
try {
for (String knowtatorXmlFile : kxmlfileList) {
File fxml = new File(knowtatorXmlFile);
String fileName = fxml.getName();
String listFile =
Messages.getString("i2b2.mallet.document.train.file.folder.list") + fileName + ".txt";
String narrativeFile =
Messages.getString("i2b2.mallet.document.train.file.folder.narrative")
+ fileName
+ ".txt";
File fl = new File(listFile);
File fn = new File(narrativeFile);
if (fl.exists() || fn.exists())
System.err.println(
"error in generate traing data for mallet document classification: 0618 pm File exist!");
BufferedWriter fList = null;
BufferedWriter fNarrative = null;
fList = new BufferedWriter(new FileWriter(listFile));
fNarrative = new BufferedWriter(new FileWriter(narrativeFile));
Parser p = new Parser(knowtatorXmlFile);
ArrayList<String> narratives = p.GetListNarrativeSentences("narrative");
ArrayList<String> lists = p.GetListNarrativeSentences("list");
for (String sent : narratives) {
// fNarrative.write(knowtatorXmlFile + "\t" + sent + "\n");
fNarrative.write(sent + "\n");
}
for (String sent : lists) {
// fList.write(knowtatorXmlFile + "\t" + sent + "\n");
fList.write(sent + "\n");
}
fList.close();
fNarrative.close();
}
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
System.out.println(kxmlfileList.size());
// generate mallet file
File root = new File(Messages.getString("i2b2.mallet.document.train.file.folder"));
String arffFile = Messages.getString("i2b2.mallet.ln.classification.arff.file");
MalletWekaInterface mwInterface = new MalletWekaInterface();
mwInterface.GenerateTrainingARFFfile(root, arffFile);
}