private void constructBalancedSentiCorpus() throws IOException {
balanceSentiment(
prop.getProperty("senti.corpus.file.search"),
prop.getProperty("senti.corpus.search.fields.delim"),
OMTwitterCorpusFile.fieldNameToId(prop.getProperty("senti.corpus.search.fields"), "\\s+"),
prop.getProperty("senti.corpus.file"),
prop.getProperty("senti.corpus.fields.delim"),
OMTwitterCorpusFile.fieldNameToId(prop.getProperty("senti.corpus.fields"), "\\s+"),
false);
balanceSentiment(
prop.getProperty("senti.corpus.file.sample"),
prop.getProperty("senti.corpus.sample.fields.delim"),
OMTwitterCorpusFile.fieldNameToId(prop.getProperty("senti.corpus.sample.fields"), "\\s+"),
prop.getProperty("senti.corpus.file"),
prop.getProperty("senti.corpus.fields.delim"),
OMTwitterCorpusFile.fieldNameToId(prop.getProperty("senti.corpus.fields"), "\\s+"),
true);
printCorpusStat(
prop.getProperty("senti.corpus.file"),
prop.getProperty("senti.corpus.fields.delim"),
OMTwitterCorpusFile.fieldNameToId(prop.getProperty("senti.corpus.fields"), "\\s+"));
}
@Override
public void initialize() throws ResourceInitializationException {
super.initialize();
logger = getLogger();
try {
evalCorpusReader =
new OMTwitterCorpusFileReader(
(String) getConfigParameterValue(PARAM_EVALUATION_CORPUS_FILE),
(String) getConfigParameterValue(PARAM_EVALUATION_CORPUS_DELIM),
OMTwitterCorpusFile.fieldNameToId(
(String) getConfigParameterValue(PARAM_EVALUATION_CORPUS_FIELDS), " "));
} catch (Exception e) {
logger.log(Level.SEVERE, e.getMessage());
throw new ResourceInitializationException(e);
}
printResult = (Boolean) getConfigParameterValue(PARAM_PRINT_RESULT);
String neTagsStr = (String) getConfigParameterValue(PARAM_NAMED_ENTITY_TAGS);
if (neTagsStr == null) {
throw new ResourceInitializationException();
}
labelNone = (String) getConfigParameterValue(PARAM_LABEL_NONE);
String[] neTags = neTagsStr.split(" ");
int idx = 0;
map = new HashMap<String, Integer>();
for (String tag : neTags) {
map.put(tag + "_B", idx++);
map.put(tag + "_M", idx++);
map.put(tag + "_E", idx++);
}
map.put(labelNone, idx++);
labelNoneIdx = idx - 1;
stat = new int[idx][3];
senti = new int[3][3];
classifiedEntityCnt = new int[idx / 3];
answerEntityCnt = new int[idx / 3];
}