/**
* Loads a similarity metric if it isn't already loaded.
*
* @param name
* @return
* @throws ConfigurationException
* @throws IOException
*/
public SimilarityMetric loadMetric(String name) throws ConfigurationException, IOException {
if (env.hasMetric(name)) {
return env.getMetric(name);
}
info("loading metric " + name);
String type = requireString(configuration.getMetric(name), "type");
SimilarityMetric metric;
if (type.equals("category")) {
metric = createCategorySimilarity(name);
} else if (type.equals("text")) {
metric = createTextSimilarity(name);
} else if (type.equals("esa")) {
metric = createEsaSimilarity(name);
} else if (type.equals("links")) {
metric = createLinkSimilarity(name);
} else if (type.equals("pairwise")) {
metric = createPairwiseSimilarity(name);
} else if (type.equals("ensemble")) {
metric = loadEnsembleMetric(name);
} else {
throw new ConfigurationException("Unknown metric type: " + type);
}
metric.setName(name);
JSONObject params = configuration.getMetric(name);
if (params.containsKey("normalizer")) {
Normalizer norm = parseNormalizer(requireString(params, "normalizer"));
if (!norm.equals(null)) {
metric.setNormalizer(norm);
}
}
env.addMetric(name, metric);
return metric;
}
/**
* Loads an index if it is not already loaded.
*
* @param name - The name of the
* @return
* @throws IOException
* @throws ConfigurationException
*/
private IndexHelper loadIndex(String name) throws IOException, ConfigurationException {
if (env.hasIndex(name)) {
return env.getIndex(name);
}
info("loading index " + name);
JSONObject indexConfig = configuration.getIndex(name);
File parentDir = requireDirectory(configuration.getIndexes(), "outputDir");
IndexHelper helper = new IndexHelper(new File(parentDir, name), true);
if (indexConfig.containsKey("similarity")) {
String sim = requireString(indexConfig, "similarity");
if (sim.equals("ESA")) {
helper.getSearcher().setSimilarity(new ESASimilarity.LuceneSimilarity());
} else {
throw new ConfigurationException("unknown similarity type: " + sim);
}
}
if (indexConfig.containsKey("analyzer")) {
String analyzer = requireString(indexConfig, "analyzer");
if (analyzer.equals("ESA")) {
helper.setAnalyzer(new ESAAnalyzer());
} else {
throw new ConfigurationException("unknown analyzer type: " + analyzer);
}
}
env.addIndex(name, helper);
return helper;
}
private List<KnownSim> loadGold() throws ConfigurationException, IOException {
JSONObject params = configuration.getGold();
String path = requireString(params, "path");
List<KnownSim> g = KnownSim.read(new File(path));
env.setGold(g);
return g;
}
private SimilarityMetric createLinkSimilarity(String name)
throws ConfigurationException, IOException {
JSONObject params = configuration.getMetric(name);
SimilarityMetric metric;
String field = requireString(params, "field");
LinkSimilarity lmetric =
new LinkSimilarity(
loadMainMapper(),
loadIndex(requireString(params, "lucene")),
env.getMainIndex(),
field);
if (params.containsKey("similarity")) {
String sim = requireString(params, "similarity");
if (sim.equals("tfidf")) {
lmetric.setSimilarity(LinkSimilarity.SimFn.TFIDF);
} else if (sim.equals("google")) {
lmetric.setSimilarity(LinkSimilarity.SimFn.GOOGLE);
} else if (sim.equals("logodds")) {
lmetric.setSimilarity(LinkSimilarity.SimFn.LOGODDS);
} else if (sim.equals("jacard")) {
lmetric.setSimilarity(LinkSimilarity.SimFn.JACARD);
} else if (sim.equals("lucene")) {
lmetric.setSimilarity(LinkSimilarity.SimFn.LUCENE);
} else {
throw new IllegalArgumentException("unknown similarity: " + sim);
}
}
if (params.containsKey("minDocFreq")) {
lmetric.setMinDocFreq(requireInteger(params, "minDocFreq"));
}
metric = lmetric;
return metric;
}
/**
* Loads the mapper with the specified name if it is not already loaded.
*
* @param name
* @return The requested mapper.
* @throws ConfigurationException
* @throws IOException
*/
public synchronized ConceptMapper loadMapper(String name)
throws ConfigurationException, IOException {
if (env.hasMapper(name)) {
return env.getMapper(name);
}
info("loading mapper " + name);
JSONObject params = configuration.getMapper(name);
String type = requireString(params, "type");
ConceptMapper mapper;
if (type.equals("dictionary")) {
mapper = getDictionaryMapper(name);
} else if (type.equals("lucene")) {
mapper = getLuceneMapper(name);
} else if (type.equals("ensemble")) {
mapper = getEnsembleMapper(name);
} else if (type.equals("title")) {
mapper = getTitleMapper(name);
} else {
throw new ConfigurationException("unknown type for mapper " + name + ": " + type);
}
env.addMapper(name, mapper);
return mapper;
}
private SimilarityMetric loadEnsembleMetric(String key)
throws IOException, ConfigurationException {
info("loading ensemble metric " + key);
Map<String, Object> params = (Map<String, Object>) configuration.getMetric(key);
setDoEnsembles(false);
List<SimilarityMetric> metrics = loadMetrics();
EnsembleSimilarity similarity =
new EnsembleSimilarity(new SvmEnsemble(), loadMainMapper(), env.getMainIndex());
similarity.setComponents(metrics);
similarity.read(requireDirectory(params, "model"));
similarity.setName(key);
if (params.containsKey("minComponents")) {
similarity.setMinComponents(requireInteger(params, "minComponents"));
}
return similarity;
}