/// <summary>
/// This factory method is used to create the NERClassifierCombiner used in NERCombinerAnnotator
/// (and, thence, in StanfordCoreNLP).
/// </summary>
/// <param name="name">
/// A "x.y" format property name prefix (the "x" part). This is commonly null,
/// and then "ner" is used. If it is the empty string, then no property prefix is used.
/// </param>
/// <param name="passDownProperties">
/// Property names for which the property should be passed down
/// to the NERClassifierCombiner. The default is not to pass down, but pass down is
/// useful for things like charset encoding.
/// </param>
/// <param name="properties">
/// Various properties, including a list in "ner.model".
/// The used ones start with name + "." or are in passDownProperties
/// </param>
/// <returns>An NERClassifierCombiner with the given properties</returns>
public static NERClassifierCombiner CreateNERClassifierCombiner(string name, ICollection <string> passDownProperties, Properties properties)
{
string prefix = (name == null) ? "ner." : name.IsEmpty() ? string.Empty : name + '.';
string modelNames = properties.GetProperty(prefix + "model");
if (modelNames == null)
{
modelNames = DefaultPaths.DefaultNerThreeclassModel + ',' + DefaultPaths.DefaultNerMucModel + ',' + DefaultPaths.DefaultNerConllModel;
}
// but modelNames can still be empty string is set explicitly to be empty!
string[] models;
if (!modelNames.IsEmpty())
{
models = modelNames.Split(",");
}
else
{
// Allow for no real NER model - can just use numeric classifiers or SUTime
log.Info("WARNING: no NER models specified");
models = StringUtils.EmptyStringArray;
}
NERClassifierCombiner nerCombiner;
try
{
bool applyNumericClassifiers = PropertiesUtils.GetBool(properties, prefix + ApplyNumericClassifiersPropertyBase, ApplyNumericClassifiersDefault);
bool useSUTime = PropertiesUtils.GetBool(properties, prefix + NumberSequenceClassifier.UseSutimePropertyBase, NumberSequenceClassifier.UseSutimeDefault);
bool applyRegexner = PropertiesUtils.GetBool(properties, NERClassifierCombiner.ApplyGazetteProperty, NERClassifierCombiner.ApplyGazetteDefault);
Properties combinerProperties;
if (passDownProperties != null)
{
combinerProperties = PropertiesUtils.ExtractSelectedProperties(properties, passDownProperties);
if (useSUTime)
{
// Make sure SUTime parameters are included
Properties sutimeProps = PropertiesUtils.ExtractPrefixedProperties(properties, NumberSequenceClassifier.SutimeProperty + ".", true);
PropertiesUtils.OverWriteProperties(combinerProperties, sutimeProps);
}
}
else
{
// if passDownProperties is null, just pass everything through
combinerProperties = properties;
}
//Properties combinerProperties = PropertiesUtils.extractSelectedProperties(properties, passDownProperties);
NERClassifierCombiner.Language nerLanguage = NERClassifierCombiner.Language.FromString(properties.GetProperty(prefix + "language"), NERClassifierCombiner.Language.English);
nerCombiner = new NERClassifierCombiner(applyNumericClassifiers, nerLanguage, useSUTime, applyRegexner, combinerProperties, models);
}
catch (IOException e)
{
throw new RuntimeIOException(e);
}
return(nerCombiner);
}
/// <exception cref="System.IO.IOException"/>
public NERCombinerAnnotator(Properties properties)
{
IList <string> models = new List <string>();
string modelNames = properties.GetProperty("ner.model");
if (modelNames == null)
{
modelNames = DefaultPaths.DefaultNerThreeclassModel + ',' + DefaultPaths.DefaultNerMucModel + ',' + DefaultPaths.DefaultNerConllModel;
}
if (!modelNames.IsEmpty())
{
Sharpen.Collections.AddAll(models, Arrays.AsList(modelNames.Split(",")));
}
if (models.IsEmpty())
{
// Allow for no real NER model - can just use numeric classifiers or SUTime.
// Have to unset ner.model, so unlikely that people got here by accident.
log.Info("WARNING: no NER models specified");
}
bool applyNumericClassifiers = PropertiesUtils.GetBool(properties, NERClassifierCombiner.ApplyNumericClassifiersProperty, NERClassifierCombiner.ApplyNumericClassifiersDefault);
bool applyRegexner = PropertiesUtils.GetBool(properties, NERClassifierCombiner.ApplyGazetteProperty, NERClassifierCombiner.ApplyGazetteDefault);
bool useSUTime = PropertiesUtils.GetBool(properties, NumberSequenceClassifier.UseSutimeProperty, NumberSequenceClassifier.UseSutimeDefault);
// option for setting doc date to be the present during each annotation
usePresentDateForDocDate = PropertiesUtils.GetBool(properties, "ner." + "usePresentDateForDocDate", false);
// option for setting doc date from a provided string
providedDocDate = PropertiesUtils.GetString(properties, "ner." + "providedDocDate", string.Empty);
Pattern p = Pattern.Compile("[0-9]{4}\\-[0-9]{2}\\-[0-9]{2}");
Matcher m = p.Matcher(providedDocDate);
if (!m.Matches())
{
providedDocDate = string.Empty;
}
NERClassifierCombiner.Language nerLanguage = NERClassifierCombiner.Language.FromString(PropertiesUtils.GetString(properties, NERClassifierCombiner.NerLanguageProperty, null), NERClassifierCombiner.NerLanguageDefault);
bool verbose = PropertiesUtils.GetBool(properties, "ner." + "verbose", false);
string[] loadPaths = Sharpen.Collections.ToArray(models, new string[models.Count]);
Properties combinerProperties = PropertiesUtils.ExtractSelectedProperties(properties, NERClassifierCombiner.DefaultPassDownProperties);
if (useSUTime)
{
// Make sure SUTime parameters are included
Properties sutimeProps = PropertiesUtils.ExtractPrefixedProperties(properties, NumberSequenceClassifier.SutimeProperty + '.', true);
PropertiesUtils.OverWriteProperties(combinerProperties, sutimeProps);
}
NERClassifierCombiner nerCombiner = new NERClassifierCombiner(applyNumericClassifiers, nerLanguage, useSUTime, applyRegexner, combinerProperties, loadPaths);
this.nThreads = PropertiesUtils.GetInt(properties, "ner.nthreads", PropertiesUtils.GetInt(properties, "nthreads", 1));
this.maxTime = PropertiesUtils.GetLong(properties, "ner.maxtime", 0);
this.maxSentenceLength = PropertiesUtils.GetInt(properties, "ner.maxlen", int.MaxValue);
this.language = LanguageInfo.GetLanguageFromString(PropertiesUtils.GetString(properties, "ner.language", "en"));
// in case of Spanish, use the Spanish number regexner annotator
if (language.Equals(LanguageInfo.HumanLanguage.Spanish))
{
Properties spanishNumberRegexNerProperties = new Properties();
spanishNumberRegexNerProperties["spanish.number.regexner.mapping"] = spanishNumberRegexRules;
spanishNumberRegexNerProperties["spanish.number.regexner.validpospattern"] = "^(NUM).*";
spanishNumberRegexNerProperties["spanish.number.regexner.ignorecase"] = "true";
spanishNumberAnnotator = new TokensRegexNERAnnotator("spanish.number.regexner", spanishNumberRegexNerProperties);
}
// set up fine grained ner
SetUpFineGrainedNER(properties);
// set up additional rules ner
SetUpAdditionalRulesNER(properties);
// set up entity mentions
SetUpEntityMentionBuilding(properties);
Verbose = verbose;
this.ner = nerCombiner;
}