/**
* extract adv information from NIH AdvEntry record, and add to a simplenlg
* WordElement For now just extract modifier type
*
* @param wordElement
* @param AdvEntry
*/
private void addAdverbInfo(WordElement wordElement, AdvEntry advEntry)
{
var verbModifier = false;
var sentenceModifier = false;
var intensifier = false;
List <string> modifications = advEntry.GetModification();
foreach (var modification in modifications)
{
if (modification.startsWith("verb_modifier"))
{
verbModifier = true;
}
else if (modification.startsWith("sentence_modifier"))
{
sentenceModifier = true;
}
else if (modification.startsWith("intensifier"))
{
intensifier = true;
}
// ignore other modification types
}
// ignore (for now) other info in record
wordElement.setFeature(LexicalFeature.VERB_MODIFIER, verbModifier);
wordElement.setFeature(LexicalFeature.SENTENCE_MODIFIER,
sentenceModifier);
wordElement.setFeature(LexicalFeature.INTENSIFIER, intensifier);
return;
}
/**
* extract adj information from NIH AdjEntry record, and add to a simplenlg
* WordElement For now just extract position info
*
* @param wordElement
* @param AdjEntry
*/
private void addAdjectiveInfo(WordElement wordElement, AdjEntry adjEntry)
{
var qualitativeAdj = false;
var colourAdj = false;
var classifyingAdj = false;
var predicativeAdj = false;
List <string> positions = adjEntry.GetPosition();
foreach (var position in positions)
{
if (position.startsWith("attrib(1)"))
{
qualitativeAdj = true;
}
else if (position.startsWith("attrib(2)"))
{
colourAdj = true;
}
else if (position.startsWith("attrib(3)"))
{
classifyingAdj = true;
}
else if (position.startsWith("pred"))
{
predicativeAdj = true;
}
// ignore other positions
}
// ignore (for now) other info in record
wordElement.setFeature(LexicalFeature.QUALITATIVE, qualitativeAdj);
wordElement.setFeature(LexicalFeature.COLOUR, colourAdj);
wordElement.setFeature(LexicalFeature.CLASSIFYING, classifyingAdj);
wordElement.setFeature(LexicalFeature.PREDICATIVE, predicativeAdj);
return;
}
/**
* Extract info about the spelling variants of a word from an NIH record,
* and add to the simplenlg Woordelement.
*
* <P>
* Spelling variants are represented as lists of strings, retrievable via
* {@link LexicalFeature#SPELL_VARS}
*
* @param wordElement
* @param record
*/
private void addSpellingVariants(WordElement wordElement, LexRecord record)
{
Vector <string> vars = record.GetSpellingVars();
if (vars != null && !vars.isEmpty())
{
var wordVars = new List <string>();
wordVars.addAll(vars);
wordElement.setFeature(LexicalFeature.SPELL_VARS, wordVars);
}
// we set the default spelling var as the baseForm
wordElement.setFeature(LexicalFeature.DEFAULT_SPELL, wordElement
.getBaseForm());
}
/**
* extract information about acronyms from NIH record, and add to a
* simplenlg WordElement.
*
* <P>
* Acronyms are represented as lists of word elements. Any acronym will have
* a list of full form word elements, retrievable via
* {@link LexicalFeature#ACRONYM_OF}
*
* @param wordElement
* @param record
*/
private void addAcronymInfo(WordElement wordElement, LexRecord record)
{
// NB: the acronyms are actually the full forms of which the word is an
// acronym
List <string> acronyms = record.GetAcronyms();
if (!acronyms.isEmpty())
{
// the list of full forms of which this word is an acronym
List <INLGElement> acronymOf = wordElement
.getFeatureAsElementList(LexicalFeature.ACRONYM_OF);
// keep all acronym full forms and set them up as wordElements
foreach (var fullForm in acronyms)
{
if (fullForm.contains("|"))
{
// get the acronym id
string acronymID = fullForm.substring(
fullForm.indexOf("|") + 1, fullForm.length());
// create the full form element
WordElement fullFormWE = this.getWordByID(acronymID);
if (fullForm != null)
{
// add as full form of this acronym
acronymOf.add(fullFormWE);
// List<NLGElement> fullFormAcronyms = fullFormWE
// .getFeatureAsElementList(LexicalFeature.ACRONYMS);
// fullFormAcronyms.add(wordElement);
// fullFormWE.setFeature(LexicalFeature.ACRONYMS,
// fullFormAcronyms);
}
}
}
// set all the full forms for this acronym
wordElement.setFeature(LexicalFeature.ACRONYM_OF, acronymOf);
}
// if (!acronyms.isEmpty()) {
//
// string acronym = acronyms.get(0);
// // remove anything after a |, this will be an NIH ID
// if (acronym.contains("|"))
// acronym = acronym.substring(0, acronym.indexOf("|"));
// wordElement.setFeature(LexicalFeature.ACRONYM_OF, acronym);
// }
return;
}
/**
* create a simplenlg WordElement from a Word node in a lexicon XML file
*
* @param wordNode
* @return
* @throws XPathUtilException
*/
private WordElement convertNodeToWord(XmlNode wordNode)
{
// if this isn't a Word node, ignore it
if (!wordNode.LocalName.equalsIgnoreCase(XML_WORD))
{
return(null);
}
// // if there is no base, flag an error and return null
// string base = XPathUtil.extractValue(wordNode, Constants.XML_BASE);
// if (base == null) {
// Console.WriteLine("Error in loading XML lexicon: Word with no base");
// return null;
// }
// create word
var word = new WordElement();
var inflections = new List <Inflection>();
// now copy features
var nodes = wordNode.SelectNodes("*");
for (var i = 0; i < nodes.Count; i++)
{
var featureNode = nodes[i];
if (featureNode.NodeType == XmlNodeType.Element)
{
var feature = featureNode.LocalName.trim();
var value = featureNode.InnerText;
if (value != null)
{
value = value.trim();
}
if (feature == null)
{
Debug.WriteLine("Error in XML lexicon node for " + word);
break;
}
if (feature.equalsIgnoreCase(XML_BASE))
{
word.setBaseForm(value);
}
else if (feature.equalsIgnoreCase(XML_CATEGORY))
{
var c = LexicalCategoryExtensions.valueOf(value.toUpperCase());
word.setCategory(c);
}
else if (feature.equalsIgnoreCase(XML_ID))
{
word.setId(value);
}
else if (value == null || value.Equals(""))
{
// if this is an infl code, add it to inflections
Tuple <bool, Inflection> infl = InflectionExtensions.getInflCode(feature);
if (infl.Item1)
{
inflections.Add(infl.Item2);
}
else
{
word.setFeature(feature, true);
}
}
else
{
word.setFeature(feature, value);
}
}
}
// if no infl specified, assume regular
if (inflections.isEmpty())
{
inflections.Add(Inflection.REGULAR);
}
// default inflection code is "reg" if we have it, else random pick form
// infl codes available
var defaultInfl = inflections.Contains(Inflection.REGULAR)
? Inflection.REGULAR
: inflections[0];
word.setFeature(LexicalFeature.DEFAULT_INFL, defaultInfl);
word.setDefaultInflectionalVariant(defaultInfl);
foreach (var infl in inflections)
{
word.addInflectionalVariant(infl);
}
// done, return word
return(word);
}
/**
* extract verb information from NIH VerbEntry record, and add to a
* simplenlg WordElement For now just extract transitive, instransitive,
* and/or ditransitive
*
* @param wordElement
* @param verbEntry
*/
private void addVerbInfo(WordElement wordElement, VerbEntry verbEntry)
{
if (verbEntry == null)
{
// should only happen for aux verbs, which have
// auxEntry instead of verbEntry in NIH Lex
// just flag as transitive and return
wordElement.setFeature(LexicalFeature.INTRANSITIVE, false);
wordElement.setFeature(LexicalFeature.TRANSITIVE, true);
wordElement.setFeature(LexicalFeature.DITRANSITIVE, false);
return;
}
bool intransitiveVerb = notEmpty(verbEntry.GetIntran());
bool transitiveVerb = notEmpty(verbEntry.GetTran()) ||
notEmpty(verbEntry.GetCplxtran());
bool ditransitiveVerb = notEmpty(verbEntry.GetDitran());
wordElement.setFeature(LexicalFeature.INTRANSITIVE, intransitiveVerb);
wordElement.setFeature(LexicalFeature.TRANSITIVE, transitiveVerb);
wordElement.setFeature(LexicalFeature.DITRANSITIVE, ditransitiveVerb);
// add the inflectional variants
List <string> variants = verbEntry.GetVariants();
if (!variants.isEmpty())
{
var wordVariants = new List <Inflection>();
foreach (var v in variants)
{
int index = v.indexOf("|");
string code;
Inflection infl;
if (index > -1)
{
code = v.substring(0, index).toLowerCase().trim();
infl = Inflection.getInflCode(code);
}
else
{
infl = Inflection.getInflCode(v.toLowerCase().trim());
}
if (infl != null)
{
wordElement.addInflectionalVariant(infl);
wordVariants.add(infl);
}
}
// if the variants include "reg", this is the default, otherwise
// just a random pick
Inflection defaultVariant = wordVariants
.contains(Inflection.REGULAR) ||
wordVariants.isEmpty()
? Inflection.REGULAR
: wordVariants.get(0);
// wordElement.setFeature(LexicalFeature.INFLECTIONS, wordVariants);
// wordElement.setFeature(LexicalFeature.DEFAULT_INFL, defaultVariant);
wordElement.setDefaultInflectionalVariant(defaultVariant);
}
// ignore (for now) other info in record
return;
}
/**
* extract noun information from NIH NounEntry record, and add to a
* simplenlg WordElement For now just extract whether count/non-count and
* whether proper or not
*
* @param wordElement
* @param nounEntry
*/
private void addNounInfo(WordElement wordElement, NounEntry nounEntry)
{
bool proper = nounEntry.IsProper();
// bool nonCountVariant = false;
// bool regVariant = false;
// add the inflectional variants
List <string> variants = nounEntry.GetVariants();
if (!variants.isEmpty())
{
var wordVariants = new List <Inflection>();
foreach (var v in variants)
{
int index = v.indexOf("|");
string code;
if (index > -1)
{
code = v.substring(0, index).toLowerCase().trim();
}
else
{
code = v.toLowerCase().trim();
}
Inflection infl = Inflection.getInflCode(code);
if (infl != null)
{
wordVariants.add(infl);
wordElement.addInflectionalVariant(infl);
}
}
// if the variants include "reg", this is the default, otherwise
// just a random pick
Inflection defaultVariant = wordVariants
.Contains(Inflection.REGULAR) ||
wordVariants.isEmpty()
? Inflection.REGULAR
: wordVariants.get(0);
wordElement.setFeature(LexicalFeature.DEFAULT_INFL, defaultVariant);
wordElement.setDefaultInflectionalVariant(defaultVariant);
}
// for (string variant : variants) {
// if (variant.startsWith("uncount")
// || variant.startsWith("groupuncount"))
// nonCountVariant = true;
//
// if (variant.startsWith("reg"))
// regVariant = true;
// // ignore other variant info
// }
// lots of words have both "reg" and "unCount", indicating they
// can be used in either way. Regard such words as normal,
// only flag as nonCount if unambiguous
// wordElement.setFeature(LexicalFeature.NON_COUNT, nonCountVariant
// && !regVariant);
wordElement.setFeature(LexicalFeature.PROPER, proper);
// ignore (for now) other info in record
return;
}
/**
* make a WordElement from a lexical record. Currently just specifies basic
* params and inflections Should do more in the future!
*
* @param record
* @return
*/
private WordElement makeWord(LexRecord record)
{
// get basic data
string baseForm = record.GetBase();
ILexicalCategory category = getSimplenlgCategory(record);
string id = record.GetEui();
// create word class
var wordElement = new WordElement(baseForm, (LexicalCategory)category, id);
// now add type information
switch (category.lexType)
{
case LexicalCategoryEnum.ADJECTIVE:
addAdjectiveInfo(wordElement, record.GetCatEntry().GetAdjEntry());
break;
case LexicalCategoryEnum.ADVERB:
addAdverbInfo(wordElement, record.GetCatEntry().GetAdvEntry());
break;
case LexicalCategoryEnum.NOUN:
addNounInfo(wordElement, record.GetCatEntry().GetNounEntry());
break;
case LexicalCategoryEnum.VERB:
addVerbInfo(wordElement, record.GetCatEntry().GetVerbEntry());
break;
// ignore closed class words
}
var defaultInfl = (Inflection)wordElement
.getDefaultInflectionalVariant();
// now add inflected forms
// if (keepStandardInflections || !standardInflections(record,
// category)) {
foreach (InflVar inflection in record.GetInflVarsAndAgreements()
.GetInflValues())
{
string simplenlgInflection = getSimplenlgInflection(inflection
.GetInflection());
if (simplenlgInflection != null)
{
string inflectedForm = inflection.GetVar();
Inflection inflType = Inflection.getInflCode(inflection
.GetType());
// store all inflectional variants, except for regular ones
// unless explicitly set
if (inflType != null &&
!(Inflection.REGULAR.Equals(inflType) && !this.keepStandardInflections))
{
wordElement.addInflectionalVariant(inflType,
simplenlgInflection, inflectedForm);
}
// if the infl variant is the default, also set this feature on
// the word
if (defaultInfl == null ||
(defaultInfl.Equals(inflType) && !(Inflection.REGULAR
.Equals(inflType) && !this.keepStandardInflections)))
{
wordElement.setFeature(simplenlgInflection, inflectedForm);
}
// wordElement
// .setFeature(simplenlgInflection, inflection.GetVar());
}
}
// }
// add acronym info
addAcronymInfo(wordElement, record);
// now add spelling variants
addSpellingVariants(wordElement, record);
return(wordElement);
}
