public GenderAttribute GuessGender(InformedPhrase word, IWordLookup informer)
{
List <string> parts = word.Generate();
string part = parts[0];
ProbableStrength given = IsGivenName(informer, part);
if (given.strength < .25 && given.weight > 0.75)
{
GenderAttribute attr = new GenderAttribute(GenderAttribute.GenderOptions.Neuter);
attr.Strength = given.InverseProbability();
return(attr);
}
ProbableStrength female = IsFemaleName(part);
if (female.strength > 0.5)
{
GenderAttribute attr = new GenderAttribute(GenderAttribute.GenderOptions.Female);
attr.Strength = (new ProbableStrength(2.0 * (female.strength - 0.5), female.weight)).Relative(given);
return(attr);
}
else if (female.strength < 0.5 && female.weight > 0.25)
{
GenderAttribute attr = new GenderAttribute(GenderAttribute.GenderOptions.Male);
attr.Strength = (new ProbableStrength(2.0 * (0.5 - female.strength), female.weight)).Relative(given);
return(attr);
}
else
{
GenderAttribute attr = new GenderAttribute(GenderAttribute.GenderOptions.Human);
attr.Strength = given;
return(attr);
}
}
public static ProbableStrength SeemsReferee(InformedPhrase phrase, IWordLookup informer, ProbableStrength anaphora)
{
List<string> words = phrase.Generate();
double weight = 0;
foreach (string word in words)
weight += informer.GetWeight(word, false);
/* words weight result
* 1 0 0
* 1 1 1
* 2 0 0
* 2 .5 .4
* 2 1 2/3
* 2 1.5 .86
* 2 2 1
* 3 .5 .3
* 3 1 .5
* 3 2 .8
*/
double myweight = 2.0 * weight / (words.Count + weight);
if (words.Count == 1)
myweight /= 2.0; // down-score 1-word answers
if (words.Count > 2)
myweight /= Math.Log(words.Count);
ProbableStrength notanaphora = anaphora.InverseProbability();
return new ProbableStrength(Math.Sqrt(myweight * notanaphora.strength), notanaphora.weight + .5 - notanaphora.weight * .5);
}
// Try to determine if this is any kind of given name
public ProbableStrength IsGivenName(IWordLookup informer, string word)
{
word = word.ToLower();
// Try to look it up in me
string nametype = null;
if (TryGetValue(word, out nametype))
{
return(ProbableStrength.Full);
}
double weight = informer.GetWeight(word, false);
return(new ProbableStrength(weight * weight, 0.5));
}
// Inform everything that looks like a name
public void GenderInformAll(InformedPhrase phrase, IWordLookup informer)
{
foreach (KeyValuePair <PhraseSense, double> sense in phrase.Senses)
{
// Could this be a person? Have to check all noun phrases, because DummyEnglishParser doesn't properly specify Nouns
if (nounparts.Contains(sense.Key.SpeechPart()))
{
GenderInform(sense.Key, informer);
}
else
{
// Otherwise drill down
foreach (InformedPhrase subphr in sense.Key.Phrases)
{
GenderInformAll(subphr, informer);
}
}
}
}
public static ProbableStrength SeemsReferee(InformedPhrase phrase, IWordLookup informer, ProbableStrength anaphora)
{
List <string> words = phrase.Generate();
double weight = 0;
foreach (string word in words)
{
weight += informer.GetWeight(word, false);
}
/* words weight result
* 1 0 0
* 1 1 1
* 2 0 0
* 2 .5 .4
* 2 1 2/3
* 2 1.5 .86
* 2 2 1
* 3 .5 .3
* 3 1 .5
* 3 2 .8
*/
double myweight = 2.0 * weight / (words.Count + weight);
if (words.Count == 1)
{
myweight /= 2.0; // down-score 1-word answers
}
if (words.Count > 2)
{
myweight /= Math.Log(words.Count);
}
ProbableStrength notanaphora = anaphora.InverseProbability();
return(new ProbableStrength(Math.Sqrt(myweight * notanaphora.strength), notanaphora.weight + .5 - notanaphora.weight * .5));
}
public static List <PhraseAttribute> TreebankToAttributes(string part, IWordLookup informer, string word, bool simplified)
{
List <PhraseAttribute> attributes = new List <PhraseAttribute>();
InformedPhrase informed = (informer == null || word == null || word.Contains(" ")) ? null : informer.GetInformed(word, false);
if (part == "NN")
{
attributes.Add(new PartOSAttribute(Noun));
attributes.Add(new NumberAttribute(NumberAttribute.NumberOptions.One));
PhraseSense sense = informed != null?informed.FindSense(SpeechPart.Noun, true) : null;
return(OverrideAttributes(sense, attributes));
}
if (part == "NNS")
{
attributes.Add(new PartOSAttribute(Noun));
attributes.Add(new NumberAttribute(NumberAttribute.NumberOptions.Many));
PhraseSense sense = informed != null?informed.FindSense(SpeechPart.Noun, false) : null;
return(OverrideAttributes(sense, attributes));
}
if (part == "NNP")
{
attributes.Add(new PartOSAttribute(ProperNoun));
attributes.Add(new NumberAttribute(NumberAttribute.NumberOptions.One));
return(attributes);
}
if (part == "NNPS")
{
attributes.Add(new PartOSAttribute(ProperNoun));
attributes.Add(new NumberAttribute(NumberAttribute.NumberOptions.Many));
return(attributes);
}
if (part == "PRP" || part == "PRP$" || part == "WDT" || part == "WP" || part == "WP$" || part == "WDT" || part == "WRB")
{
if (informer == null || word == null || informed == null)
{
if (part == "PRP")
{
attributes.Add(new PartOSAttribute(PersonalPronoun));
}
else if (part == "PRP$")
{
attributes.Add(new PartOSAttribute(PossessivePronoun));
}
else if (part == "WP")
{
attributes.Add(new PartOSAttribute(WhPronoun));
}
else if (part == "WP$")
{
attributes.Add(new PartOSAttribute(PossiveWhPronoun));
}
else
{
attributes.Add(new PartOSAttribute(part));
}
return(attributes);
}
PhraseSense sense = informed.FindSense(PersonalPronoun, true);
if (sense == null)
{
sense = informed.FindSense(ArticulatePronoun, true);
}
if (sense == null)
{
sense = informed.Senses[0].Key;
}
return(sense.Attributes);
}
// Convert phrases
if (part.StartsWith("NP"))
{
part = "NN_P" + part.Substring(2);
}
else if (part.StartsWith("VP"))
{
part = "VB_P" + part.Substring(2);
}
else if (part.StartsWith("PP"))
{
part = "IN_P" + part.Substring(2);
}
if (simplified)
{
int dash = part.IndexOf('-');
if (dash > 0)
{
part = part.Substring(0, dash);
}
}
if (!catalog.ContainsKey(part))
{
attributes.Add(new PartOSAttribute(new SpeechPart(part)));
}
else
{
attributes.Add(new PartOSAttribute(part));
}
return(attributes);
}
public ScoreCalculator(IWordLookup wordLookup)
{
_wordLookup = wordLookup ?? throw new ArgumentNullException(nameof(wordLookup));
}
// Given a noun phrase sense
public void GenderInform(PhraseSense sense, IWordLookup informer)
{
// Drill down to the first noun word, looking for a current attribute
PhraseSense first = sense;
GenderAttribute bypart = (GenderAttribute)first.FindAttribute(typeof(GenderAttribute));
while (bypart == null)
{
// Find a noun sub-phrase
PhraseSense next = null;
foreach (InformedPhrase subfirst in first.Phrases)
{
foreach (KeyValuePair <PhraseSense, double> firstsense in subfirst.Senses)
{
if (nounparts.Contains(firstsense.Key.SpeechPart()))
{
next = firstsense.Key;
break;
}
}
if (next != null)
{
break;
}
}
if (next == null)
{
break;
}
bypart = (GenderAttribute)next.FindAttribute(typeof(GenderAttribute));
if (next.Phrases.Count == 0)
{
break;
}
first = next;
}
if (first.SpeechPart() != SpeechPart.ProperNoun)
{
if (bypart != null && bypart.Strength.weight > 0.5)
{
return; // we seem to know!
}
List <string> words;
if (first.Phrases.Count > 0)
{
words = first.Phrases[0].Generate();
}
else
{
words = new List <string>();
words.Add(first.Name());
}
if (informer.GetWeight(words[0], false) < 0.5)
{
return; // not worth the lookup
}
if (bypart == null)
{
GenderAttribute dummy = new GenderAttribute();
bypart = (GenderAttribute)dummy.Guess(sense.Phrases[0]);
}
}
// Now guess using the given name
InformedPhrase tocheck;
if (first.Phrases.Count > 0)
{
tocheck = first.Phrases[0];
}
else
{
List <KeyValuePair <PhraseSense, double> > senses = new List <KeyValuePair <PhraseSense, double> >();
senses.Add(new KeyValuePair <PhraseSense, double>(first, 1.0));
tocheck = new InformedPhrase(sense.Name(), senses);
}
GenderAttribute gender = GuessGender(tocheck, informer);
if (bypart != null)
{
gender = MergeGenderAttributes(gender, bypart);
}
// Update relevant gender attributes
sense.AddOrUpdateAttribute(gender);
if (sense != first)
{
first.AddOrUpdateAttribute(gender);
}
}
public static List<PhraseAttribute> TreebankToAttributes(string part, IWordLookup informer, string word, bool simplified)
{
List<PhraseAttribute> attributes = new List<PhraseAttribute>();
InformedPhrase informed = (informer == null || word == null || word.Contains(" ")) ? null : informer.GetInformed(word, false);
if (part == "NN")
{
attributes.Add(new PartOSAttribute(Noun));
attributes.Add(new NumberAttribute(NumberAttribute.NumberOptions.One));
PhraseSense sense = informed != null ? informed.FindSense(SpeechPart.Noun, true) : null;
return OverrideAttributes(sense, attributes);
}
if (part == "NNS")
{
attributes.Add(new PartOSAttribute(Noun));
attributes.Add(new NumberAttribute(NumberAttribute.NumberOptions.Many));
PhraseSense sense = informed != null ? informed.FindSense(SpeechPart.Noun, false) : null;
return OverrideAttributes(sense, attributes);
}
if (part == "NNP")
{
attributes.Add(new PartOSAttribute(ProperNoun));
attributes.Add(new NumberAttribute(NumberAttribute.NumberOptions.One));
return attributes;
}
if (part == "NNPS")
{
attributes.Add(new PartOSAttribute(ProperNoun));
attributes.Add(new NumberAttribute(NumberAttribute.NumberOptions.Many));
return attributes;
}
if (part == "PRP" || part == "PRP$" || part == "WDT" || part == "WP" || part == "WP$" || part == "WDT" || part == "WRB")
{
if (informer == null || word == null || informed == null)
{
if (part == "PRP")
attributes.Add(new PartOSAttribute(PersonalPronoun));
else if (part == "PRP$")
attributes.Add(new PartOSAttribute(PossessivePronoun));
else if (part == "WP")
attributes.Add(new PartOSAttribute(WhPronoun));
else if (part == "WP$")
attributes.Add(new PartOSAttribute(PossiveWhPronoun));
else
attributes.Add(new PartOSAttribute(part));
return attributes;
}
PhraseSense sense = informed.FindSense(PersonalPronoun, true);
if (sense == null)
sense = informed.FindSense(ArticulatePronoun, true);
if (sense == null)
sense = informed.Senses[0].Key;
return sense.Attributes;
}
// Convert phrases
if (part.StartsWith("NP"))
part = "NN_P" + part.Substring(2);
else if (part.StartsWith("VP"))
part = "VB_P" + part.Substring(2);
else if (part.StartsWith("PP"))
part = "IN_P" + part.Substring(2);
if (simplified)
{
int dash = part.IndexOf('-');
if (dash > 0)
part = part.Substring(0, dash);
}
if (!catalog.ContainsKey(part))
attributes.Add(new PartOSAttribute(new SpeechPart(part)));
else
attributes.Add(new PartOSAttribute(part));
return attributes;
}