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 override PhraseAttribute Guess(InformedPhrase word)
{
List <KeyValuePair <SpeechPart, double> > kvparts = word.SpeechParts();
if (kvparts.Count != 1)
{
GenderAttribute attr = new GenderAttribute(GenderOptions.Neuter);
attr.strength = new ProbableStrength(0.0, 0.0); // we don't know anything
return(attr);
}
// If it's a proper noun, we guess it might be a human
SpeechPart part = kvparts[0].Key;
if (part == SpeechPart.ProperNoun)
{
GenderAttribute attr = new GenderAttribute(GenderOptions.Human);
attr.strength = new ProbableStrength(1.0, 0.5);
return(attr);
}
else
{
GenderAttribute attr = new GenderAttribute(GenderOptions.Neuter);
attr.strength = new ProbableStrength(1.0, 0.1);
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);
}
public InformedPhrase ReplaceByName(string before, InformedPhrase after)
{
if (name == before)
{
return(after);
}
List <KeyValuePair <PhraseSense, double> > aftersenses = new List <KeyValuePair <PhraseSense, double> >();
foreach (KeyValuePair <PhraseSense, double> sense in senses)
{
// Recurse on each subphrase
List <InformedPhrase> afterphrases = new List <InformedPhrase>();
foreach (InformedPhrase subphrase in sense.Key.Phrases)
{
afterphrases.Add(subphrase.ReplaceByName(before, after));
}
PhraseSense aftersense = new PhraseSense(sense.Key.Definition, afterphrases, sense.Key.Attributes);
aftersenses.Add(new KeyValuePair <PhraseSense, double>(aftersense, sense.Value));
}
string aftername = " " + name + " ";
aftername = aftername.Replace(" " + before + " ", " " + after.name + " ").Trim();
return(new InformedPhrase(aftername, aftersenses));
}
// Is this phrase identical to ours?
public bool IsIdentical(InformedPhrase other)
{
if (name != other.name || senses.Count != other.senses.Count)
{
return(false);
}
// Each sense must match one in the other
foreach (KeyValuePair <PhraseSense, double> sense in senses)
{
bool found = false;
foreach (KeyValuePair <PhraseSense, double> othersense in other.senses)
{
if (sense.Key.IsIdentical(othersense.Key))
{
found = true;
break;
}
}
if (!found)
{
return(false);
}
}
return(true);
}
// If the replacement is found in any of the senses, that's used
public List <string> GenerateWithReplacement(InformedPhrase before, List <string> after, ref bool found)
{
// Replace!
if (this == before)
{
return(after);
}
// Just plug in the word
if (IsTerminal)
{
List <string> single = new List <string>();
single.Add(name);
return(single);
}
List <string> result = new List <string>();
// Try this on each sense, looking for the replacement
foreach (KeyValuePair <PhraseSense, double> sense in senses)
{
// Recurse on each subphrase
result = new List <string>();
foreach (InformedPhrase subphrase in sense.Key.Phrases)
{
result.AddRange(subphrase.GenerateWithReplacement(before, after, ref found));
}
if (found || before == null)
{
return(result); // if the replacement was found, use this result
}
}
return(result);
}
// Guess whether or not a phrase is plural
public static ProbableStrength SeemsPlural(InformedPhrase phrase)
{
ProbableStrength result = new ProbableStrength(0, 0);
double strengthFactor = result.ImproveStrengthStart();
// Noun count is well determined by ending in s
ProbableStrength nounness = PartOSAttribute.SeemsA(phrase, SpeechPart.Noun);
ProbableStrength nisplural;
if (phrase.Name.ToLower().EndsWith("s"))
nisplural = new ProbableStrength(1.0, 0.8);
else
nisplural = new ProbableStrength(0.0, 0.5);
result.ImproveStrength(nounness.Relative(nisplural), ref strengthFactor);
// Verbs that end in s are probably not plural
ProbableStrength verbness = PartOSAttribute.SeemsA(phrase, SpeechPart.Verb);
if (phrase.Name.ToLower().EndsWith("s"))
{
ProbableStrength visplural = new ProbableStrength(0.0, 0.8);
result.ImproveStrength(verbness.Relative(visplural), ref strengthFactor);
}
result.ImproveStrengthFinish(strengthFactor);
return result;
}
// Does this appear to be a reference for something?
// Checks the part of our senses, and our sub-parts
public static ProbableStrength SeemsAnaphora(InformedPhrase phrase)
{
ProbableStrength total = new ProbableStrength(0, 0);
double strengthFactor = total.ImproveStrengthStart();
foreach (KeyValuePair<PhraseSense, double> sense in phrase.Senses)
{
if (sense.Key.Phrases.Count == 1)
total.ImproveStrength(SeemsAnaphora(sense.Key.Phrases[0]), ref strengthFactor);
else
{
ProbableStrength singletotal = sense.Key.SpeechPart().SeemsAnaphora().DownWeight(1.0 / (sense.Key.Phrases.Count + 1));
// Combine together results from subphrases
ProbableStrength phrasetotal = new ProbableStrength(0, 0);
double phraseFactor = phrasetotal.ImproveStrengthStart();
foreach (InformedPhrase subphrase in sense.Key.Phrases)
phrasetotal.ImproveStrength(SeemsAnaphora(subphrase).DownWeight(1.0 / sense.Key.Phrases.Count), ref phraseFactor);
phrasetotal.ImproveStrengthFinish(phraseFactor);
// Use both our and the recursive result
total.ImproveStrength(singletotal.Combine(phrasetotal), ref strengthFactor);
}
}
total.ImproveStrengthFinish(strengthFactor);
return total;
}
public virtual ProbableStrength Describes(PhraseSense sense)
{
ProbableStrength total = new ProbableStrength(0, 0);
double strengthFactor = total.ImproveStrengthStart();
foreach (PhraseAttribute attribute in sense.Attributes)
{
total.ImproveStrength(Match(attribute), ref strengthFactor);
}
if (strengthFactor == 0)
{
// do we match subphrases then?
foreach (InformedPhrase subphrase in sense.Phrases)
{
total.ImproveStrength(Describes(subphrase).DownWeight(1.0 / sense.Phrases.Count), ref strengthFactor);
}
}
if (strengthFactor == 0)
{
// We never found an appropriate attribute-- so guess!
List <KeyValuePair <PhraseSense, double> > senses = new List <KeyValuePair <PhraseSense, double> >();
senses.Add(new KeyValuePair <PhraseSense, double>(sense, 1.0));
InformedPhrase dummy = new InformedPhrase(sense.Name(), senses);
ProbableStrength result = Match(Guess(dummy));
total.ImproveStrength(result, ref strengthFactor);
}
total.ImproveStrengthFinish(strengthFactor);
return(total);
}
// Does this appear to be a reference for something?
// Checks the part of our senses, and our sub-parts
public static ProbableStrength SeemsAnaphora(InformedPhrase phrase)
{
ProbableStrength total = new ProbableStrength(0, 0);
double strengthFactor = total.ImproveStrengthStart();
foreach (KeyValuePair <PhraseSense, double> sense in phrase.Senses)
{
if (sense.Key.Phrases.Count == 1)
{
total.ImproveStrength(SeemsAnaphora(sense.Key.Phrases[0]), ref strengthFactor);
}
else
{
ProbableStrength singletotal = sense.Key.SpeechPart().SeemsAnaphora().DownWeight(1.0 / (sense.Key.Phrases.Count + 1));
// Combine together results from subphrases
ProbableStrength phrasetotal = new ProbableStrength(0, 0);
double phraseFactor = phrasetotal.ImproveStrengthStart();
foreach (InformedPhrase subphrase in sense.Key.Phrases)
{
phrasetotal.ImproveStrength(SeemsAnaphora(subphrase).DownWeight(1.0 / sense.Key.Phrases.Count), ref phraseFactor);
}
phrasetotal.ImproveStrengthFinish(phraseFactor);
// Use both our and the recursive result
total.ImproveStrength(singletotal.Combine(phrasetotal), ref strengthFactor);
}
}
total.ImproveStrengthFinish(strengthFactor);
return(total);
}
public static ProbableStrength SeemsReferencialVerb(InformedPhrase phrase)
{
if (phrase.IsTerminal)
{
string verb = phrase.Name.ToLower();
if (verb == "do" || verb == "did" || verb == "does" || verb == "done" || verb == "doing")
{
return(ProbableStrength.Full);
}
if ((verb == "have" || verb == "had" || verb == "has" || verb == "having") ||
(verb == "was" || verb == "were" || verb == "am" || verb == "is" || verb == "are" || verb == "being" || verb == "been" || verb == "be"))
{
return(ProbableStrength.Half);
}
return(ProbableStrength.Zero);
}
else
{
if (phrase.Senses.Count == 1)
{
KeyValuePair <PhraseSense, double> sense = phrase.Senses[0];
if (sense.Key.Phrases.Count == 1)
{
return(SeemsReferencialVerb(sense.Key.Phrases[0]));
}
}
// Could be, but not so simple!
return(ProbableStrength.None);
}
}
// Merge new sense structures into existing ones
public InformedPhrase Merge(InformedPhrase other)
{
if (this == other || other == null)
{
return(this);
}
InformedPhrase merged = new InformedPhrase(name, senses);
// Add each sense
foreach (KeyValuePair <PhraseSense, double> othersense in other.senses)
{
// Does this sense already exist in our list?
bool found = false;
foreach (KeyValuePair <PhraseSense, double> sense in senses)
{
if (sense.Key.IsIdentical(othersense.Key))
{
if (sense.Value == othersense.Value)
{
found = true;
break;
}
}
}
if (!found)
{
merged.senses.Add(othersense);
}
}
return(merged);
}
// Guess whether or not a phrase is plural
public static ProbableStrength SeemsPlural(InformedPhrase phrase)
{
ProbableStrength result = new ProbableStrength(0, 0);
double strengthFactor = result.ImproveStrengthStart();
// Noun count is well determined by ending in s
ProbableStrength nounness = PartOSAttribute.SeemsA(phrase, SpeechPart.Noun);
ProbableStrength nisplural;
if (phrase.Name.ToLower().EndsWith("s"))
nisplural = new ProbableStrength(1.0, 0.8);
else
nisplural = new ProbableStrength(0.0, 0.5);
result.ImproveStrength(nounness.Relative(nisplural), ref strengthFactor);
// Verbs that end in s are probably not plural
ProbableStrength verbness = PartOSAttribute.SeemsA(phrase, SpeechPart.Verb);
if (phrase.Name.ToLower().EndsWith("s"))
{
ProbableStrength visplural = new ProbableStrength(0.0, 0.8);
result.ImproveStrength(verbness.Relative(visplural), ref strengthFactor);
}
result.ImproveStrengthFinish(strengthFactor);
return result;
}
public bool IsContained(InformedPhrase phrase)
{
foreach (InformedPhrase subphrase in phrases)
{
if (subphrase.IsContained(phrase))
{
return(true);
}
}
return(false);
}
// Is this phrase a subphrase of us?
public bool IsContained(InformedPhrase phrase)
{
if (IsIdentical(phrase))
{
return(true); // a kind of contains
}
// Look through each phrase
foreach (KeyValuePair <PhraseSense, double> sense in senses)
{
if (sense.Key.IsContained(phrase))
{
return(true);
}
}
return(false);
}
// Could this be an anaphora for the given phrase? Check the attributes
public static ProbableStrength AnaphoraOf(InformedPhrase anaphora, PhraseSense prime)
{
if (anaphora.IsContained(prime) || prime.IsContained(anaphora))
return ProbableStrength.Zero; // can't refer to included structure
ProbableStrength total = new ProbableStrength(0, 0);
double strengthFactor = total.ImproveStrengthStart();
foreach (KeyValuePair<PhraseSense, double> sense in anaphora.Senses)
{
ProbableStrength match = AnaphoraOf(sense.Key, prime);
total.ImproveStrength(match.DownWeight(sense.Value), ref strengthFactor);
}
total.ImproveStrengthFinish(strengthFactor);
return total;
}
public override PhraseAttribute Guess(InformedPhrase word)
{
ProbableStrength plural = SeemsPlural(word);
if (plural.strength > .5)
{
NumberAttribute result = new NumberAttribute(NumberOptions.Many);
result.strength.strength = 2.0 * (plural.strength - 0.5);
result.strength.weight = plural.weight;
return result;
}
else
{
NumberAttribute result = new NumberAttribute(NumberOptions.One);
result.strength.strength = 2.0 * (.5 - plural.strength);
result.strength.weight = plural.weight;
return result;
}
}
// 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);
}
}
}
}
// Could this be an anaphora for the given phrase? Check the attributes
public static ProbableStrength AnaphoraOf(InformedPhrase anaphora, PhraseSense prime)
{
if (anaphora.IsContained(prime) || prime.IsContained(anaphora))
{
return(ProbableStrength.Zero); // can't refer to included structure
}
ProbableStrength total = new ProbableStrength(0, 0);
double strengthFactor = total.ImproveStrengthStart();
foreach (KeyValuePair <PhraseSense, double> sense in anaphora.Senses)
{
ProbableStrength match = AnaphoraOf(sense.Key, prime);
total.ImproveStrength(match.DownWeight(sense.Value), ref strengthFactor);
}
total.ImproveStrengthFinish(strengthFactor);
return(total);
}
// Does this attribute describe the word? Consider overriding this
public ProbableStrength Describes(InformedPhrase word)
{
ProbableStrength total = new ProbableStrength(0, 0);
double strengthFactor = total.ImproveStrengthStart();
foreach (KeyValuePair <PhraseSense, double> sense in word.Senses)
{
total.ImproveStrength(Describes(sense.Key).DownWeight(sense.Value), ref strengthFactor);
}
if (strengthFactor == 0)
{
// We never found an appropriate attribute-- so guess!
ProbableStrength result = Match(Guess(word));
total.ImproveStrength(result, ref strengthFactor);
}
total.ImproveStrengthFinish(strengthFactor);
return(total);
}
// Is this actual object in our structure?
public bool IsObjectContained(InformedPhrase phrase)
{
if (Equals(phrase))
{
return(true); // a kind of contains
}
// Look through each phrase
foreach (KeyValuePair <PhraseSense, double> sense in senses)
{
foreach (InformedPhrase subphrase in sense.Key.Phrases)
{
if (subphrase.IsObjectContained(phrase))
{
return(true);
}
}
}
return(false);
}
public static ProbableStrength SeemsReferencePhrase(InformedPhrase phrase, bool not)
{
ProbableStrength total = new ProbableStrength(0, 0);
double strengthFactor = total.ImproveStrengthStart();
foreach (KeyValuePair <PhraseSense, double> sense in phrase.Senses)
{
if (sense.Key.Phrases.Count == 1)
{
total.ImproveStrength(SeemsReferencialVerb(sense.Key.Phrases[0]), ref strengthFactor);
}
else
{
// Is this a "helper" verb followed by an anaphora?
ProbableStrength foundHelper = ProbableStrength.None, foundBoth = ProbableStrength.None;
foreach (InformedPhrase subphr in sense.Key.Phrases)
{
if (!not)
{
foundHelper = foundHelper.Better(SeemsReferencialVerb(subphr));
foundBoth = foundBoth.Better(foundHelper.Relative(SeemsAnaphora(subphr)));
}
else
{
foundHelper = foundHelper.Better(SeemsReferencialVerb(subphr).InverseProbability());
foundBoth = foundBoth.Better(foundHelper.Relative(SeemsAnaphora(subphr).InverseProbability()));
}
}
foundBoth.weight = foundBoth.weight + .5 - foundBoth.weight * .5;
// Use both our and the recursive result
total.ImproveStrength(foundBoth, ref strengthFactor);
}
}
total.ImproveStrengthFinish(strengthFactor);
return(total);
}
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 bool IsContained(InformedPhrase phrase)
{
foreach (InformedPhrase subphrase in phrases)
if (subphrase.IsContained(phrase))
return true;
return false;
}
// What does it look like the attribute for this phrase is? Consider overriding this.
public virtual PhraseAttribute Guess(InformedPhrase word)
{
return new PhraseAttribute(0, 0);
}
// Is this actual object in our structure?
public bool IsObjectContained(InformedPhrase phrase)
{
if (Equals(phrase))
return true; // a kind of contains
// Look through each phrase
foreach (KeyValuePair<PhraseSense, double> sense in senses)
foreach (InformedPhrase subphrase in sense.Key.Phrases)
if (subphrase.IsObjectContained(phrase))
return true;
return false;
}
// Merge new sense structures into existing ones
public InformedPhrase Merge(InformedPhrase other)
{
if (this == other || other == null)
return this;
InformedPhrase merged = new InformedPhrase(name, senses);
// Add each sense
foreach (KeyValuePair<PhraseSense, double> othersense in other.senses)
{
// Does this sense already exist in our list?
bool found = false;
foreach (KeyValuePair<PhraseSense, double> sense in senses)
if (sense.Key.IsIdentical(othersense.Key))
if (sense.Value == othersense.Value)
{
found = true;
break;
}
if (!found)
merged.senses.Add(othersense);
}
return merged;
}
// Does this phrase match a given part of speech?
public static ProbableStrength SeemsA(InformedPhrase phrase, SpeechPart part)
{
return (new PartOSAttribute(part)).Describes(phrase);
}
// Is this phrase identical to ours?
public bool IsIdentical(InformedPhrase other)
{
if (name != other.name || senses.Count != other.senses.Count)
return false;
// Each sense must match one in the other
foreach (KeyValuePair<PhraseSense, double> sense in senses)
{
bool found = false;
foreach (KeyValuePair<PhraseSense, double> othersense in other.senses)
if (sense.Key.IsIdentical(othersense.Key))
{
found = true;
break;
}
if (!found)
return false;
}
return true;
}
// What does it look like the attribute for this phrase is? Consider overriding this.
public virtual PhraseAttribute Guess(InformedPhrase word)
{
return(new PhraseAttribute(0, 0));
}
public InformedPhrase ReplaceByName(string before, InformedPhrase after)
{
if (name == before)
return after;
List<KeyValuePair<PhraseSense, double>> aftersenses = new List<KeyValuePair<PhraseSense, double>>();
foreach (KeyValuePair<PhraseSense, double> sense in senses)
{
// Recurse on each subphrase
List<InformedPhrase> afterphrases = new List<InformedPhrase>();
foreach (InformedPhrase subphrase in sense.Key.Phrases)
afterphrases.Add(subphrase.ReplaceByName(before, after));
PhraseSense aftersense = new PhraseSense(sense.Key.Definition, afterphrases, sense.Key.Attributes);
aftersenses.Add(new KeyValuePair<PhraseSense, double>(aftersense, sense.Value));
}
string aftername = " " + name + " ";
aftername = aftername.Replace(" " + before + " ", " " + after.name + " ").Trim();
return new InformedPhrase(aftername, aftersenses);
}
public static ProbableStrength SeemsReferencePhrase(InformedPhrase phrase, bool not)
{
ProbableStrength total = new ProbableStrength(0, 0);
double strengthFactor = total.ImproveStrengthStart();
foreach (KeyValuePair<PhraseSense, double> sense in phrase.Senses)
{
if (sense.Key.Phrases.Count == 1)
total.ImproveStrength(SeemsReferencialVerb(sense.Key.Phrases[0]), ref strengthFactor);
else
{
// Is this a "helper" verb followed by an anaphora?
ProbableStrength foundHelper = ProbableStrength.None, foundBoth = ProbableStrength.None;
foreach (InformedPhrase subphr in sense.Key.Phrases)
{
if (!not)
{
foundHelper = foundHelper.Better(SeemsReferencialVerb(subphr));
foundBoth = foundBoth.Better(foundHelper.Relative(SeemsAnaphora(subphr)));
}
else
{
foundHelper = foundHelper.Better(SeemsReferencialVerb(subphr).InverseProbability());
foundBoth = foundBoth.Better(foundHelper.Relative(SeemsAnaphora(subphr).InverseProbability()));
}
}
foundBoth.weight = foundBoth.weight + .5 - foundBoth.weight * .5;
// Use both our and the recursive result
total.ImproveStrength(foundBoth, ref strengthFactor);
}
}
total.ImproveStrengthFinish(strengthFactor);
return total;
}
public static ProbableStrength SeemsReferencialVerb(InformedPhrase phrase)
{
if (phrase.IsTerminal)
{
string verb = phrase.Name.ToLower();
if (verb == "do" || verb == "did" || verb == "does" || verb == "done" || verb == "doing")
return ProbableStrength.Full;
if ((verb == "have" || verb == "had" || verb == "has" || verb == "having") ||
(verb == "was" || verb == "were" || verb == "am" || verb == "is" || verb == "are" || verb == "being" || verb == "been" || verb == "be"))
return ProbableStrength.Half;
return ProbableStrength.Zero;
}
else
{
if (phrase.Senses.Count == 1)
{
KeyValuePair<PhraseSense, double> sense = phrase.Senses[0];
if (sense.Key.Phrases.Count == 1)
return SeemsReferencialVerb(sense.Key.Phrases[0]);
}
// Could be, but not so simple!
return ProbableStrength.None;
}
}
// 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 override PhraseAttribute Guess(InformedPhrase word)
{
ProbableStrength plural = SeemsPlural(word);
if (plural.strength > .5)
{
NumberAttribute result = new NumberAttribute(NumberOptions.Many);
result.strength.strength = 2.0 * (plural.strength - 0.5);
result.strength.weight = plural.weight;
return result;
}
else
{
NumberAttribute result = new NumberAttribute(NumberOptions.One);
result.strength.strength = 2.0 * (.5 - plural.strength);
result.strength.weight = plural.weight;
return result;
}
}
// Does this phrase match a given part of speech?
public static ProbableStrength SeemsA(InformedPhrase phrase, SpeechPart part)
{
return((new PartOSAttribute(part)).Describes(phrase));
}
public override PhraseAttribute Guess(InformedPhrase word)
{
List<KeyValuePair<SpeechPart, double>> kvparts = word.SpeechParts();
if (kvparts.Count != 1)
{
GenderAttribute attr = new GenderAttribute(GenderOptions.Neuter);
attr.strength = new ProbableStrength(0.0, 0.0); // we don't know anything
return attr;
}
// If it's a proper noun, we guess it might be a human
SpeechPart part = kvparts[0].Key;
if (part == SpeechPart.ProperNoun)
{
GenderAttribute attr = new GenderAttribute(GenderOptions.Human);
attr.strength = new ProbableStrength(1.0, 0.5);
return attr;
}
else
{
GenderAttribute attr = new GenderAttribute(GenderOptions.Neuter);
attr.strength = new ProbableStrength(1.0, 0.1);
return attr;
}
}
// Does this attribute describe the word? Consider overriding this
public ProbableStrength Describes(InformedPhrase word)
{
ProbableStrength total = new ProbableStrength(0, 0);
double strengthFactor = total.ImproveStrengthStart();
foreach (KeyValuePair<PhraseSense, double> sense in word.Senses)
total.ImproveStrength(Describes(sense.Key).DownWeight(sense.Value), ref strengthFactor);
if (strengthFactor == 0)
{
// We never found an appropriate attribute-- so guess!
ProbableStrength result = Match(Guess(word));
total.ImproveStrength(result, ref strengthFactor);
}
total.ImproveStrengthFinish(strengthFactor);
return total;
}
public virtual ProbableStrength Describes(PhraseSense sense)
{
ProbableStrength total = new ProbableStrength(0, 0);
double strengthFactor = total.ImproveStrengthStart();
foreach (PhraseAttribute attribute in sense.Attributes)
total.ImproveStrength(Match(attribute), ref strengthFactor);
if (strengthFactor == 0)
{
// do we match subphrases then?
foreach (InformedPhrase subphrase in sense.Phrases)
total.ImproveStrength(Describes(subphrase).DownWeight(1.0 / sense.Phrases.Count), ref strengthFactor);
}
if (strengthFactor == 0)
{
// We never found an appropriate attribute-- so guess!
List<KeyValuePair<PhraseSense, double>> senses = new List<KeyValuePair<PhraseSense, double>>();
senses.Add(new KeyValuePair<PhraseSense,double>(sense, 1.0));
InformedPhrase dummy = new InformedPhrase(sense.Name(), senses);
ProbableStrength result = Match(Guess(dummy));
total.ImproveStrength(result, ref strengthFactor);
}
total.ImproveStrengthFinish(strengthFactor);
return total;
}
// If the replacement is found in any of the senses, that's used
public List<string> GenerateWithReplacement(InformedPhrase before, List<string> after, ref bool found)
{
// Replace!
if (this == before)
return after;
// Just plug in the word
if (IsTerminal)
{
List<string> single = new List<string>();
single.Add(name);
return single;
}
List<string> result = new List<string>();
// Try this on each sense, looking for the replacement
foreach (KeyValuePair<PhraseSense, double> sense in senses)
{
// Recurse on each subphrase
result = new List<string>();
foreach (InformedPhrase subphrase in sense.Key.Phrases)
result.AddRange(subphrase.GenerateWithReplacement(before, after, ref found));
if (found || before == null)
return result; // if the replacement was found, use this result
}
return result;
}
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);
}
// Is this phrase a subphrase of us?
public bool IsContained(InformedPhrase phrase)
{
if (IsIdentical(phrase))
return true; // a kind of contains
// Look through each phrase
foreach (KeyValuePair<PhraseSense, double> sense in senses)
if (sense.Key.IsContained(phrase))
return true;
return false;
}