/*
* private void loadGenderNumber(String file, String neutralWordsFile) {
* try {
* getWordsFromFile(neutralWordsFile, neutralWords, false);
* Map<List<String>, Gender> temp = IOUtils.readObjectFromURLOrClasspathOrFileSystem(file);
* genderNumber.putAll(temp);
* } catch (IOException e) {
* throw new RuntimeIOException(e);
* } catch (ClassNotFoundException e) {
* throw new RuntimeIOException(e);
* }
* }
*/
/// <summary>Load Bergsma and Lin (2006) gender and number list.</summary>
private void LoadGenderNumber(string file, string neutralWordsFile)
{
try
{
using (BufferedReader reader = IOUtils.ReaderFromString(file))
{
GetWordsFromFile(neutralWordsFile, neutralWords, false);
string[] split = new string[2];
string[] countStr = new string[3];
for (string line; (line = reader.ReadLine()) != null;)
{
StringUtils.SplitOnChar(split, line, '\t');
StringUtils.SplitOnChar(countStr, split[1], ' ');
int male = System.Convert.ToInt32(countStr[0]);
int female = System.Convert.ToInt32(countStr[1]);
int neutral = System.Convert.ToInt32(countStr[2]);
Dictionaries.Gender gender = Dictionaries.Gender.Unknown;
if (male * 0.5 > female + neutral && male > 2)
{
gender = Dictionaries.Gender.Male;
}
else
{
if (female * 0.5 > male + neutral && female > 2)
{
gender = Dictionaries.Gender.Female;
}
else
{
if (neutral * 0.5 > male + female && neutral > 2)
{
gender = Dictionaries.Gender.Neutral;
}
}
}
if (gender == Dictionaries.Gender.Unknown)
{
continue;
}
string[] words = split[0].Split(" ");
IList <string> tokens = Arrays.AsList(words);
genderNumber[tokens] = gender;
}
}
}
catch (IOException e)
{
throw new RuntimeIOException(e);
}
}
/// <summary>This constructor is used to recreate a CorefMention following serialization.</summary>
public CorefMention(Dictionaries.MentionType mentionType, Dictionaries.Number number, Dictionaries.Gender gender, Dictionaries.Animacy animacy, int startIndex, int endIndex, int headIndex, int corefClusterID, int mentionID, int sentNum, IntTuple
position, string mentionSpan)
{
this.mentionType = mentionType;
this.number = number;
this.gender = gender;
this.animacy = animacy;
this.startIndex = startIndex;
this.endIndex = endIndex;
this.headIndex = headIndex;
this.corefClusterID = corefClusterID;
this.mentionID = mentionID;
this.sentNum = sentNum;
this.position = position;
this.mentionSpan = mentionSpan;
}
/// <summary>This constructor builds the external CorefMention class from the internal Mention.</summary>
public CorefMention(Mention m, IntTuple pos)
{
mentionType = m.mentionType;
number = m.number;
gender = m.gender;
animacy = m.animacy;
startIndex = m.startIndex + 1;
endIndex = m.endIndex + 1;
headIndex = m.headIndex + 1;
corefClusterID = m.corefClusterID;
sentNum = m.sentNum + 1;
mentionID = m.mentionID;
mentionSpan = m.SpanToString();
// index starts from 1
position = new IntTuple(2);
position.Set(0, pos.Get(0) + 1);
position.Set(1, pos.Get(1) + 1);
m.headWord.Set(typeof(CorefCoreAnnotations.CorefClusterIdAnnotation), corefClusterID);
}
/// <summary>Loads the CorefChain objects from the serialized buffer.</summary>
/// <param name="reader">the buffer</param>
/// <returns>A map from cluster id to clusters</returns>
/// <exception cref="System.IO.IOException"/>
private static IDictionary <int, CorefChain> LoadCorefChains(BufferedReader reader)
{
string line = reader.ReadLine().Trim();
if (line.IsEmpty())
{
return(null);
}
int clusterCount = System.Convert.ToInt32(line);
IDictionary <int, CorefChain> chains = Generics.NewHashMap();
// read each cluster
for (int c = 0; c < clusterCount; c++)
{
line = reader.ReadLine().Trim();
string[] bits = line.Split("\\s");
int cid = System.Convert.ToInt32(bits[0]);
int mentionCount = System.Convert.ToInt32(bits[1]);
IDictionary <IntPair, ICollection <CorefChain.CorefMention> > mentionMap = Generics.NewHashMap();
CorefChain.CorefMention representative = null;
// read each mention in this cluster
for (int m = 0; m < mentionCount; m++)
{
line = reader.ReadLine();
bits = line.Split("\\s");
IntPair key = new IntPair(System.Convert.ToInt32(bits[0]), System.Convert.ToInt32(bits[1]));
bool rep = bits[2].Equals("1");
Dictionaries.MentionType mentionType = ParseMentionType(bits[3]);
Dictionaries.Number number = ParseNumber(bits[4]);
Dictionaries.Gender gender = ParseGender(bits[5]);
Dictionaries.Animacy animacy = ParseAnimacy(bits[6]);
int startIndex = System.Convert.ToInt32(bits[7]);
int endIndex = System.Convert.ToInt32(bits[8]);
int headIndex = System.Convert.ToInt32(bits[9]);
int clusterID = System.Convert.ToInt32(bits[10]);
int mentionID = System.Convert.ToInt32(bits[11]);
int sentNum = System.Convert.ToInt32(bits[12]);
int posLen = System.Convert.ToInt32(bits[13]);
int[] posElems = new int[posLen];
for (int i = 0; i < posLen; i++)
{
posElems[i] = System.Convert.ToInt32(bits[14 + i]);
}
IntTuple position = new IntTuple(posElems);
string span = UnescapeSpace(bits[14 + posLen]);
CorefChain.CorefMention mention = new CorefChain.CorefMention(mentionType, number, gender, animacy, startIndex, endIndex, headIndex, clusterID, mentionID, sentNum, position, span);
ICollection <CorefChain.CorefMention> mentionsWithThisHead = mentionMap[key];
if (mentionsWithThisHead == null)
{
mentionsWithThisHead = Generics.NewHashSet();
mentionMap[key] = mentionsWithThisHead;
}
mentionsWithThisHead.Add(mention);
if (rep)
{
representative = mention;
}
}
// construct the cluster
CorefChain chain = new CorefChain(cid, mentionMap, representative);
chains[cid] = chain;
}
reader.ReadLine();
return(chains);
}
// static class
/// <exception cref="System.IO.IOException"/>
public static void Main(string[] args)
{
string input = null;
string output = null;
for (int argIndex = 0; argIndex < args.Length;)
{
if (Sharpen.Runtime.EqualsIgnoreCase(args[argIndex], "-input"))
{
input = args[argIndex + 1];
argIndex += 2;
}
else
{
if (Sharpen.Runtime.EqualsIgnoreCase(args[argIndex], "-output"))
{
output = args[argIndex + 1];
argIndex += 2;
}
else
{
throw new ArgumentException("Unknown argument " + args[argIndex]);
}
}
}
if (input == null)
{
throw new ArgumentException("Must specify input with -input");
}
if (output == null)
{
throw new ArgumentException("Must specify output with -output");
}
IDictionary <IList <string>, Dictionaries.Gender> genderNumber = Generics.NewHashMap();
BufferedReader reader = IOUtils.ReaderFromString(input);
for (string line; (line = reader.ReadLine()) != null;)
{
string[] split = line.Split("\t");
string[] countStr = split[1].Split(" ");
int male = System.Convert.ToInt32(countStr[0]);
int female = System.Convert.ToInt32(countStr[1]);
int neutral = System.Convert.ToInt32(countStr[2]);
Dictionaries.Gender gender = Dictionaries.Gender.Unknown;
if (male * 0.5 > female + neutral && male > 2)
{
gender = Dictionaries.Gender.Male;
}
else
{
if (female * 0.5 > male + neutral && female > 2)
{
gender = Dictionaries.Gender.Female;
}
else
{
if (neutral * 0.5 > male + female && neutral > 2)
{
gender = Dictionaries.Gender.Neutral;
}
}
}
if (gender == Dictionaries.Gender.Unknown)
{
continue;
}
string[] words = split[0].Split(" ");
IList <string> tokens = Arrays.AsList(words);
genderNumber[tokens] = gender;
}
IOUtils.WriteObjectToFile(genderNumber, output);
}
