// Initialization
public LanguageOutputTests()
{
_vocabulary = new Vocabulary(NumberOfWordVectorDimensions);
// Ensure there are some words in the vocabulary
_vocabulary.Add("hello");
_vocabulary.Add("beautiful");
_vocabulary.Add("world");
}
private void AddReviewToVocabulary(Review review)
{
foreach (string token in review.tokens)
{
vocabulary.Add(token, review);
}
}
public void FindNNeighbors()
{
const int numberOfNeighbors = 5;
const string word = "beautiful";
float[] queryVector = _vocabulary.VectorFromLabel(word);
// Make increasingly remote neighbors
float[][] nearbyVectors = new float[numberOfNeighbors][];
int iVector = 0;
for (iVector = 0; iVector < numberOfNeighbors; iVector++)
{
nearbyVectors[iVector] = new float[NumberOfWordVectorDimensions];
for (int i = 0; i < queryVector.Length; i++)
{
nearbyVectors[iVector][i] = queryVector[i] + Single.Epsilon * iVector;
}
_vocabulary.Add(iVector.ToString(), nearbyVectors[iVector]);
}
// Get nearest neighbors
var retrievedNeighbors = _vocabulary.FindNearestNeighbors(queryVector, numberOfNeighbors);
// Compare to expected neighbors
iVector = 0;
foreach (var tuple in retrievedNeighbors)
{
Assert.True(tuple.Item2.Vector.SequenceEqual(nearbyVectors[iVector++]));
}
}
private void LoadGrimmDictionary(string path)
{
foreach (string line in System.IO.File.ReadLines(path))
{
string[] parts = line.Split(HashArray, 2);
Vocabulary.Add(new Word(parts[0]));
}
Vocabulary.Sort((w1, w2) => - w1.Power.CompareTo(w2.Power));
}
private void IncVocabulary(FrequencyDocument doc)
{
foreach (var token in doc.Tokens.Keys)
{
if (!Vocabulary.Contains(token))
{
Vocabulary.Add(token);
}
}
}
/// <summary>
/// Load a stored embedding map
/// </summary>
/// <param name="path">Path to the database</param>
public void Load(string path)
{
using (var db = new LiteDatabase(path)) {
var collection = db.GetCollection <EmbeddedWord>("vocabulary");
foreach (var word in collection.FindAll())
{
if (!Vocabulary.ContainsKey(word.Word))
{
Vocabulary.Add(word.Word, Vectors.DenseOfArray(word.Vector));
}
}
}
}
public void ExportData()
{
string[] files = Directory.GetFiles(_sourceDir);
for (int i = 0; i < files.Length; i++)
{
log.InfoFormat("========== Processing file {0} of {1} - {2} =============",
i + 1, files.Length,
Path.GetFileName(files[i]));
List <KeyValuePair <string, string> > data = new List <KeyValuePair <string, string> >();
int fileCounter = 0;
uint dataCount = 0;
string targetPath = GetTargetFilePath(files[i], fileCounter);
StreamWriter writer = new StreamWriter(targetPath);
Action <string, string> exportDelegate = (label, target) => {
writer.WriteLine(string.Format("{0} {1}", _vocabulary.IndexOf(label), _vocabulary.IndexOf(target)));
if (++dataCount % _maxDataCountPerFile == 0)
{
writer.Close();
writer.Dispose();
targetPath = GetTargetFilePath(files[i], ++fileCounter);
writer = new StreamWriter(targetPath);
}
;
};
Action <List <string> > vocabMerger = (words) => {
_vocabulary.Add(words);
};
//write the data here
foreach (var sampler in _samplers)
{
sampler.Sample(files[i], exportDelegate, vocabMerger);
}
if (writer != null)
{
writer.Close();
writer.Dispose();
}
_count = _count + dataCount;
log.InfoFormat("Vocabulary size: {0} words", _vocabulary.Count);
log.InfoFormat("Saved {0} examples across {1} partitions.", dataCount, fileCounter + 1);
}
ExportVocabulary();
log.InfoFormat("Finished exporting the dataset - {0} examples in total.", Count);
}
private static Vocabulary ParseFile(string filepath)
{
var contents = System.IO.File.ReadAllLines(filepath);
IEnumerable <WordEntry> pairs;
try
{
pairs = contents.Select(s => ParseString(s)).ToList();
}
catch (Exception e)
{
throw new ParsingException($"When processing file {filepath} got error: {e.Message}");
}
var vocab = new Vocabulary();
foreach (var w in pairs)
{
vocab.Add(w);
}
return(vocab);
}
/// <summary>
/// Defines the appropriate constants for a word (W?FOO, A?FOO, ACT?FOO, PREP?FOO),
/// creating the IWordBuilder if needed.
/// </summary>
/// <param name="word">The Word.</param>
///
void DefineWord([NotNull] IWord word)
{
string rawWord = word.Atom.Text;
if (!Vocabulary.ContainsKey(word))
{
var wAtom = ZilAtom.Parse("W?" + rawWord, Context);
if (Constants.TryGetValue(wAtom, out var constantValue) == false)
{
var wb = Game.DefineVocabularyWord(rawWord);
Vocabulary.Add(word, wb);
Constants.Add(wAtom, wb);
}
else
{
if (constantValue is IWordBuilder wb)
{
Vocabulary.Add(word, wb);
}
else
{
throw new CompilerError(CompilerMessages.Nonvocab_Constant_0_Conflicts_With_Vocab_Word_1, wAtom, word.Atom);
}
}
}
foreach (var pair in Context.ZEnvironment.VocabFormat.GetVocabConstants(word))
{
var atom = ZilAtom.Parse(pair.Key, Context);
if (!Constants.ContainsKey(atom))
{
Constants.Add(atom,
Game.DefineConstant(pair.Key,
Game.MakeOperand(pair.Value)));
}
}
}
