public PhraseGraph(JudgeOfRelevancy JoR_, String text, String q, HashSet <String> q_content_words)
{
JoR = JoR_;
phrases = new SortedDictionary <String, Phrase>();
query = q;
query_content_words = q_content_words;
String[] sentences = get_sentences(text.ToLower());
Phrase last_sentence = null;
foreach (var sentence in sentences)
{
Phrase sentence_phrase = ensure_phrase(sentence);
// link neighboring sentences
if (last_sentence != null)
{
// 0.5 < 0.6 so weight will drift back
// earlier sentences are better
last_sentence.add_recipient(sentence_phrase, 0.5);
sentence_phrase.add_recipient(last_sentence, 0.6);
}
String[] words = get_words(sentence);
// link sentences with their words
foreach (var word in words)
{
if (word == "")
{
continue;
}
Phrase word_phrase = ensure_phrase(word);
// weight flows from long sentences to words, to short sentences
word_phrase.add_recipient(sentence_phrase, 1.0);
sentence_phrase.add_recipient(word_phrase, 1.0);
}
last_sentence = sentence_phrase;
}
// link synonyms
//foreach (var word in phrases.Keys)
//{
// // don't include sentences
// if (word.Split().Count() >= 2) { continue; }
// HashSet<String> syns = JoR.get_synonyms(word);
// foreach (String synonym in syns)
// {
// // TODO: fix this :D can't modify the phrases collection while we're enumerating phrases.Keys!
// Phrase word_phrase = ensure_phrase(word);
// Phrase syn_phrase = ensure_phrase(synonym);
// syn_phrase.add_recipient(word_phrase, 1.0 * syns.Count());
// word_phrase.add_recipient(syn_phrase, 1.0 * syns.Count());
// }
//}
}
public PhraseGraph(JudgeOfRelevancy JoR_, String text, String q, HashSet <String> q_content_words)
{
JoR = JoR_;
phrases = new SortedDictionary <String, Phrase>();
query = q;
query_content_words = q_content_words;
List <Tuple <String, List <String> > > sentence_words = TextCleaner.get_sentence_words(text);
Phrase last_sentence = null;
foreach (Tuple <String, List <String> > sentence_and_words in sentence_words)
{
String sentence = sentence_and_words.Item1;
List <String> words = sentence_and_words.Item2;
Phrase sentence_phrase = ensure_phrase(sentence);
if (last_sentence != null) // Link neighbor sentences
{
last_sentence.add_recipient(sentence_phrase, 0.5); // 0.5 < 0.6, so weight will drift back:
sentence_phrase.add_recipient(last_sentence, 0.6); // so earlier sentences better
}
foreach (var word in words) // Link sentences with their words
{
if (word == "")
{
continue;
}
Phrase word_phrase = ensure_phrase(word);
word_phrase.add_recipient(sentence_phrase, 10.0 / words.Count()); // weight flows from long sentences
sentence_phrase.add_recipient(word_phrase, 1.0); // to words, to short sentences
}
last_sentence = sentence_phrase;
}
foreach (String word in phrases.Keys) // Link synonyms
{
if (word.Split().Count() >= 2)
{
continue;
} // don't include sentences
HashSet <String> syns = JoR.get_synonyms(word);
foreach (String synonym in syns)
{
Phrase word_phrase = ensure_phrase(word);
Phrase syn_phrase = ensure_phrase(synonym);
syn_phrase.add_recipient(word_phrase, 1.0 * syns.Count());
word_phrase.add_recipient(syn_phrase, 1.0 * syns.Count());
}
}
}
async public static Task <PhraseGraph> build_PG(String text, String q, HashSet <String> q_content_words)
{
JudgeOfRelevancy JoR_ = await JudgeOfRelevancy.build_JoR();
return(new PhraseGraph(JoR_, text, q, q_content_words));
}
public PhraseGraph(JudgeOfRelevancy JoR_, String text, String q, HashSet<String> q_content_words)
{
JoR = JoR_;
phrases = new SortedDictionary<String, Phrase>();
query = q;
query_content_words = q_content_words;
List<Tuple<String, List<String>>> sentence_words = TextCleaner.get_sentence_words(text);
Phrase last_sentence = null;
foreach (Tuple<String, List<String>> sentence_and_words in sentence_words)
{
String sentence = sentence_and_words.Item1;
List<String> words = sentence_and_words.Item2;
Phrase sentence_phrase = ensure_phrase(sentence);
if (last_sentence != null) // Link neighbor sentences
{
last_sentence.add_recipient(sentence_phrase, 0.5); // 0.5 < 0.6, so weight will drift back:
sentence_phrase.add_recipient(last_sentence, 0.6); // so earlier sentences better
}
foreach(var word in words) // Link sentences with their words
{
if(word=="") {continue;}
Phrase word_phrase = ensure_phrase(word);
word_phrase.add_recipient(sentence_phrase, 10.0 / words.Count()); // weight flows from long sentences
sentence_phrase.add_recipient(word_phrase, 1.0); // to words, to short sentences
}
last_sentence = sentence_phrase;
}
foreach(String word in phrases.Keys) // Link synonyms
{
if (word.Split().Count() >= 2) { continue; } // don't include sentences
HashSet<String> syns = JoR.get_synonyms(word);
foreach(String synonym in syns)
{
Phrase word_phrase = ensure_phrase(word);
Phrase syn_phrase = ensure_phrase(synonym);
syn_phrase.add_recipient(word_phrase, 1.0*syns.Count());
word_phrase.add_recipient(syn_phrase, 1.0*syns.Count());
}
}
}
public PhraseGraph(JudgeOfRelevancy JoR_, String text, String q, HashSet<String> q_content_words)
{
JoR = JoR_;
phrases = new SortedDictionary<String, Phrase>();
query = q;
query_content_words = q_content_words;
String[] sentences = get_sentences(text.ToLower());
Phrase last_sentence = null;
foreach (var sentence in sentences)
{
Phrase sentence_phrase = ensure_phrase(sentence);
// link neighboring sentences
if (last_sentence != null)
{
// 0.5 < 0.6 so weight will drift back
// earlier sentences are better
last_sentence.add_recipient(sentence_phrase, 0.5);
sentence_phrase.add_recipient(last_sentence, 0.6);
}
String[] words = get_words(sentence);
// link sentences with their words
foreach (var word in words)
{
if (word == "") { continue; }
Phrase word_phrase = ensure_phrase(word);
// weight flows from long sentences to words, to short sentences
word_phrase.add_recipient(sentence_phrase, 1.0);
sentence_phrase.add_recipient(word_phrase, 1.0);
}
last_sentence = sentence_phrase;
}
// link synonyms
//foreach (var word in phrases.Keys)
//{
// // don't include sentences
// if (word.Split().Count() >= 2) { continue; }
// HashSet<String> syns = JoR.get_synonyms(word);
// foreach (String synonym in syns)
// {
// // TODO: fix this :D can't modify the phrases collection while we're enumerating phrases.Keys!
// Phrase word_phrase = ensure_phrase(word);
// Phrase syn_phrase = ensure_phrase(synonym);
// syn_phrase.add_recipient(word_phrase, 1.0 * syns.Count());
// word_phrase.add_recipient(syn_phrase, 1.0 * syns.Count());
// }
//}
}