public List<Document> proccessCalculateVectorSpaceModel() { List<Document> documentRetrievals = new ArrayList<Document>(); processQuery(); List<Word> queryWords = query.getWords(); if (queryWords == null || queryWords.size() == 0) return documentRetrievals; Collections.sort(queryWords); double maximumFrequency = queryWords.get(0).getTf(); // calculate TF-IDF for query for (Word wordQ : queryWords) { // check ton tai gia tri cua word(Q) trong word(allDocument) int index = words.indexOf(wordQ); // System.out.println("INDEX ========== " + index); if (index > -1) { wordQ.setIdf(words.get(index).getIdf()); wordQ.setProcessTF(wordQ.getTf() / maximumFrequency); // wordQ.setTf(wordQ.getTf()/maximumFrequency); // System.out.println("IDF: " + words.get(index).getIdf()); } } System.out.println("Query:" + queryWords); for (Document document : documents) { // calculate document with query double cosinQD = 0D; for (Word wordInQuery : query.getWords()) { // tim word cua query trong document. int idWordInD = document.getWords().indexOf(wordInQuery); if (idWordInD > -1) { cosinQD += wordInQuery.getTfidf() * document.getWords().get(idWordInD).getTfidf(); } } document.setCosinWithQuery(cosinQD / (document.getDocumentLeng() * query.getDocumentLeng())); if (document.getCosinWithQuery() > 0) { documentRetrievals.add(document.clone()); } } return documentRetrievals; }
/** * This method will remove all stop words in all documents. see details in http://www.tfidf.com/ * * @param totalDocument ( = 0 -> use default value of total document) */ public void processDocumentsAndCalculateTFIDF() { // remove stop words if (isSemanticSearch) { for (Document document : documents) { document.getWords().removeAll(stopwords); } } // end remove stop words // calculate word appear in all document (add to words) for (Document document : documents) { for (Word word : document.getWords()) { Word w = new Word(0L, word.getWord(), word.getTypeWord(), 0L, 1D); addWordIntoWords(w); } } // update IDF for each word in each document for (Document document : documents) { // calculate IDF for word in the all document. for (Word word : document.getWords()) { int wordId = words.indexOf(word); double idfWord = Math.log(documents.size() / words.get(wordId).getTf()); // set idf for word in each document word.setIdf(idfWord); word.setProcessTF(word.getTf() / document.getWords().size()); words.get(wordId).setIdf(idfWord); } Collections.sort(document.getWords()); if (document.getWords().size() > 10) { // System.out.println("url: " + document.getUrl() + "---" + document.getWords().subList(0, // 10)); } } // System.out.println("word in the all document: " + words); }