public List extractNounPhraseProductNameCandidate()

in opennlp-similarity/src/main/java/opennlp/tools/apps/relevanceVocabs/PhraseProcessor.java [114:210]
37 lines of code
16 McCabe index (conditional complexity)

	public List<String> extractNounPhraseProductNameCandidate(String sentence) {

		List<String> queryArrayStr = new ArrayList<>();

		if (sentence.split(" ").length ==1) { // this is a word, return empty
			//queryArrayStr.add( sentence);
			return queryArrayStr;
		}
		String quoted1 = StringUtils.substringBetween(sentence, "\"", "\"");
		String quoted2 = StringUtils.substringBetween(sentence, "\'", "\'");
		List<List<ParseTreeChunk>> groupedChunks = nlProc.formGroupedPhrasesFromChunksForPara(sentence);
		if (groupedChunks.size()<1)
			return queryArrayStr;

		List<ParseTreeChunk> nPhrases = groupedChunks.get(0);

		for (ParseTreeChunk ch : nPhrases) {
			StringBuilder query = new StringBuilder();
			int size = ch.getLemmas().size();
			boolean phraseBeingFormed = false;
			for (int i = 0; i < size; i++) {
				if ((ch.getPOSs().get(i).startsWith("N") || ch.getPOSs().get(i)
						.startsWith("J") || ch.getPOSs().get(i).startsWith("CD") ) )
				//		&& StringUtils.isAlpha(ch.getLemmas().get(i)))
				{
					query.append(ch.getLemmas().get(i)).append(" ");
					phraseBeingFormed = true;
				} else
					if ((ch.getPOSs().get(i).startsWith("PR") || ch.getPOSs().get(i).startsWith("IN") || ch.getPOSs().get(i).startsWith("TO")  )
							&& phraseBeingFormed )
						break;
					else if (ch.getPOSs().get(i).startsWith("DT") || ch.getPOSs().get(i).startsWith("CC"))
						continue;
			}
			query = new StringBuilder(query.toString().trim());
			int len = query.toString().split(" ").length;
			if (len > 5 || len < 2) // too long or too short
				continue;

			/*
			if (len < 4 && len>1) { // every word should start with capital
				String[] qs = query.split(" ");
				boolean bAccept = true;
				for (String w : qs) {
					if (w.toLowerCase().equals(w)) // idf only two words then
													// has to be person name,
													// title or geolocation
						bAccept = false;
				}
				if (!bAccept)
					continue;
			}
			*/
			 // individual word, possibly a frequent word
			// if len==1 do nothing

			query = new StringBuilder(query.toString().trim());
			queryArrayStr.add(query.toString());

			}
			/*
			if (queryArrayStr.size() < 1) { // release constraints on NP down to 2
											// keywords
				for (ParseTreeChunk ch : nPhrases) {
					String query = "";
					int size = ch.getLemmas().size();

					for (int i = 0; i < size; i++) {
						if (ch.getPOSs().get(i).startsWith("N")
								|| ch.getPOSs().get(i).startsWith("J")) {
							query += ch.getLemmas().get(i) + " ";
						}
					}
					query = query.trim();
					int len = query.split(" ").length;
					if (len < 2)
						continue;

					query = TextProcessor.fastTokenize(query.toLowerCase(), false)
							.toString().replace('[', ' ').replace(']', ' ').trim();
					if (query.length() > 6)
						queryArrayStr.add(query);
				}
			}
			//queryArrayStr = Utils
			//		.removeDuplicatesFromQueries(queryArrayStr);
			if (quoted1 != null
					&& ((quoted1.length() > 5 && !stopList.isCommonWord(quoted1)) || quoted1
							.length() > 10))
				queryArrayStr.add(quoted1);
			if (quoted2 != null
					&& ((quoted2.length() > 5 && !stopList.isCommonWord(quoted2)) || quoted2
							.length() > 10))
				queryArrayStr.add(quoted2);
			*/
		return queryArrayStr;
	}