In labeller, nos of pairs that MATCH and DO_NOT_MATCH are printed zin…

…ggAI#87

core/src/main/java/zingg/Labeller.java

@@ -24,6 +24,7 @@ public class Labeller extends ZinggBase {
 
 	protected static String name = "zingg.Labeller";
 	public static final Log LOG = LogFactory.getLog(Labeller.class);
+	long positivePairsCount, negativePairsCount;
 
 	public Labeller() {
 		setZinggOptions(ZinggOptions.LABEL);
@@ -55,6 +56,8 @@ public Dataset<Row> getUnmarkedRecords() throws ZinggClientException {
 				unmarkedRecords = unmarkedRecords.join(markedRecords,
 						unmarkedRecords.col(ColName.CLUSTER_COLUMN).equalTo(markedRecords.col(ColName.CLUSTER_COLUMN)),
 						"left_anti");
+				positivePairsCount = markedRecords.filter(markedRecords.col(ColName.MATCH_FLAG_COL).equalTo(ColValues.MATCH_TYPE_MATCH)).count() / 2;
+				negativePairsCount = markedRecords.filter(markedRecords.col(ColName.MATCH_FLAG_COL).equalTo(ColValues.MATCH_TYPE_NOT_A_MATCH)).count() / 2;
 			} 
 		} catch (Exception e) {
 			LOG.warn("No unmarked record for labelling");
@@ -64,7 +67,7 @@ public Dataset<Row> getUnmarkedRecords() throws ZinggClientException {
 
 	public void processRecordsCli(Dataset<Row> lines) throws ZinggClientException {
 		LOG.info("Processing Records for CLI Labelling");
-
+		printMarkedRecordsStat();
 		if (lines == null || lines.count() == 0) {
 			LOG.info("It seems there are no unmarked records at this moment. Please run findTrainingData job to build some pairs to be labelled and then run this labeler.");
 			return;
@@ -88,14 +91,15 @@ public void processRecordsCli(Dataset<Row> lines) throws ZinggClientException {
 
 				score = currentPair.head().getAs(ColName.SCORE_COL);
 				prediction = currentPair.head().getAs(ColName.PREDICTION_COL);
- 
+	
 				msg1 = String.format("\tRecord pair %d out of %d records to be labelled by the user.\n", index, totalPairs);
 				String matchType = LabelMatchType.get(prediction).msg;
 				msg2 = String.format("\tZingg predicts the records %s with a similarity score of %.2f\n", 
 					matchType, score);
 				String msgHeader = msg1 + msg2;
 
 				selected_option = displayRecordsAndGetUserInput(DSUtil.select(currentPair, displayCols), msgHeader);
+				updateLabellerStat(selected_option);
 				if (selected_option == 9) {
 					LOG.info("User has quit in the middle. Updating the records.");
 					break;
@@ -188,6 +192,22 @@ int readCliInput() {
 		return selection;
 	}
 
+	private void updateLabellerStat(int selected_option) {
+		if (selected_option == ColValues.IS_MATCH_PREDICTION) {
+			++positivePairsCount;
+		}
+		else if (selected_option == ColValues.IS_NOT_A_MATCH_PREDICTION) {
+			++negativePairsCount;
+		}
+		printMarkedRecordsStat();
+	}
+
+	private void printMarkedRecordsStat() {
+ 		System.out.print("\tIn all, " + positivePairsCount + " pair(s) have been marked " + LabelMatchType.MATCH);
+		System.out.print(" and " + negativePairsCount + " pair(s) have been marked " + LabelMatchType.DO_NOT_MATCH);
+		System.out.println();
+	}
+
 	void writeLabelledOutput(Dataset<Row> records) {
 		if (records == null) {
 			LOG.warn("No records to be labelled.");