Examples with NlpEntityOperator edu.uci.ics.texera.dataflow.nlp.entity.NlpEntityOperator used on opensource projects

Example 1 with NlpEntityOperator

use of edu.uci.ics.texera.dataflow.nlp.entity.NlpEntityOperator in project textdb by TextDB.

the class LogicalPlanTest method testGetOutputSchema2.

/*
     * Test getOutputSchema on a valid operator graph.
     *                  -> RegexMatcher -->
     * KeywordSource --<                     >-- Join --> TupleSink
     *                  -> NlpEntityOperator -->
     *
     */
@Test
public void testGetOutputSchema2() throws Exception {
    LogicalPlan logicalPlan = getLogicalPlan2();
    Plan queryPlan = logicalPlan.buildQueryPlan();
    HashMap<String, ISink> sinkHashMap = queryPlan.getSinkMap();
    Assert.assertEquals(1, sinkHashMap.size());
    ISink tupleSink = null;
    for (HashMap.Entry<String, ISink> entry : sinkHashMap.entrySet()) {
        tupleSink = entry.getValue();
    }
    Assert.assertNotNull(tupleSink);
    IOperator join = ((TupleSink) tupleSink).getInputOperator();
    IOperator joinInput1 = ((Join) join).getInnerInputOperator();
    IOperator joinInput2 = ((Join) join).getOuterInputOperator();
    IOperator connectorOut1 = ((RegexMatcher) joinInput1).getInputOperator();
    IOperator connectorOut2 = ((NlpEntityOperator) joinInput2).getInputOperator();
    OneToNBroadcastConnector connector1 = ((ConnectorOutputOperator) connectorOut1).getOwnerConnector();
    OneToNBroadcastConnector connector2 = ((ConnectorOutputOperator) connectorOut2).getOwnerConnector();
    IOperator keywordSource = connector1.getInputOperator();
    join.open();
    Schema expectedJoinOutputSchema = join.getOutputSchema();
    Schema expectedSourceOutputSchema = keywordSource.getOutputSchema();
    Schema expectedMatcherOutputSchema = joinInput1.getOutputSchema();
    Schema expectedNlpEntityOutputSchema = joinInput2.getOutputSchema();
    join.close();
    Schema joinOutputSchema = logicalPlan.getOperatorOutputSchema(JOIN_DISTANCE_ID);
    Schema sourceOutputSchema = logicalPlan.getOperatorOutputSchema(KEYWORD_SOURCE_ID);
    Schema matcherOutputSchema = logicalPlan.getOperatorOutputSchema(REGEX_ID);
    Schema nlpEntityOutputSchema = logicalPlan.getOperatorOutputSchema(NLP_ENTITY_ID);
    Assert.assertEquals(expectedJoinOutputSchema, joinOutputSchema);
    Assert.assertEquals(expectedSourceOutputSchema, sourceOutputSchema);
    Assert.assertEquals(expectedMatcherOutputSchema, matcherOutputSchema);
    Assert.assertEquals(expectedNlpEntityOutputSchema, nlpEntityOutputSchema);
}

Example 2 with NlpEntityOperator

use of edu.uci.ics.texera.dataflow.nlp.entity.NlpEntityOperator in project textdb by TextDB.

the class LogicalPlanTest method testLogicalPlan2.

/*
     * Test a valid operator graph.
     *                  -> RegexMatcher -->
     * KeywordSource --<                     >-- Join --> TupleSink
     *                  -> NlpEntityOperator -->
     *
     */
@Test
public void testLogicalPlan2() throws Exception {
    LogicalPlan logicalPlan = getLogicalPlan2();
    Plan queryPlan = logicalPlan.buildQueryPlan();
    HashMap<String, ISink> sinkHashMap = queryPlan.getSinkMap();
    Assert.assertEquals(1, sinkHashMap.size());
    ISink tupleSink = null;
    for (HashMap.Entry<String, ISink> entry : sinkHashMap.entrySet()) {
        tupleSink = entry.getValue();
    }
    Assert.assertTrue(tupleSink instanceof TupleSink);
    IOperator join = ((TupleSink) tupleSink).getInputOperator();
    Assert.assertTrue(join instanceof Join);
    IOperator joinInput1 = ((Join) join).getInnerInputOperator();
    Assert.assertTrue(joinInput1 instanceof RegexMatcher);
    IOperator joinInput2 = ((Join) join).getOuterInputOperator();
    Assert.assertTrue(joinInput2 instanceof NlpEntityOperator);
    IOperator connectorOut1 = ((RegexMatcher) joinInput1).getInputOperator();
    Assert.assertTrue(connectorOut1 instanceof ConnectorOutputOperator);
    IOperator connectorOut2 = ((NlpEntityOperator) joinInput2).getInputOperator();
    Assert.assertTrue(connectorOut2 instanceof ConnectorOutputOperator);
    HashSet<Integer> connectorIndices = new HashSet<>();
    connectorIndices.add(((ConnectorOutputOperator) connectorOut1).getOutputIndex());
    connectorIndices.add(((ConnectorOutputOperator) connectorOut2).getOutputIndex());
    Assert.assertEquals(connectorIndices.size(), 2);
    OneToNBroadcastConnector connector1 = ((ConnectorOutputOperator) connectorOut1).getOwnerConnector();
    OneToNBroadcastConnector connector2 = ((ConnectorOutputOperator) connectorOut2).getOwnerConnector();
    Assert.assertSame(connector1, connector2);
    IOperator keywordSource = connector1.getInputOperator();
    Assert.assertTrue(keywordSource instanceof KeywordMatcherSourceOperator);
}

Example 3 with NlpEntityOperator

use of edu.uci.ics.texera.dataflow.nlp.entity.NlpEntityOperator in project textdb by TextDB.

the class LogicalPlanTest method testLogicalPlan3.

/*
     * Test a valid operator graph.
     *
     *                  --> RegexMatcher -->
     *                  |                    >-- Join1
     * KeywordSource --< -> NlpEntityOperator -->          >-- Join2 --> TupleSink
     *                  |                           /
     *                  --> FuzzyTokenMatcher ----->
     *
     */
@Test
public void testLogicalPlan3() throws Exception {
    LogicalPlan logicalPlan = getLogicalPlan3();
    Plan queryPlan = logicalPlan.buildQueryPlan();
    HashMap<String, ISink> sinkHashMap = queryPlan.getSinkMap();
    Assert.assertEquals(1, sinkHashMap.size());
    ISink tupleSink = null;
    for (HashMap.Entry<String, ISink> entry : sinkHashMap.entrySet()) {
        tupleSink = entry.getValue();
    }
    Assert.assertTrue(tupleSink instanceof TupleSink);
    IOperator join2 = ((TupleSink) tupleSink).getInputOperator();
    Assert.assertTrue(join2 instanceof Join);
    IOperator join2Input1 = ((Join) join2).getOuterInputOperator();
    Assert.assertTrue(join2Input1 instanceof Join);
    IOperator join2Input2 = ((Join) join2).getInnerInputOperator();
    Assert.assertTrue(join2Input2 instanceof FuzzyTokenMatcher);
    IOperator join1Input1 = ((Join) join2Input1).getInnerInputOperator();
    Assert.assertTrue(join1Input1 instanceof RegexMatcher);
    IOperator join1Input2 = ((Join) join2Input1).getOuterInputOperator();
    Assert.assertTrue(join1Input2 instanceof NlpEntityOperator);
    IOperator connectorOut1 = ((RegexMatcher) join1Input1).getInputOperator();
    Assert.assertTrue(connectorOut1 instanceof ConnectorOutputOperator);
    IOperator connectorOut2 = ((NlpEntityOperator) join1Input2).getInputOperator();
    Assert.assertTrue(connectorOut2 instanceof ConnectorOutputOperator);
    IOperator connectorOut3 = ((FuzzyTokenMatcher) join2Input2).getInputOperator();
    Assert.assertTrue(connectorOut3 instanceof ConnectorOutputOperator);
    HashSet<Integer> connectorIndices = new HashSet<>();
    connectorIndices.add(((ConnectorOutputOperator) connectorOut1).getOutputIndex());
    connectorIndices.add(((ConnectorOutputOperator) connectorOut2).getOutputIndex());
    connectorIndices.add(((ConnectorOutputOperator) connectorOut3).getOutputIndex());
    Assert.assertEquals(connectorIndices.size(), 3);
    OneToNBroadcastConnector connector1 = ((ConnectorOutputOperator) connectorOut1).getOwnerConnector();
    OneToNBroadcastConnector connector2 = ((ConnectorOutputOperator) connectorOut2).getOwnerConnector();
    OneToNBroadcastConnector connector3 = ((ConnectorOutputOperator) connectorOut3).getOwnerConnector();
    Assert.assertSame(connector1, connector2);
    Assert.assertSame(connector1, connector3);
    IOperator keywordSource = connector1.getInputOperator();
    Assert.assertTrue(keywordSource instanceof KeywordMatcherSourceOperator);
}

Example 4 with NlpEntityOperator

use of edu.uci.ics.texera.dataflow.nlp.entity.NlpEntityOperator in project textdb by TextDB.

the class NlpExtractorPerformanceTest method matchNLP.

/*
     * This function does match based on tokenType
     */
public static void matchNLP(String tableName, NlpEntityType tokenType) throws Exception {
    List<String> attributeNames = Arrays.asList(MedlineIndexWriter.ABSTRACT);
    ISourceOperator sourceOperator = new ScanBasedSourceOperator(new ScanSourcePredicate(tableName));
    NlpEntityPredicate nlpEntityPredicate = new NlpEntityPredicate(tokenType, attributeNames, SchemaConstants.SPAN_LIST);
    NlpEntityOperator nlpEntityOperator = new NlpEntityOperator(nlpEntityPredicate);
    nlpEntityOperator.setInputOperator(sourceOperator);
    long startMatchTime = System.currentTimeMillis();
    nlpEntityOperator.open();
    Tuple nextTuple = null;
    int counter = 0;
    while ((nextTuple = nlpEntityOperator.getNextTuple()) != null) {
        ListField<Span> spanListField = nextTuple.getField(SchemaConstants.SPAN_LIST);
        List<Span> spanList = spanListField.getValue();
        counter += spanList.size();
    }
    nlpEntityOperator.close();
    long endMatchTime = System.currentTimeMillis();
    double matchTime = (endMatchTime - startMatchTime) / 1000.0;
    totalMatchingTime += matchTime;
    totalResults += counter;
}