/**
* Create a disMaxJunc query rule based on the given search terms as well as the information from
* given ontology terms
*
* @param ontologyTerms
* @param searchTerms
* @return disMaxJunc queryRule
*/
public QueryRule createDisMaxQueryRuleForAttribute(
Set<String> searchTerms, Collection<OntologyTerm> ontologyTerms) {
List<String> queryTerms = new ArrayList<String>();
if (searchTerms != null) {
searchTerms
.stream()
.filter(searchTerm -> StringUtils.isNotBlank(searchTerm))
.forEach(searchTerm -> queryTerms.add(parseQueryString(searchTerm)));
}
// Handle tags with only one ontologyterm
ontologyTerms
.stream()
.filter(ontologyTerm -> !ontologyTerm.getIRI().contains(","))
.forEach(
ot -> {
queryTerms.addAll(parseOntologyTermQueries(ot));
});
QueryRule disMaxQueryRule = createDisMaxQueryRuleForTerms(queryTerms);
// Handle tags with multiple ontologyterms
ontologyTerms
.stream()
.filter(ontologyTerm -> ontologyTerm.getIRI().contains(","))
.forEach(
ot -> {
disMaxQueryRule.getNestedRules().add(createShouldQueryRule(ot.getIRI()));
});
return disMaxQueryRule;
}
/**
* Create a disMaxQueryRule with corresponding boosted value
*
* @param queryTerms
* @param boostValue
* @return a disMaxQueryRule with boosted value
*/
public QueryRule createDisMaxQueryRuleForTermsWithBoost(
List<String> queryTerms, Double boostValue) {
QueryRule finalDisMaxQuery = createDisMaxQueryRuleForTerms(queryTerms);
if (boostValue != null && boostValue.intValue() != 0) {
finalDisMaxQuery.setValue(boostValue);
}
return finalDisMaxQuery;
}
/**
* Create a boolean should query for composite tags containing multiple ontology terms
*
* @param multiOntologyTermIri
* @return return a boolean should queryRule
*/
public QueryRule createShouldQueryRule(String multiOntologyTermIri) {
QueryRule shouldQueryRule = new QueryRule(new ArrayList<QueryRule>());
shouldQueryRule.setOperator(Operator.SHOULD);
for (String ontologyTermIri : multiOntologyTermIri.split(",")) {
OntologyTerm ontologyTerm = ontologyService.getOntologyTerm(ontologyTermIri);
List<String> queryTerms = parseOntologyTermQueries(ontologyTerm);
Double termFrequency = termFrequencyService.getTermFrequency(ontologyTerm.getLabel());
shouldQueryRule
.getNestedRules()
.add(createDisMaxQueryRuleForTermsWithBoost(queryTerms, termFrequency));
}
return shouldQueryRule;
}
/**
* Create disMaxJunc query rule based a list of queryTerm. All queryTerms are lower cased and stop
* words are removed
*
* @param queryTerms
* @return disMaxJunc queryRule
*/
public QueryRule createDisMaxQueryRuleForTerms(List<String> queryTerms) {
List<QueryRule> rules = new ArrayList<QueryRule>();
queryTerms
.stream()
.filter(query -> StringUtils.isNotEmpty(query))
.map(QueryParser::escape)
.map(this::reverseEscapeLuceneChar)
.forEach(
query -> {
rules.add(
new QueryRule(AttributeMetaDataMetaData.LABEL, Operator.FUZZY_MATCH, query));
rules.add(
new QueryRule(
AttributeMetaDataMetaData.DESCRIPTION, Operator.FUZZY_MATCH, query));
});
QueryRule finalDisMaxQuery = new QueryRule(rules);
finalDisMaxQuery.setOperator(Operator.DIS_MAX);
return finalDisMaxQuery;
}
private static Attribute getQueryRuleAttribute(QueryRule queryRule, EntityType entityType) {
String queryRuleField = queryRule.getField();
if (queryRuleField == null) {
throw new MolgenisValidationException(
new ConstraintViolation(
format(
"Query rule with operator [%s] is missing required field",
queryRule.getOperator().toString())));
}
Attribute attr = entityType.getAttribute(queryRuleField);
if (attr == null) {
throw new MolgenisValidationException(
new ConstraintViolation(
format(
"Query rule field [%s] refers to unknown attribute in entity type [%s]",
queryRuleField, entityType.getName())));
}
return attr;
}
private void validateQueryRule(QueryRule queryRule, EntityType entityType) {
QueryRule.Operator operator = queryRule.getOperator();
switch (operator) {
case AND:
case NOT:
case OR:
break;
case EQUALS:
case FUZZY_MATCH:
case FUZZY_MATCH_NGRAM:
case GREATER:
case GREATER_EQUAL:
case LESS:
case LESS_EQUAL:
case LIKE:
{
Attribute attr = getQueryRuleAttribute(queryRule, entityType);
Object value = toQueryRuleValue(queryRule.getValue(), attr);
queryRule.setValue(value);
break;
}
case SEARCH:
{
Object queryRuleValue = queryRule.getValue();
if (queryRuleValue != null && !(queryRuleValue instanceof String)) {
// fix value type
queryRule.setValue(queryRuleValue.toString());
}
break;
}
case IN:
case RANGE:
{
Attribute attr = getQueryRuleAttribute(queryRule, entityType);
Object queryRuleValue = queryRule.getValue();
if (queryRuleValue != null) {
if (!(queryRuleValue instanceof Iterable<?>)) {
throw new MolgenisValidationException(
new ConstraintViolation(
format(
"Query rule with operator [%s] value is of type [%s] instead of [Iterable]",
operator, queryRuleValue.getClass().getSimpleName())));
}
// fix value types
Iterable<?> queryRuleValues = (Iterable<?>) queryRuleValue;
List<Object> values =
stream(queryRuleValues.spliterator(), false)
.map(value -> toQueryRuleValue(value, attr))
.collect(toList());
queryRule.setValue(values);
}
break;
}
case DIS_MAX:
case NESTED:
case SHOULD:
queryRule
.getNestedRules()
.forEach(nestedQueryRule -> validateQueryRule(nestedQueryRule, entityType));
break;
default:
throw new RuntimeException(format("Unknown query operator [%s]", operator.toString()));
}
}
@Test
public void testFindAttributes() {
DefaultEntityMetaData sourceEntityMetaData = new DefaultEntityMetaData("sourceEntityMetaData");
EntityMetaData targetEntityMetaData = new DefaultEntityMetaData("targetEntityMetaData");
DefaultAttributeMetaData targetAttribute = new DefaultAttributeMetaData("targetAttribute");
// Mock the id's of the attribute entities that should be searched
List<String> attributeIdentifiers = Arrays.asList("1", "2");
when(semanticSearchServiceHelper.getAttributeIdentifiers(sourceEntityMetaData))
.thenReturn(attributeIdentifiers);
// Mock the createDisMaxQueryRule method
List<QueryRule> rules = new ArrayList<QueryRule>();
QueryRule targetQueryRuleLabel =
new QueryRule(AttributeMetaDataMetaData.LABEL, Operator.FUZZY_MATCH, "height");
rules.add(targetQueryRuleLabel);
QueryRule targetQueryRuleOntologyTermTag =
new QueryRule(AttributeMetaDataMetaData.LABEL, Operator.FUZZY_MATCH, "standing height");
rules.add(targetQueryRuleOntologyTermTag);
QueryRule targetQueryRuleOntologyTermTagSyn =
new QueryRule(AttributeMetaDataMetaData.LABEL, Operator.FUZZY_MATCH, "length");
rules.add(targetQueryRuleOntologyTermTagSyn);
QueryRule disMaxQueryRule = new QueryRule(rules);
disMaxQueryRule.setOperator(Operator.DIS_MAX);
when(semanticSearchServiceHelper.createDisMaxQueryRuleForAttribute(
targetEntityMetaData, targetAttribute))
.thenReturn(disMaxQueryRule);
MapEntity entity1 =
new MapEntity(
ImmutableMap.of(
AttributeMetaDataMetaData.NAME,
"height_0",
AttributeMetaDataMetaData.LABEL,
"height",
AttributeMetaDataMetaData.DESCRIPTION,
"this is a height measurement in m!"));
List<Entity> attributeMetaDataEntities = Arrays.<Entity>asList(entity1);
List<QueryRule> disMaxQueryRules =
Lists.newArrayList(
new QueryRule(AttributeMetaDataMetaData.IDENTIFIER, Operator.IN, attributeIdentifiers),
new QueryRule(Operator.AND),
disMaxQueryRule);
AttributeMetaData attributeHeight = new DefaultAttributeMetaData("height_0");
AttributeMetaData attributeWeight = new DefaultAttributeMetaData("weight_0");
sourceEntityMetaData.addAttributeMetaData(attributeHeight);
sourceEntityMetaData.addAttributeMetaData(attributeWeight);
// Case 1
when(dataService.findAll(
AttributeMetaDataMetaData.ENTITY_NAME, new QueryImpl(disMaxQueryRules)))
.thenReturn(attributeMetaDataEntities);
Iterable<AttributeMetaData> termsActual1 =
semanticSearchService.findAttributes(
sourceEntityMetaData, targetEntityMetaData, targetAttribute);
Iterable<AttributeMetaData> termsExpected1 = Arrays.<AttributeMetaData>asList(attributeHeight);
assertEquals(termsActual1, termsExpected1);
// Case 2
when(dataService.findAll(
AttributeMetaDataMetaData.ENTITY_NAME, new QueryImpl(disMaxQueryRules)))
.thenReturn(Arrays.<Entity>asList());
Iterable<AttributeMetaData> termsActual2 =
semanticSearchService.findAttributes(
sourceEntityMetaData, targetEntityMetaData, targetAttribute);
Iterable<AttributeMetaData> termsExpected2 = Arrays.<AttributeMetaData>asList();
assertEquals(termsActual2, termsExpected2);
Mockito.reset(ontologyService);
attribute.setDescription("Standing height (Ångstrøm)");
when(ontologyService.findOntologyTerms(
ontologies, ImmutableSet.of("standing", "height", "ångstrøm"), 100))
.thenReturn(ontologyTerms);
Hit<OntologyTerm> result = semanticSearchService.findTags(attribute, ontologies);
assertEquals(result, Hit.<OntologyTerm>create(standingHeight, 0.76471f));
}