Search in sources :

Example 1 with OpenNlpAnalysedContentFactory

use of org.apache.stanbol.enhancer.engines.keywordextraction.linking.impl.OpenNlpAnalysedContentFactory in project stanbol by apache.

the class KeywordLinkingEngineTest method testTaxonomyLinker.

/**
     * This tests the EntityLinker functionality (if the expected Entities
     * are linked)
     * @throws Exception
     */
@Test
public void testTaxonomyLinker() throws Exception {
    OpenNlpAnalysedContentFactory acf = OpenNlpAnalysedContentFactory.getInstance(openNLP, new TextAnalyzerConfig());
    EntityLinkerConfig config = new EntityLinkerConfig();
    config.setRedirectProcessingMode(RedirectProcessingMode.FOLLOW);
    EntityLinker linker = new EntityLinker(acf.create(TEST_TEXT, "en"), searcher, config);
    linker.process();
    Map<String, List<String>> expectedResults = new HashMap<String, List<String>>();
    expectedResults.put("Patrick Marshall", new ArrayList<String>(Arrays.asList("urn:test:PatrickMarshall")));
    expectedResults.put("geologist", new ArrayList<String>(//the redirected entity
    Arrays.asList("urn:test:redirect:Geologist")));
    expectedResults.put("New Zealand", new ArrayList<String>(Arrays.asList("urn:test:NewZealand")));
    expectedResults.put("University of Otago", new ArrayList<String>(Arrays.asList("urn:test:UniversityOfOtago", "urn:test:UniversityOfOtago_Texas")));
    for (LinkedEntity linkedEntity : linker.getLinkedEntities().values()) {
        List<String> expectedSuggestions = expectedResults.remove(linkedEntity.getSelectedText());
        assertNotNull("LinkedEntity " + linkedEntity.getSelectedText() + "is not an expected Result (or was found twice)", expectedSuggestions);
        linkedEntity.getSuggestions().iterator();
        assertEquals("Number of suggestions " + linkedEntity.getSuggestions().size() + " != number of expected suggestions " + expectedSuggestions.size() + "for selection " + linkedEntity.getSelectedText(), linkedEntity.getSuggestions().size(), expectedSuggestions.size());
        double score = linkedEntity.getScore();
        for (int i = 0; i < expectedSuggestions.size(); i++) {
            Suggestion suggestion = linkedEntity.getSuggestions().get(i);
            assertEquals("Expecced Suggestion at Rank " + i + " expected: " + expectedSuggestions.get(i) + " suggestion: " + suggestion.getRepresentation().getId(), expectedSuggestions.get(i), suggestion.getRepresentation().getId());
            assertTrue("Score of suggestion " + i + "(" + suggestion.getScore() + " > as of the previous one (" + score + ")", score >= suggestion.getScore());
            score = suggestion.getScore();
        }
    }
}
Also used : LinkedEntity(org.apache.stanbol.enhancer.engines.keywordextraction.impl.LinkedEntity) EntityLinkerConfig(org.apache.stanbol.enhancer.engines.keywordextraction.impl.EntityLinkerConfig) OpenNlpAnalysedContentFactory(org.apache.stanbol.enhancer.engines.keywordextraction.linking.impl.OpenNlpAnalysedContentFactory) HashMap(java.util.HashMap) EntityLinker(org.apache.stanbol.enhancer.engines.keywordextraction.impl.EntityLinker) Suggestion(org.apache.stanbol.enhancer.engines.keywordextraction.impl.Suggestion) List(java.util.List) ArrayList(java.util.ArrayList) TextAnalyzerConfig(org.apache.stanbol.commons.opennlp.TextAnalyzer.TextAnalyzerConfig) Test(org.junit.Test)

Aggregations

ArrayList (java.util.ArrayList)1 HashMap (java.util.HashMap)1 List (java.util.List)1 TextAnalyzerConfig (org.apache.stanbol.commons.opennlp.TextAnalyzer.TextAnalyzerConfig)1 EntityLinker (org.apache.stanbol.enhancer.engines.keywordextraction.impl.EntityLinker)1 EntityLinkerConfig (org.apache.stanbol.enhancer.engines.keywordextraction.impl.EntityLinkerConfig)1 LinkedEntity (org.apache.stanbol.enhancer.engines.keywordextraction.impl.LinkedEntity)1 Suggestion (org.apache.stanbol.enhancer.engines.keywordextraction.impl.Suggestion)1 OpenNlpAnalysedContentFactory (org.apache.stanbol.enhancer.engines.keywordextraction.linking.impl.OpenNlpAnalysedContentFactory)1 Test (org.junit.Test)1