use of com.yahoo.text.interpretation.AnnotationClass in project vespa by vespa-engine.
the class AnnotationTestCase method testAnnotationAPI.
public void testAnnotationAPI() {
Interpretation a = new Interpretation("new york hotel");
a.annotate(0, 3, "token");
a.annotate(0, 8, "state").put("name", "New York");
a.annotate(0, 8, "state").put("country", "US");
a.annotate(0, 8, "state").put("coast", "east");
a.annotate(9, 14, "business");
a.annotate(4, 8, "token");
a.annotate(9, 14, "token");
for (Span span : a.getTokens()) {
assertTrue(span.hasClass(new AnnotationClass("token")));
}
Set<AnnotationClass> annotationClasses = a.getClasses(0, 3);
Set<AnnotationClass> testClass = new HashSet<>(Arrays.asList(new AnnotationClass("token"), new AnnotationClass("state")));
assertEquals(testClass, annotationClasses);
assertNull(a.get("state", "country"));
assertEquals("US", a.get(0, 8, "state", "country"));
assertEquals("new york", a.root().getSubSpans().get(0).getText());
assertEquals("hotel", a.root().getSubSpans().get(1).getText());
assertEquals(2, a.root().getSubSpans().size());
// Test scoring
a.setProbability(5);
Interpretation b = new Interpretation("new york hotel");
b.setProbability(3);
// Test the interpretation API
a.annotate("vespa_query");
assertNotNull(a.get("vespa_query"));
// This is bad about the API, getTokens may not necessairily return what a user thinks a token is
// But it should still be tested
a.annotate(0, 1, "n");
Set<String> testSet = new HashSet<>(Arrays.asList("n", "york", "hotel"));
for (Span span : a.getTokens()) {
assertTrue(testSet.remove(span.getText()));
}
assertEquals(0, testSet.size());
}
Aggregations