Beispiel #1
0
        public void ThesaurusBuildsNormalizedSynonymToCanonicalFormDictionaryAndIgnoresEmptyLemmas()
        {
            const string canonicalAcronym   = "acronym";
            const string canonicalMicrosoft = "Microsoft";
            var          synonyms           = new Thesaurus(new[]
            {
                new[] { canonicalAcronym, "acornym", "acronyms" },
                Array.Empty <string>(),
                new[] { canonicalMicrosoft, "Microsoft Corporation", "Microsoft corp.", "MSFT" }
            }).Synonyms;

            Assert.AreEqual(7, synonyms.Count());
            Assert.AreEqual(canonicalAcronym, synonyms["acronym"]);
            Assert.AreEqual(canonicalAcronym, synonyms["acornym"]);
            Assert.AreEqual(canonicalAcronym, synonyms["acronyms"]);
            Assert.AreEqual(canonicalMicrosoft, synonyms["microsoft"]);
            Assert.AreEqual(canonicalMicrosoft, synonyms["microsoftcorporation"]);
            Assert.AreEqual(canonicalMicrosoft, synonyms["microsoftcorp"]);
            Assert.AreEqual(canonicalMicrosoft, synonyms["msft"]);
        }