通过 Nest 在 Elasticsearch 中使用同义词

Question

I'm having real trouble getting synonyms to work in Elastic using the Nest API.

I've set up my index and all of the appropriate settings, but when I query based a term that should be a synonym, the results appear as if they haven't been applied at all. Here is my set up:

m_objNode = new Uri(Properties.Settings.Default.strLocalElasticSearchURL);
m_objConnectionSettings = new ConnectionSettings(m_objNode, defaultIndex: "myIndex");
m_objElasticClient = new ElasticClient(m_objConnectionSettings);

IndexSettings indexSettings = new IndexSettings();
indexSettings.NumberOfReplicas = 1;
indexSettings.NumberOfShards = 1;

CustomAnalyzer exclamation = new CustomAnalyzer();
exclamation.Tokenizer = "exclamationTokenizer";

indexSettings.Analysis.Tokenizers.Add("exclamationTokenizer", new PatternTokenizer {
    Pattern = @"!"
});

indexSettings.Analysis.Analyzers.Add("exclamation", exclamation);
indexSettings.Analysis.TokenFilters.Add("synonym", new SynonymTokenFilter { Synonyms = new[] { "tire => tyre", "aluminum => aluminium" }, IgnoreCase = true, Tokenizer = "whitespace" });

m_objElasticClient.CreateIndex(c => c
    .Index("myIndex")
    .InitializeUsing(indexSettings)
    .AddMapping<myClass>(m => m
        .MapFromAttributes()
        .IndexAnalyzer("english")
        .SearchAnalyzer("english")                              
        ));

And the objects I'm indexing look like this:

[ElasticType(IdProperty = "JAUniqueKey")]
public class myClass {

    public string JAUniqueKey { get; set; }
    public int JAItemID { get; set; }
    public string JATitle { get; set; }
    public string JABody { get; set; }
}

I'm trying to get the fields JATitle and JABody to be aligned with the synonyms.

Any ideas sure would be welcome.

Thanks, ScrappyT

Answer 1

You've created token filters correctly but you didn't add it into filters for your custom analyzer.

IndexSettings indexSettings = new IndexSettings();
indexSettings.NumberOfReplicas = 1;
indexSettings.NumberOfShards = 1;

CustomAnalyzer exclamation = new CustomAnalyzer();
exclamation.Tokenizer = "exclamationTokenizer";
exclamation.Filter = new List<string> {"synonym"};
indexSettings.Analysis.Tokenizers.Add(
   "exclamationTokenizer",
    new PatternTokenizer { });

indexSettings.Analysis.Analyzers.Add("exclamation", exclamation);
indexSettings.Analysis.TokenFilters.Add(
    "synonym",
    new SynonymTokenFilter
    {
        Synonyms = new[] { "tire => tyre", "aluminum => aluminium" },
        IgnoreCase = true,
        Tokenizer = "whitespace"
    });

Hope it helps.