Java 如何提高hibernate搜索中整个术语的出现率

对于这样的定义

@AnalyzerDef(name = "standard", charFilters = {
    @CharFilterDef(factory = HTMLStripCharFilterFactory.class) },
    tokenizer = @TokenizerDef(factory = StandardTokenizerFactory.class),
    filters = {
        @TokenFilterDef(factory = StandardFilterFactory.class),
        @TokenFilterDef(factory = LowerCaseFilterFactory.class),
        @TokenFilterDef(factory = StopFilterFactory.class, params = {
            @Parameter(name = "words", value = "/org/apache/lucene/analysis/snowball/english_stop.txt")}),
        @TokenFilterDef(factory = EdgeNGramFilterFactory.class, params = {
            @Parameter(name = "maxGramSize", value = "1"),
            @Parameter(name = "maxGramSize", value = "15")})
    }),

我有两个文档，如

5456

及

5459在丛林中

，搜索词如5459，我想返回结果中高于第一个文档的第二个文档。但是第二个的

fieldNorm

低于第一个

当整个搜索词出现在文档中时，与仅部分出现的搜索词相比，如何提升文档

---

我认为这看起来很相关

春季yaml配置

spring.jpa.properties.hibernate.search.default.similarity : com.example.search.CustomSimilarity

---

我将使用两个字段，一个没有ngrams并增强，另一个有ngrams但未增强

@AnalyzerDefs({
    @AnalyzerDef(
        name = "ngram",
        charFilters = {
            @CharFilterDef(factory = HTMLStripCharFilterFactory.class)
        },
        tokenizer = @TokenizerDef(factory = StandardTokenizerFactory.class),
        filters = {
            @TokenFilterDef(factory = StandardFilterFactory.class),
            @TokenFilterDef(factory = LowerCaseFilterFactory.class),
            @TokenFilterDef(factory = StopFilterFactory.class, params = {
                @Parameter(name = "words", value = "/org/apache/lucene/analysis/snowball/english_stop.txt")}),
            @TokenFilterDef(factory = EdgeNGramFilterFactory.class, params = {
                @Parameter(name = "maxGramSize", value = "1"),
                @Parameter(name = "maxGramSize", value = "15")
            })
        }
    ),
    @AnalyzerDef(
        name = "standard",
        charFilters = {
            @CharFilterDef(factory = HTMLStripCharFilterFactory.class)
        },
        tokenizer = @TokenizerDef(factory = StandardTokenizerFactory.class),
        filters = {
            @TokenFilterDef(factory = StandardFilterFactory.class),
            @TokenFilterDef(factory = LowerCaseFilterFactory.class),
            @TokenFilterDef(factory = StopFilterFactory.class, params = {
                @Parameter(name = "words", value = "/org/apache/lucene/analysis/snowball/english_stop.txt")})
        }
    ),
})
@Indexed
@Entity
public class MyEntity {

    @Fields({
        @Field("myfield_ngram", analyzer = @Analyzer(definition = "ngram")),
        @Field("myfield_standard", analyzer = @Analyzer(definition = "standard"))
    })
    private String myField;

    // ...
}

然后按此方式查询：

QueryBuilder qb = fullTextSession.getSearchFactory()
        .buildQueryBuilder()
        .forEntity( MyEntity.class )
        .overridesForField( "myField_ngram", "standard" ) // Don't generate ngrams when querying, it serves no purpose
        .get();
Query query = qb.keyword()
        .onField( "myField_standard" ).boostedTo(2.0f)
        .andField( "myField_ngram" )
        .matching( "5459 In the Jungle" )
        .createQuery();

---

免责声明：我没有测试这段代码。

hibernate/lucene的人能评论一下这是否很好吗？

QueryBuilder qb = fullTextSession.getSearchFactory()
        .buildQueryBuilder()
        .forEntity( MyEntity.class )
        .overridesForField( "myField_ngram", "standard" ) // Don't generate ngrams when querying, it serves no purpose
        .get();
Query query = qb.keyword()
        .onField( "myField_standard" ).boostedTo(2.0f)
        .andField( "myField_ngram" )
        .matching( "5459 In the Jungle" )
        .createQuery();