休眠搜索以查找短语的部分匹配项

Question

在我的项目中，我们使用带有 lucene 分析器和太阳能的 hibernate search 4.5。 我为我的客户提供一个文本字段。当他们输入短语时，我想查找名称中包含给定短语的所有 User 实体。

例如，考虑在数据库中有以下标题的条目列表：

[ Alan Smith, John Cane, Juno Taylor, Tom Caner Junior ]

jun 应该返回 Juno Taylor 和 Tom Caner Junior

an 应该返回 Alan Smith、John Cane 和 Tom Caner Junior

    @AnalyzerDef(name = "customanalyzer", tokenizer = @TokenizerDef(factory = WhitespaceTokenizerFactory.class), filters = {
            @TokenFilterDef(factory = LowerCaseFilterFactory.class),
            @TokenFilterDef(factory = SnowballPorterFilterFactory.class, params = { @Parameter(name = "language", value = "English") })

    })
@Analyzer(definition = "customanalyzer")
    public class Student implements Serializable {

        @Column(name = "Fname")
        @Field(index = Index.YES, store = Store.YES, analyze = Analyze.YES)
        private String fname;

        @Column(name = "Lname")
        @Field(index = Index.YES, store = Store.YES, analyze = Analyze.YES)
        private String lname;

    }

我尝试过通配符搜索，但是

Wildcard queries do not apply the analyzer on the matching terms. Otherwise the risk of * or ? being mangled is too high.

Query luceneQuery = mythQB
    .keyword()
      .wildcard()
    .onFields("fname")
    .matching("ju*")
    .createQuery();

我怎样才能做到这一点？

Answer 1

首先，您没有将分析器分配给您的字段，因此目前没有使用它。你应该使用@Field.analyzer。

其次，回答你的问题，这种文本最好用EdgeNGramFilter分析。您应该将此过滤器添加到您的分析器定义中。

编辑：此外，为了防止“sathya”等查询匹配“sanchana”，您应该在查询时使用不同的分析器。

下面是一个完整的例子。

@AnalyzerDef(name = "customanalyzer", tokenizer = @TokenizerDef(factory = WhitespaceTokenizerFactory.class), filters = {
        @TokenFilterDef(factory = LowerCaseFilterFactory.class),
        @TokenFilterDef(factory = SnowballPorterFilterFactory.class, params = { @Parameter(name = "language", value = "English") })
        @TokenFilterDef(factory = EdgeNGramFilterFactory.class, params = { @Parameter(name = "maxGramSize", value = "15") })

})
@AnalyzerDef(name = "customanalyzer_query", tokenizer = @TokenizerDef(factory = WhitespaceTokenizerFactory.class), filters = {
        @TokenFilterDef(factory = LowerCaseFilterFactory.class),
        @TokenFilterDef(factory = SnowballPorterFilterFactory.class, params = { @Parameter(name = "language", value = "English") })

})
public class Student implements Serializable {

    @Column(name = "Fname")
    @Field(index = Index.YES, store = Store.YES, analyze = Analyze.YES, analyzer = @Analyzer(definition = "customanalyzer"))
    private String fname;

    @Column(name = "Lname")
    @Field(index = Index.YES, store = Store.YES, analyze = Analyze.YES, analyzer = @Analyzer(definition = "customanalyzer")))
    private String lname;

}

然后特别提到你想在构建查询时使用这个“查询”分析器：

QueryBuilder queryBuilder = fullTextEntityManager.getSearchFactory().buildQueryBuilder().forEntity(Student.class)
    // Here come the assignments of "query" analyzers
    .overridesForField( "fname", "customanalyzer_query" )
    .overridesForField( "lname", "customanalyzer_query" )
    .get();
// Then it's business as usual
Query luceneQuery = queryBuilder.keyword().onFields("fname", "lname").matching("sathya").createQuery();
FullTextQuery query = fullTextEntityManager.createFullTextQuery(luceneQuery, Student.class);

另请参阅：https://stackoverflow.com/a/43047342/6692043

---

顺便说一句，如果您的数据仅包含名字和姓氏，则不应使用词干提取 (SnowballPorterFilterFactory)：它只会无缘无故地降低搜索的准确性。

Answer 2

为什么不使用标准的TypedQuery？

（String term 是您的搜索词）

TypedQuery<Student> q = em.createQuery(
        "SELECT s " +
        "FROM Student s " +
        "WHERE s.fname like :search " +
        "OR s.lname like :search";
q.setParameter("search", "%" + term + "%");

没有测试这个，但是这样的东西应该可以解决问题。