Question

我的ElasticSearch有一堆索引的电视剧集。

每集都标有一系列描述内容主要特征的标签。

现在我想实现一个“类似”的功能，我想搜索给定剧集中标签最多重叠（但不一定全部）的所有剧集。

示例：

Original Episode Tags: ["a","b","c","d"]

Some Other Episode 1: ["a","b"] // should match, 2 matching tags
Some Other Episode 2: ["a","b","c","x","y"] // should match higher, 3 matching tags 
Some Other Episode 2: ["a"] // should match lower, only 1 matching tags
Some Other Episode 3: ["e","f","g"] // shouldn't match, no matching tags

我尝试使用带有should子句的布尔查询，但问题是，一旦达到minimum_should_match要求，文档就会匹配，其余的子句似乎会从得分计算中忽略。

Answer 1

我想我找到了（a）正确的方法：

{
    "query": {
        "function_score": {
            "query": {
                "bool": {
                    "should": [
                        {"term":{"tags":"a"}},
                        {"term":{"tags":"b"}},
                        {"term":{"tags":"c"}}
                    ]
                }
            },
            "functions": [
                {"filter":{"term":{"tags":"a"}},"weight": 5},
                {"filter":{"term":{"tags":"b"}},"weight": 5},
                {"filter":{"term":{"tags":"c"}},"weight": 5}
            ]
        }
    }
}

should子句确保匹配文档中至少有一个标记匹配，而functions子句为每个匹配标记将匹配文档的得分提高5。

根据标签提升搜索结果

1 个答案: