elasticsearch入门系列7----补全功能(completion suggester)

1. 建索引

PUT business_line
{
    "mappings": {
        "business" : {
            "properties" : {
                "name" : {
                    "type" : "completion",
                    "analyzer": "ik_max_word",
                    "search_analyzer": "ik_max_word"
                },
                "desc" : {
                    "type": "text",
                    "analyzer": "ik_max_word",
                    "search_analyzer": "ik_max_word"
                }
            }
        }
    }
}

2. 添加数据

POST business_line/business?refresh
{
  "name" : "京东人寿险",
  "desc": "京东人寿险是个十分棒的业务"
}

3. 测试补全功能

POST business_line/_search?pretty
{
    "suggest": {
        "song-suggest" : {
            "prefix" : "京东",
            "completion" : {
                "field" : "name"
            }
        }
    }
}

测试发现补全功能只支持前缀补全，原理是指定type为completion类型之后，会将分词结果在内存中保存一份，保存的数据结构为tries树。同时，type类型不为completion类型的字段不支持补全查询。以上结论可以自行验证。
如果想要支持对所有输入的词汇进行联想查询，可以参照这篇文章：https://www.elastic.co/cn/blog/you-complete-me
4. 补全功能扩展：

PUT novel
{
    "mappings": {
        "novel" : {
            "properties" : {
                "all_input" : {
                    "type" : "completion",
                    "analyzer": "ik_max_word",
                    "search_analyzer": "ik_max_word"
                },
                "name" : {
                    "type": "text",
                    "analyzer": "ik_max_word",
                    "search_analyzer": "ik_max_word"
                },
                 "author" : {
                    "type": "text",
                    "analyzer": "ik_max_word",
                    "search_analyzer": "ik_max_word"
                },
                 "desc" : {
                    "type": "text",
                    "analyzer": "ik_max_word",
                    "search_analyzer": "ik_max_word"
                }
            }
        }
    }
}

input选项为分词结果

POST _analyze
{
  "analyzer": "ik_max_word",
  "text": "四大名著"
}

POST novel/novel?refresh
{
  "all_input" : {
      "input": [
        "三国演义",
        "三国",
        "演义",
        "罗贯中",
        "四大",
        "名著",
        "四大名著"
      ]
    },
  "output":"三国演义",
  "name": "三国演义",
  "author": "罗贯中",
  "desc": "四大名著"
}

POST novel/_search?pretty
{
    "suggest": {
        "song-suggest" : {
            "text" : "三国",
            "completion" : {
                "field" : "all_input"
            }
        }
    },
    "_source": "output"
}

但是这个功能个人感觉有点鸡肋，词汇的录入需要人工干预，不够智能化；而且在5.0以后，output不再支持在字段级别进行配置，需要放到_source中，所以整个补全过程需要查询出文档，拿出文档的output字段。