进阶-第13__深度探秘搜索技术_基于multi_match+most fiels策略进行multi-field搜索

 

1、为帖子数据增加content字段

从best-fields换成most-fields策略 best-fields策略，主要是说将某一个field匹配尽可能多的关键词的doc优先返回回来 most-fields策略，主要是说尽可能返回更多field匹配到某个关键词的doc，优先返回回来   添加测试数据 POST /forum/article/_bulk { "update": { "_id": "1"} } { "doc" : {"sub_title" : "learning more courses"} } { "update": { "_id": "2"} } { "doc" : {"sub_title" : "learned a lot of course"} } { "update": { "_id": "3"} } { "doc" : {"sub_title" : "we have a lot of fun"} } { "update": { "_id": "4"} } { "doc" : {"sub_title" : "both of them are good"} } { "update": { "_id": "5"} } { "doc" : {"sub_title" : "haha, hello world"} } 查看结果： GET /forum/article/_search {   "took": 2,   "timed_out": false,   "_shards": {     "total": 5,     "successful": 5,     "failed": 0   },   "hits": {     "total": 5,     "max_score": 1,     "hits": [       {         "_index": "forum",         "_type": "article",         "_id": "5",         "_score": 1,         "_source": {           "articleID": "DHJK-B-1395-#Ky5",           "userID": 3,           "hidden": false,           "postDate": "2017-03-01",           "tag": [             "elasticsearch"           ],           "tag_cnt": 1,           "view_cnt": 10,           "title": "this is spark blog",           "content": "spark is best big data solution based on scala ,an programming language similar to java",           "sub_title": "haha, hello world"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "2",         "_score": 1,         "_source": {           "articleID": "KDKE-B-9947-#kL5",           "userID": 1,           "hidden": false,           "postDate": "2017-01-02",           "tag": [             "java"           ],           "tag_cnt": 1,           "view_cnt": 50,           "title": "this is java blog",           "content": "i think java is the best programming language",           "sub_title": "learned a lot of course"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "4",         "_score": 1,         "_source": {           "articleID": "QQPX-R-3956-#aD8",           "userID": 2,           "hidden": true,           "postDate": "2017-01-02",           "tag": [             "java",             "elasticsearch"           ],           "tag_cnt": 2,           "view_cnt": 80,           "title": "this is java, elasticsearch, hadoop blog",           "content": "elasticsearch and hadoop are all very good solution, i am a beginner",           "sub_title": "both of them are good"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "1",         "_score": 1,         "_source": {           "articleID": "XHDK-A-1293-#fJ3",           "userID": 1,           "hidden": false,           "postDate": "2017-01-01",           "tag": [             "java",             "hadoop"           ],           "tag_cnt": 2,           "view_cnt": 30,           "title": "this is java and elasticsearch blog",           "content": "i like to write best elasticsearch article",           "sub_title": "learning more courses"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "3",         "_score": 1,         "_source": {           "articleID": "JODL-X-1937-#pV7",           "userID": 2,           "hidden": false,           "postDate": "2017-01-01",           "tag": [             "hadoop"           ],           "tag_cnt": 1,           "view_cnt": 100,           "title": "this is elasticsearch blog",           "content": "i am only an elasticsearch beginner",           "sub_title": "we have a lot of fun"         }       }     ]   } }     搜索sub_title 中包含learning courses 的数据 GET /forum/article/_search {   "query": {     "match": {       "sub_title": "learning courses"     }   } } 结果： {   "took": 1,   "timed_out": false,   "_shards": {     "total": 5,     "successful": 5,     "failed": 0   },   "hits": {     "total": 2,     "max_score": 0.68064547,     "hits": [       {         "_index": "forum",         "_type": "article",         "_id": "2",         "_score": 0.68064547,         "_source": {           "articleID": "KDKE-B-9947-#kL5",           "userID": 1,           "hidden": false,           "postDate": "2017-01-02",           "tag": [             "java"           ],           "tag_cnt": 1,           "view_cnt": 50,           "title": "this is java blog",           "content": "i think java is the best programming language",           "sub_title": "learned a lot of course"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "1",         "_score": 0.25316024,         "_source": {           "articleID": "XHDK-A-1293-#fJ3",           "userID": 1,           "hidden": false,           "postDate": "2017-01-01",           "tag": [             "java",             "hadoop"           ],           "tag_cnt": 2,           "view_cnt": 30,           "title": "this is java and elasticsearch blog",           "content": "i like to write best elasticsearch article",           "sub_title": "learning more courses"         }       }     ]   } }   注意：由于sub_title 是text 默认采用的是englist分词器，会进行时态等的转换。而当进行搜索时，我们希望          "sub_title": "learning more courses"能位于最前面，显然目前是达不到的。故需要对sub_title 建立子字段。 对sub_title建立mapping POST /forum/_mapping/article {   "properties": {       "sub_title": {           "type":     "string",           "analyzer": "english",//会用english分词器，会将时态还原           "fields": {               "std":   {                   "type":     "string",                   "analyzer": "standard"//不会进行时态转换               }           }       }   } }       sub_title用的是enligsh analyzer，所以还原了单词   为什么，因为如果我们用的是类似于english analyzer这种分词器的话，就会将单词还原为其最基本的形态，stemmer learning --> learn learned --> learn courses --> course   sub_titile: learning coureses --> learn course   { "doc" : {"sub_title" : "learned a lot of course"} }，就排在了{ "doc" : {"sub_title" : "learning more courses"} }的前面   很绕。。。。我自己都觉得很绕   很多东西，你看文字就觉得很绕，然后用语言去表述，也很绕，但是我觉得，用语言去说，相对来说会好一点点 进行multi-field进行搜索 GET /forum/article/_search {    "query": {         "multi_match": {             "query":  "learning courses",             "type":   "most_fields", //最多匹配出             "fields": [ "sub_title", "sub_title.std" ]//sub_title.std 采用standard分词器，不进行时态转换         }     } } 结果： {   "took": 2,   "timed_out": false,   "_shards": {     "total": 5,     "successful": 5,     "failed": 0   },   "hits": {     "total": 2,     "max_score": 1.219939,     "hits": [       {         "_index": "forum",         "_type": "article",         "_id": "2",         "_score": 1.219939,         "_source": {           "articleID": "KDKE-B-9947-#kL5",           "userID": 1,           "hidden": false,           "postDate": "2017-01-02",           "tag": [             "java"           ],           "tag_cnt": 1,           "view_cnt": 50,           "title": "this is java blog",           "content": "i think java is the best programming language",           "sub_title": "learned a lot of course"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "1",         "_score": 1.012641,         "_source": {           "articleID": "XHDK-A-1293-#fJ3",           "userID": 1,           "hidden": false,           "postDate": "2017-01-01",           "tag": [             "java",             "hadoop"           ],           "tag_cnt": 2,           "view_cnt": 30,           "title": "this is java and elasticsearch blog",           "content": "i like to write best elasticsearch article",           "sub_title": "learning more courses"         }       }     ]   } }       你问我，具体的分数怎么算出来的，很难说，因为这个东西很复杂， 还不只是TF/IDF算法。因为不同的query，不同的语法，都有不同的计算score的细节。   与best_fields的区别   （1）best_fields，是对多个field进行搜索，挑选某个field匹配度最高的那个分数，同时在多个query最高分相同的情况下，在一定程度上考虑其他query的分数。简单来说，你对多个field进行搜索，就想搜索到某一个field尽可能包含更多关键字的数据   优点：通过best_fields策略，以及综合考虑其他field，还有minimum_should_match支持，可以尽可能精准地将匹配的结果推送到最前面 缺点：除了那些精准匹配的结果，其他差不多大的结果，排序结果不是太均匀，没有什么区分度了   实际的例子：百度之类的搜索引擎，最匹配的到最前面，但是其他的就没什么区分度了   （2）most_fields，综合多个field一起进行搜索，尽可能多地让所有field的query参与到总分数的计算中来，此时就会是个大杂烩，出现类似best_fields案例最开始的那个结果，结果不一定精准，某一个document的一个field包含更多的关键字，但是因为其他document有更多field匹配到了，所以排在了前面；所以需要建立类似sub_title.std这样的field，尽可能让某一个field精准匹配query string，贡献更高的分数，将更精准匹配的数据排到前面   优点：将尽可能匹配更多field的结果推送到最前面，整个排序结果是比较均匀的 缺点：可能那些精准匹配的结果，无法推送到最前面   实际的例子：wiki，明显的most_fields策略，搜索结果比较均匀，但是的确要翻好几页才能找到最匹配的结果

查看所有的数据

GET /forum/article/_search 结果： {   "took": 10,   "timed_out": false,   "_shards": {     "total": 5,     "successful": 5,     "failed": 0   },   "hits": {     "total": 5,     "max_score": 1,     "hits": [       {         "_index": "forum",         "_type": "article",         "_id": "5",         "_score": 1,         "_source": {           "articleID": "DHJK-B-1395-#Ky5",           "userID": 3,           "hidden": false,           "postDate": "2017-03-01",           "tag": [             "elasticsearch"           ],           "tag_cnt": 1,           "view_cnt": 10,           "title": "this is spark blog",           "content": "spark is best big data solution based on scala ,an programming language similar to java"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "2",         "_score": 1,         "_source": {           "articleID": "KDKE-B-9947-#kL5",           "userID": 1,           "hidden": false,           "postDate": "2017-01-02",           "tag": [             "java"           ],           "tag_cnt": 1,           "view_cnt": 50,           "title": "this is java blog",           "content": "i think java is the best programming language"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "4",         "_score": 1,         "_source": {           "articleID": "QQPX-R-3956-#aD8",           "userID": 2,           "hidden": true,           "postDate": "2017-01-02",           "tag": [             "java",             "elasticsearch"           ],           "tag_cnt": 2,           "view_cnt": 80,           "title": "this is java, elasticsearch, hadoop blog",           "content": "elasticsearch and hadoop are all very good solution, i am a beginner"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "1",         "_score": 1,         "_source": {           "articleID": "XHDK-A-1293-#fJ3",           "userID": 1,           "hidden": false,           "postDate": "2017-01-01",           "tag": [             "java",             "hadoop"           ],           "tag_cnt": 2,           "view_cnt": 30,           "title": "this is java and elasticsearch blog",           "content": "i like to write best elasticsearch article"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "3",         "_score": 1,         "_source": {           "articleID": "JODL-X-1937-#pV7",           "userID": 2,           "hidden": false,           "postDate": "2017-01-01",           "tag": [             "hadoop"           ],           "tag_cnt": 1,           "view_cnt": 100,           "title": "this is elasticsearch blog",           "content": "i am only an elasticsearch beginner"         }       }     ]   } }

 

2、搜索title或content中包含java或solution的帖子

GET /forum/article/_search {     "query": {         "bool": {             "should": [                 { "match": { "title": "java solution" }},                 { "match": { "content":  "java solution" }}             ]         }     } } 结果： {   "took": 12,   "timed_out": false,   "_shards": {     "total": 5,     "successful": 5,     "failed": 0   },   "hits": {     "total": 4,     "max_score": 0.8849759,     "hits": [       {         "_index": "forum",         "_type": "article",         "_id": "2",         "_score": 0.8849759,         "_source": {           "articleID": "KDKE-B-9947-#kL5",           "userID": 1,           "hidden": false,           "postDate": "2017-01-02",           "tag": [             "java"           ],           "tag_cnt": 1,           "view_cnt": 50,           "title": "this is java blog",           "content": "i think java is the best programming language"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "4",         "_score": 0.7120095,         "_source": {           "articleID": "QQPX-R-3956-#aD8",           "userID": 2,           "hidden": true,           "postDate": "2017-01-02",           "tag": [             "java",             "elasticsearch"           ],           "tag_cnt": 2,           "view_cnt": 80,           "title": "this is java, elasticsearch, hadoop blog",           "content": "elasticsearch and hadoop are all very good solution, i am a beginner"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "5",         "_score": 0.56008905,         "_source": {           "articleID": "DHJK-B-1395-#Ky5",           "userID": 3,           "hidden": false,           "postDate": "2017-03-01",           "tag": [             "elasticsearch"           ],           "tag_cnt": 1,           "view_cnt": 10,           "title": "this is spark blog",           "content": "spark is best big data solution based on scala ,an programming language similar to java"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "1",         "_score": 0.26742277,         "_source": {           "articleID": "XHDK-A-1293-#fJ3",           "userID": 1,           "hidden": false,           "postDate": "2017-01-01",           "tag": [             "java",             "hadoop"           ],           "tag_cnt": 2,           "view_cnt": 30,           "title": "this is java and elasticsearch blog",           "content": "i like to write best elasticsearch article"         }       }     ]   } }

 

下面这个就是multi-field搜索，多字段搜索

3、结果分析

期望既有solution又有java

期望的是doc5，结果是doc2,doc4排在了前面

 

计算每个document的relevance score：每个query的分数，乘以matched query数量，除以总query数量

 

算一下doc4的分数

{         "_index": "forum",         "_type": "article",         "_id": "4",         "_score": 0.7120095,         "_source": {           "articleID": "QQPX-R-3956-#aD8",           "userID": 2,           "hidden": true,           "postDate": "2017-01-02",           "tag": [             "java",             "elasticsearch"           ],           "tag_cnt": 2,           "view_cnt": 80,           "title": "this is java, elasticsearch, hadoop blog",           "content": "elasticsearch and hadoop are all very good solution, i am a beginner"         }       },

 

{ "match": { "title": "java solution" }}，针对doc4，是有一个分数的

{ "match": { "content":  "java solution" }}，针对doc4，也是有一个分数的

 

所以是两个分数加起来，比如说，1.1 + 1.2 = 2.3

matched query数量 = 2(匹配到的query的数量)

总query数量 = 2(搜索的总的query数量)

 

2.3 * 2 / 2 = 2.3

 

算一下doc5的分数

{         "_index": "forum",         "_type": "article",         "_id": "5",         "_score": 0.56008905,         "_source": {           "articleID": "DHJK-B-1395-#Ky5",           "userID": 3,           "hidden": false,           "postDate": "2017-03-01",           "tag": [             "elasticsearch"           ],           "tag_cnt": 1,           "view_cnt": 10,           "title": "this is spark blog",           "content": "spark is best big data solution based on scala ,an programming language similar to java"         }       },

 

{ "match": { "title": "java solution" }}，针对doc5，是没有分数的（因为没搜索到）

{ "match": { "content":  "java solution" }}，针对doc5，是有一个分数的

 

所以说，只有一个query是有分数的，比如2.3

matched query数量 = 1  (匹配到的query的数量)

总query数量 = 2(搜索的总的query数量)

 

2.3 * 1 / 2 = 1.15

 

doc5的分数 = 1.15 < doc4的分数 = 2.3

 

4、best fields策略，dis_max

 

best fields策略，就是说，搜索到的结果，应该是某一个field中匹配到了尽可能多的关键词，被排在前面(doc5)；而不是尽可能多的field匹配到了少数的关键词，排在了前面(doc4)

 

dis_max语法，直接取多个query中，分数最高的那一个query的分数即可

 

{ "match": { "title": "java solution" }}，针对doc4，是有一个分数的，1.1

{ "match": { "content":  "java solution" }}，针对doc4，也是有一个分数的，1.2

取最大分数，1.2

 

{ "match": { "title": "java solution" }}，针对doc5，是没有分数的

{ "match": { "content":  "java solution" }}，针对doc5，是有一个分数的，2.3

取最大分数，2.3

 

然后doc4的分数 = 1.2 < doc5的分数 = 2.3，所以doc5就可以排在更前面的地方，符合我们的需要

GET /forum/article/_search {     "query": {         "dis_max": {             "queries": [                 { "match": { "title": "java solution" }},                 { "match": { "content":  "java solution" }}             ]         }     } } 结果 {   "took": 19,   "timed_out": false,   "_shards": {     "total": 5,     "successful": 5,     "failed": 0   },   "hits": {     "total": 4,     "max_score": 0.68640786,     "hits": [       {         "_index": "forum",         "_type": "article",         "_id": "2",         "_score": 0.68640786,         "_source": {           "articleID": "KDKE-B-9947-#kL5",           "userID": 1,           "hidden": false,           "postDate": "2017-01-02",           "tag": [             "java"           ],           "tag_cnt": 1,           "view_cnt": 50,           "title": "this is java blog",           "content": "i think java is the best programming language"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "5",         "_score": 0.56008905,         "_source": {           "articleID": "DHJK-B-1395-#Ky5",           "userID": 3,           "hidden": false,           "postDate": "2017-03-01",           "tag": [             "elasticsearch"           ],           "tag_cnt": 1,           "view_cnt": 10,           "title": "this is spark blog",           "content": "spark is best big data solution based on scala ,an programming language similar to java"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "4",         "_score": 0.5565415,         "_source": {           "articleID": "QQPX-R-3956-#aD8",           "userID": 2,           "hidden": true,           "postDate": "2017-01-02",           "tag": [             "java",             "elasticsearch"           ],           "tag_cnt": 2,           "view_cnt": 80,           "title": "this is java, elasticsearch, hadoop blog",           "content": "elasticsearch and hadoop are all very good solution, i am a beginner"         }       },       {         "_index": "forum",         "_type": "article",         "_id": "1",         "_score": 0.26742277,         "_source": {           "articleID": "XHDK-A-1293-#fJ3",           "userID": 1,           "hidden": false,           "postDate": "2017-01-01",           "tag": [             "java",             "hadoop"           ],           "tag_cnt": 2,           "view_cnt": 30,           "title": "this is java and elasticsearch blog",           "content": "i like to write best elasticsearch article"         }       }     ]   } }