基于term vector深入探查数据的情况
更多干货
- 分布式实战(干货)
- spring cloud 实战(干货)
- mybatis 实战(干货)
- spring boot 实战(干货)
- React 入门实战(干货)
- 构建中小型互联网企业架构(干货)
- python 学习持续更新
- ElasticSearch 笔记
概述
1、term vector介绍
获取document中的某个field内的各个term的统计信息
term information: term frequency in the field, term positions, start and end offsets, term payloads
term statistics: 设置term_statistics=true; total term frequency, 一个term在所有document中出现的频率; document frequency,有多少document包含这个term
field statistics: document count,有多少document包含这个field; sum of document frequency,一个field中所有term的df之和; sum of total term frequency,一个field中的所有term的tf之和
GET /twitter/tweet/1/_termvectors
GET /twitter/tweet/1/_termvectors?fields=text
term statistics和field statistics并不精准,不会被考虑有的doc可能被删除了
我告诉大家,其实很少用,用的时候,一般来说,就是你需要对一些数据做探查的时候。比如说,你想要看到某个term,某个词条,大话西游,这个词条,在多少个document中出现了。或者说某个field,film_desc,电影的说明信息,有多少个doc包含了这个说明信息。
2、index-iime term vector实验
term vector,涉及了很多的term和field相关的统计信息,有两种方式可以采集到这个统计信息
(1)index-time,你在mapping里配置一下,然后建立索引的时候,就直接给你生成这些term和field的统计信息了 (2)query-time,你之前没有生成过任何的Term vector信息,然后在查看term vector的时候,直接就可以看到了,会on the fly,现场计算出各种统计信息,然后返回给你
PUT /my_index
{
"mappings": {
"my_type": {
"properties": {
"text": {
"type": "text",
"term_vector": "with_positions_offsets_payloads",
"store" : true,
"analyzer" : "fulltext_analyzer"
},
"fullname": {
"type": "text",
"analyzer" : "fulltext_analyzer"
}
}
}
},
"settings" : {
"index" : {
"number_of_shards" : 1,
"number_of_replicas" : 0
},
"analysis": {
"analyzer": {
"fulltext_analyzer": {
"type": "custom",
"tokenizer": "whitespace",
"filter": [
"lowercase",
"type_as_payload"
]
}
}
}
}
}
PUT /my_index/my_type/1
{
"fullname" : "Leo Li",
"text" : "hello test test test "
}
PUT /my_index/my_type/2
{
"fullname" : "Leo Li",
"text" : "other hello test ..."
}
GET /my_index/my_type/1/_termvectors
{
"fields" : ["text"],
"offsets" : true,
"payloads" : true,
"positions" : true,
"term_statistics" : true,
"field_statistics" : true
}
{
"_index": "my_index",
"_type": "my_type",
"_id": "1",
"_version": 1,
"found": true,
"took": 10,
"term_vectors": {
"text": {
"field_statistics": {
"sum_doc_freq": 6,
"doc_count": 2,
"sum_ttf": 8
},
"terms": {
"hello": {
"doc_freq": 2,
"ttf": 2,
"term_freq": 1,
"tokens": [
{
"position": 0,
"start_offset": 0,
"end_offset": 5,
"payload": "d29yZA=="
}
]
},
"test": {
"doc_freq": 2,
"ttf": 4,
"term_freq": 3,
"tokens": [
{
"position": 1,
"start_offset": 6,
"end_offset": 10,
"payload": "d29yZA=="
},
{
"position": 2,
"start_offset": 11,
"end_offset": 15,
"payload": "d29yZA=="
},
{
"position": 3,
"start_offset": 16,
"end_offset": 20,
"payload": "d29yZA=="
}
]
}
}
}
}
}
3、query-time term vector实验
GET /my_index/my_type/1/_termvectors
{
"fields" : ["fullname"],
"offsets" : true,
"positions" : true,
"term_statistics" : true,
"field_statistics" : true
}
一般来说,如果条件允许,你就用query time的term vector就可以了,你要探查什么数据,现场去探查一下就好了
4、手动指定doc的term vector
GET /my_index/my_type/_termvectors
{
"doc" : {
"fullname" : "Leo Li",
"text" : "hello test test test"
},
"fields" : ["text"],
"offsets" : true,
"payloads" : true,
"positions" : true,
"term_statistics" : true,
"field_statistics" : true
}
手动指定一个doc,实际上不是要指定doc,而是要指定你想要安插的词条,hello test,那么就可以放在一个field中
将这些term分词,然后对每个term,都去计算它在现有的所有doc中的一些统计信息
这个挺有用的,可以让你手动指定要探查的term的数据情况,你就可以指定探查“大话西游”这个词条的统计信息
5、手动指定analyzer来生成term vector
GET /my_index/my_type/_termvectors
{
"doc" : {
"fullname" : "Leo Li",
"text" : "hello test test test"
},
"fields" : ["text"],
"offsets" : true,
"payloads" : true,
"positions" : true,
"term_statistics" : true,
"field_statistics" : true,
"per_field_analyzer" : {
"text": "standard"
}
}
6、terms filter
GET /my_index/my_type/_termvectors
{
"doc" : {
"fullname" : "Leo Li",
"text" : "hello test test test"
},
"fields" : ["text"],
"offsets" : true,
"payloads" : true,
"positions" : true,
"term_statistics" : true,
"field_statistics" : true,
"filter" : {
"max_num_terms" : 3,
"min_term_freq" : 1,
"min_doc_freq" : 1
}
}
这个就是说,根据term统计信息,过滤出你想要看到的term vector统计结果
也挺有用的,比如你探查数据把,可以过滤掉一些出现频率过低的term,就不考虑了
7、multi term vector
GET _mtermvectors
{
"docs": [
{
"_index": "my_index",
"_type": "my_type",
"_id": "2",
"term_statistics": true
},
{
"_index": "my_index",
"_type": "my_type",
"_id": "1",
"fields": [
"text"
]
}
]
}
GET /my_index/_mtermvectors
{
"docs": [
{
"_type": "test",
"_id": "2",
"fields": [
"text"
],
"term_statistics": true
},
{
"_type": "test",
"_id": "1"
}
]
}
GET /my_index/my_type/_mtermvectors
{
"docs": [
{
"_id": "2",
"fields": [
"text"
],
"term_statistics": true
},
{
"_id": "1"
}
]
}
GET /_mtermvectors
{
"docs": [
{
"_index": "my_index",
"_type": "my_type",
"doc" : {
"fullname" : "Leo Li",
"text" : "hello test test test"
}
},
{
"_index": "my_index",
"_type": "my_type",
"doc" : {
"fullname" : "Leo Li",
"text" : "other hello test ..."
}
}
]
}