数据量40亿左右，采用ngram=1 分词，查询太慢

Feature Requests
1

数据量40亿左右，采用ngram=1 分词，match命中keyword多，查询太慢，查询耗时几分钟

2

Did you try morphology=icu_chinese?

3

icu_chinese不符合我们的场景，会有些查询查不到啊

4

使用 set profiling=1；发现get_docs Durition 150.01多

5

icu_chinese不符合我们的场景，会有些查询查不到啊

There’s an issue about adding support for Jieba ICU is not a good choice for chinese · Issue #931 · manticoresoftware/manticoresearch · GitHub

Pull requests are welcome. If it’s mission critical for you and you are ready to sponsor the development, the core team can prioritize this task. Let me know then.

6

那关于使用 set profiling=1；调试发现get_docs Durition 150.01多，这是什么原因导致耗时这么长呢

7

It may be just too complex computation, because each character becomes a separate token in case of ngram_len=1, Manticore has to find all documents for each token and then intersect/combin the lists, sort them etc.

8

thank you very munch,i understand

9

嗨，能联系到你么？有些问题想请教

我有邮箱：dingtianbao1983@gmail.com

10

I’m not sure who you meant: @xyls82 or me, but you can contact me at contact@manticoresearch.com