why database and manticoresearch return different result

[l1t1]

ms returns 13 rows. db returns 18 rows.
note field at row of code 0122 includes 油料 in ms.

mysql> select code from note5 where match('油料|薯类');
+------+
| code |
+------+
| 3532 |
| 5111 |
| 0123 |
| 1371 |
| 1419 |
| 1519 |
| 3579 |
| 3572 |
| 0511 |
| 1511 |
| 5121 |
| 5112 |
| 0252 |
+------+
select column1 from note5 where note like'%油料%' or note like'%薯类%';
┌─────────┐
│ column1 │
│ varchar │
├─────────┤
│ 0122    │
│ 0511    │
│ 1331    │
│ 1332    │
│ 3463    │
│ 3532    │
│ 5111    │
│ 5119    │
│ 5639    │
│ 0123    │
│ 1371    │
│ 1419    │
│ 1511    │
│ 1519    │
│ 3572    │
│ 3579    │
│ 5121    │
│ 5112    │
├─────────┤
│ 18 rows │
└─────────┘

select * from note5 where code='0122';
|    6 | 0122 | 油料种植,包括对下列油料的种植活动：

[l1t1]

my config file

source csv_note5
{
    type = csvpipe
    csvpipe_command = type note5.csv
    
    csvpipe_field_string = code
    csvpipe_field_string = note
}
index note5 {
  charset_table = cjk
  morphology = icu_chinese
  type   = plain
  source = csv_note5
  path   = C:\Manticore\bin\note5
}

[tomatolog]

it could be better to note5.csv file or create reproducible example that could be investigated here locally

[l1t1]

note5.csv
created by
copy(select row_number()over()rn,* from note5 where note like'%油料%' or note like'%薯类%') to 'd:/note5.csv';

[l1t1]

any finding?

[tomatolog]

did not have time to check this case. Going to check this case next week

[l1t1]

have time to check it?

[l1t1]

test on version 6.2, it has the same question.

[l1t1]

and following line found by manticoresearch doesn't match ('油料|薯类')

38,0252,"非木竹材林产品采集,包括对下列非木竹材林产品的采集活动：,天然橡胶：天然橡胶乳、烟胶片、胶清片、白皱片、褐胶片、标准胶片、其他天然橡胶；,天然树>
脂、树胶：天然生漆、天然松脂、虫胶、桃胶、冷杉胶、其他天然树脂、树胶；,栲胶原料：落叶松树皮、杨梅树皮、油柑树皮、檞树皮、木麻黄树皮、黑荆树皮、橡碗、化香果、其他栲胶原料；,非直接食用果类：油桐籽、油茶籽、沙棘果、乌桕子、文冠果、山苍籽、黑椋子、麻风树果（小桐子）、黄连木果、光皮梾（光皮树）果、其他非直
接食用果类；,非木竹编结用原料：灯心草、菖蒲、葵叶、其他编织用原料；,染色、鞣革用植物原料：五倍子、地衣、蓝靛、薯茛、其他染色、鞣革用植物原料；,非木竹野
生植物活体：野生蕨菜、野生发菜、野生薇菜、野生锁阳、其他野生植物活体；,野生植物采集产品：野生植物根、野生植物茎、野生植物叶、野生植物花、野生植物果实、
其他野生植物采集产品；,其他林产品：未加工天然软木、棕片、竹笋干、黄柏桕、山苍子、桉树叶、其他未列明林产品。,不包括：,咖啡、可可等饮料作物的采集，列入01
69（其他饮料作物种植）。"

[Maxwell0010]

dfgy

[sanikolaev]

@l1t1 If you do this:

mysql> set profiling=1; select code from note5 where match('油料|薯类'); show plan;
Query OK, 0 rows affected (0.00 sec)

+------+
| code |
+------+
| 5111 |
| 3532 |
| 0511 |
| 1511 |
| 5121 |
| 5112 |
| 3572 |
| 1419 |
| 1519 |
| 3579 |
| 1371 |
| 0123 |
+------+
12 rows in set (0.00 sec)

+------------------+-------------------------------------------------------------------------------------------------------------------------+
| Variable         | Value                                                                                                                   |
+------------------+-------------------------------------------------------------------------------------------------------------------------+
| transformed_tree | AND(
  OR(
    AND(KEYWORD(油料, querypos=1)),
    AND(KEYWORD(薯, querypos=2))),
  AND(KEYWORD(类, querypos=3)))     |
+------------------+-------------------------------------------------------------------------------------------------------------------------+
1 row in set (0.00 sec)

you can see how your search string is tokenized: it searches for texts containing 类 and 油料 OR 薯. That's why the following doesn't find anything:

mysql> select code from note5 where match('油料|薯类') and code='0122';
Empty set (0.00 sec)

but if you do this, you'll get the results:

mysql> select code from note5 where match('(油料)|(薯类)') and code='0122';
+------+
| code |
+------+
| 0122 |
+------+
1 row in set (0.00 sec)

It also finds more documents in general:

mysql> select code from note5 where match('(油料)|(薯类)');
+------+
| code |
+------+
| 5111 |
| 3532 |
| 0511 |
| 1511 |
| 5121 |
| 5112 |
| 3572 |
| 1419 |
| 1519 |
| 3579 |
| 1371 |
| 0123 |
| 0122 |
| 3463 |
| 5639 |
+------+
15 rows in set (0.00 sec)