[Koha] The origin of "Record Matching Rules" ?

[zolot]

Hi !

The so-called  "record matching rule"  determines the method of comparison  of  "incoming record"  and  "existing record"  when importing MARC record into Koha. 
(See "Koha 3.14 Manual (en)", chapter "2. Administration", section "4.7. Record Matching Rules".)

But what does "to compare an incoming record against an existing record" mean? Are structure or content of the different subfields of these records compared ?

What is it being made for ?

In general: 
what problem arise during import process and is being solved with the help of this rule ?

These questions are straight related to work in Koha.

For example, what threshold value is better: 100 or 100000 ?  Why? What does a threshold depend on? 
Et cetera.

Thank you.

Wishes, 
Gennady