How to count non-matching values ​​of 2 columns?

Question

Hadoop

It is necessary to calculate non-matching values ​​of 2 columns in a table in Hive without regard to case and symbols and, preferably, taking into account transliteration. That is, at the output, I would like options like L'OREAL - Loreal, MARS - Mars, PARMALAT - Parmalat to be considered as matching, and options like L'OREAL - Parmalat as not matching.
The query below, which is enough for me, of course, considers them all as not matching. Tried a hundred different additions to it, it's not the same.

select count (item_id)
from table_1
where column_1 <> table_2