1 Reply ・ Started by n3m1 at 2023-01-06 22:32:30 UTC ・ Last reply by Kimtaro Admin at 2023-01-11 05:58:49 UTC

What is the logic of ranking results

Been wondering what is the lotic for result ranking. Let's say for example I am searching for 頭. To no surprise the first will bring up the reading あたま and meaning head.

The readings and the results seem to match the data in JmDict for the most part so I am assuming that isbthe primary source of data for reading and sense.

However the data has a number of kanji elements with an exact match in different entries. However even looking at the priority field for the kanji elements, it is not straight forward why the first results is the one I get.

Not that I am not looking for an "intuitive" answer why, more and algorithmic logic

Kimtaro Admin at 2023-01-11 05:58:49 UTC

The ranking is a mix of several indicators, like frequency data from JMdict, kanji frequency, whether the word has example sentences, part of speech, and more. It's hand tuned with different importance given to different indicators.

to reply.