I was updating many smith records recently and got fed up with the amount of time needed to locate a record for a particular smith (by name, era, province, school etc). When trying to find one smith I always end up with filling out missing Kanji, signatures and whatever else is missing for few other smiths with the same name.
It is very distracting, so I've decided to solve this problem (at least to some extent) by stopping for the moment and filling all the missing Kanji in formal smith names. It's not a small job: 2156 smiths (out of 12258 record in total) don't have Kanji representation of their name in Swordsmith Index at the moment. It's clear from the past experience that doing 100 record per day is already a very challenging task. So I've decided I'll do my best to finish this work by Christmas.
Update
Kanji Marathon is finished!
All the smiths have Kanji representation of the name now.
In the meantime I've created a little automated solution (more details to follow) which verifies Kanji agains Romaji. It has already found 311 inconsistencies (some of them are legitimate though). My next assignment would be to go through the list of inconsistences and fix them.
Most of them are straightforward, e.g. using the reading 'ie' instead of 'iye' (which has been decided to be THE reading used in the Index), long vowels which I don't check yet (e.g. uu), hiragana and katakana in names (I only check against Kanji). There are also few cases when similar looking characters are mixed up or old and new forms used inconsistently, e.g. 舍 and 舎, 藏 and 蔵.
Obviously it can't tell if a wrong Kanji used against correct Romaji reading. But that's something which can only be checked manually.