Users' FAQ
"Very High", "High", "Normal", and "Low" are word usage frequency. This frequency is based on many sources such as newspapers, novels, blogs, etc.
A word marked as "Unranked" means that it's not counted in our database. This does not always mean that it is not used often. For example, "そうかもしれない" is a common phrase, but it's a combination of multiple words so it's not counted in the database.
These are readings only apply to Kanji next to them. For example, please try searching 顔. You will see "がん (顔)" under "Restricted readings". This means that the reading "がん" only applies to "顔" and does not apply to "貌".
For common sentences, the accuracy is very good. However, unusual sentences or grammatically questionable ones such as "きょねんはさむかったですね" can give inaccurate results. However, you can press on the green triangle button on the right to re-analyze those sentences. In the future, we plan to add more NLP engines and allow users to change them.
It's best to explain this by using example sentences such as "娘の生まれは東京だ。" and "娘が生まれ、妻の実家から帰った後のエピソードとなります。" If you try searching for "生まれ" of the first sentence, the app will only show "生まれ". However, for the second sentence, the app will return "生まれる" first because it detects this word as a verb. In case the detection is wrong, "生まれ" is also shown as the second result.
Sometimes all readings in the definition of a word are totally different with its reading shown in Katakana. In this case, the Katakana reading might be wrong and you should use the readings in the definition or check with other sources. This is totally normal as Japanese is a highly contextual language. In addition, the app uses many algorithms to guess the correct meaning of a word in a sentence. Thus, sometimes results return from these algorithms can conflict with each other.
Please try to capture more text, as this provides OCR engine more patterns to work on. For example, instead of capturing one or two words, please try to capture the whole sentence.