Lapis uses word endings to determine if a word is kifuku or not, e.g. if it ends on ["い", "う", "く", "す", "つ", "ぶ", "む", "る"] then it's set as kifuku (except if it's heiban). JPMN on the other hand assigns a word as kifuku by utilizing the Yomitan {tags} data, and assigns a word as kifuku if it contains a verb tag (except if it's heiban).

How accurate are each of these approaches? Are there cases when they are wrong?

Thanks!