This article is prepared for submission to Competition on Legal Information Extraction/Entailment (COLIEE 2022), an international competitive event organized to focus on information processing and retrieval. The proposed method tackles on how to construct an answering system capable of responding Yes/No legal questions, ultimately recognizing entailment between legal queries from past Japanese bar exams and relevant articles of Japan Civil Code (both in Japanese). We first attempted to extract disjunctive union text from each training query and relevant article(s) with corresponding ‘Y/N’ answers as their labels, eventually forming our reference database (training set). Then the same process was repeated on a sample of different queries and relevant articles yet without a ‘Y/N’ label as the input (testing set). Finally, when constructing our model, the similarity ratio between the test disjunctive union and the training disjunctive union by longest common subsequence was calculated as its basis. As a result, this model achieved an accuracy of 0.6055 in Task 4 (rank 3rd as a team, and 7th as a trial). This is an extremely simple and efficient model capable of satisfactory performance.