The bibliographic data comprising the identification information for each book are sometimes not unified across libraries in Japan. In order to improve this situation, we investigate the current bibliographic confusion in Japan and to manually iden-tify similar books using crowdsourcing versus automatic identification. Sets of books which that were previously difficult to judge can now be processed via bibliographic identification through crowdsourcing. In particular, it is noteworthy that the crowdsourcing method was correct when the sets of comprised nouns with high specificity or low frequency of use. Among the sets with divided re-sults, 14 were caused by formatting differences between the two books. It would be possible to raise the rate of decision accuracy further by adding information about size that was not displayed on the screen in this investigation. However, as there are many differences in terms of attention and knowledge among people who participate in crowdsourcing, it is necessary to use large numbers of partici-pants in order to more successfully identify sets of books.