This paper proposed \textbf{Comic-Shelf (CS) vectors}, which convolve the co-occurrence of comic titles on the bookshelves ordered by ranking, as a method for modeling sensibilities toward comic titles. By extracting semantic relationships from the orderings based on readers’ subjective evaluations and representing them as numerical vectors, we aim to establish a new information representation that reflects user sensibilities. In vector mapping analysis, it was revealed that the comic vectors of titles stored on the same bookshelf were plotted relatively close to one another. Assuming that the affection toward titles included on the same bookshelf is similar, it was inferred that higher vector similarity corresponds to comics that are closer in human affection. Furthermore, it was demonstrated that not only similarities between individual titles but also similarities between bookshelf themes could be visually captured. In a mock recommendation, we investigated whether CS vectors could select titles that aligned with participants’ preferences. The results showed that using CS vectors allowed for the selection of comics that better aligned with participants’ preferences compared to other methods, demonstrating the effectiveness of the CS vectors.