Criminal Deduction Using Similarity Analysis Between Mystery Stories

The " Knowledge Graph Reasoning Challenge" an annual event in which participants try to infer a mystery novel using Knowledge Graph and other tools, has been attracting a lot of attention. The purpose of this research is to propose and implement a new method using BERT for this challenge. To prove this, we focused on the degree of similarity between the two novels and estimated the culprit. As a result of the analysis focusing on the words and actions of the culprit, we succeeded in extracting the culprit's name. However, the successful extraction was based on a specific condition, and further discussion is needed to pursue more precise accuracy in the future.