In enterprises operating large-scale equipment, such as plants, maintenance workers must quickly and accurately find and understand the information in the equipment maintenance documents to perform maintenance tasks effectively. If the equipment maintenance documents include sentences with semantically ambiguous expressions, it will interfere with the maintenance knowledge search, and it may affect the maintenance performance of engineers. In order to solve these problems, text-based research of maintenance documents have been done to extract the key information or knowledge from these documents. Previous studies focused on finding the technical terminologies or calculating the similarity of documents using named entity recognition approaches. This paper proposes a method to extract knowledge of not only the technical terminologies but also their relations. The proposed method uses a rule-based approach that can be applied to the results of a named entity recognition approach and a dependency parsing approach. The named entity recognition approach found technical terms and the dependency parsing approach provided sentence structure information, so that the proposed method showed that a set of rules can extract maintenance knowledge, including entities and their relations. Trouble-shooting documents in the field were used as an experiment to demonstrate the effectiveness of the proposed method, and the experiment showed the possibility of practical use of the proposed method.

In this study, we propose a valuation methodology for Non-Fungible Tokens (NFTs), focusing on the profile picture (PFP) NFT category represented by the Bored Ape Yacht Club (BAYC). To identify the attributes that influence the value of individual BAYC NFTs, we develop a hedonic pricing model that uses the NFT’s value as the dependent variable and its properties as independent variables. We apply Term Frequency-Inverse Document Frequency (TF-IDF) to quantify attributes of NFTs. Three hedonic models—linear, quadratic, and full quadratic—are proposed. For the full quadratic model, we introduce a systematic procedure to select first-order, second-order, and interaction terms in the model. To evaluate the performance of the proposed models, we carried out comparative computational experiments. We collected actual BAYC transaction data and split it into a training set (70%) and a validation set (30%). For benchmarking purposes, we compare the proposed models against four machine learning algorithms: Random Forest, Support Vector Regression (SVR), XGBoost, and LightGBM. The machine learning models perform well on the training set, however, this was largely due to overfitting. In contrast, the proposed hedonic models maintained consistent performance with minimal degradation from the training to the validation set. Among them, the full quadratic model demonstrates the highest explanatory power on the validation set in terms of adjusted R² and other evaluation metrics.