RESUMEN
In this data article, we present an FAQ dataset written in Japanese and its translation to English in order to train chatbot models for e-learning systems. We first collected raw Q&A data reported as the difficulties from April 2015 to July 2018 by users of the e-learning system introduced at Tokyo Metropolitan University. We then divided them into 11 categories according to features provided by the e-learning system. Finally, we integrated questions with the same answers in order to create the FAQ form. The dataset contains 427 questions and 79 answers that were examined by experts with experience in using the e-learning system for more than three years. Using this dataset, we performed statistical analyses to evaluate the qualities of the FAQ dataset. The proposed applications of the dataset include not only academic research but also activities; for example, translating from Japanese to another one like Chinese, adapting/modifying our dataset for another e-learning system, and developing language models to obtain highly accurate responses from chatbots.
RESUMEN
In conventional image trading systems, images are usually stored unprotected on a server, rendering them vulnerable to untrusted server providers and malicious intruders. This paper proposes a conceptual image trading framework that enables secure storage and retrieval over Internet services. The process involves three parties: an image publisher, a server provider, and an image buyer. The aim is to facilitate secure storage and retrieval of original images for commercial transactions, while preventing untrusted server providers and unauthorized users from gaining access to true contents. The framework exploits the Discrete Cosine Transform (DCT) coefficients and the moment invariants of images. Original images are visually protected in the DCT domain, and stored on a repository server. Small representation of the original images, called thumbnails, are generated and made publicly accessible for browsing. When a buyer is interested in a thumbnail, he/she sends a query to retrieve the visually protected image. The thumbnails and protected images are matched using the DC component of the DCT coefficients and the moment invariant feature. After the matching process, the server returns the corresponding protected image to the buyer. However, the image remains visually protected unless a key is granted. Our target application is the online market, where publishers sell their stock images over the Internet using public cloud servers.