This paper introduces Silverchain, a fluent API generator developed by the author of this paper. Silverchain accepts rules on the chaining order of API methods and generates type definitions for the fluent API where the Java type checker can validate the chaining order. The output definitions of Silverchain are a skeleton of the fluent API. The creator of the fluent API (the user of Silverchain) can complete the fluent API by adding some code such as actions performed in the API methods. This paper particularly describes the design efforts made to the input and output of Silverchain, as well as the efforts made to promote Silverchain.

Demands on the performance of database systems continue to increase. In state-of-the-art database systems, the storage engine is a major source of performance bottlenecks, and it is important to harness parallelism by avoiding mutual exlusion mechanisms such as locks. In this paper, we outline the design of LineairDB, a transactional storage engine open-sourced by the authors. LineairDB employs epoch-based synchronization to all components, which harnesses a high degree of multicore parallelism in handling data structures. We describe how each component of LineairDB is optimized by epoch-based synchronization. Our experimental results show that the optimizations in LineairDB lead to about a 75x increase in throughput in YCSB benchmark. In addition, we implement LineairDB-storage-engine, which is a plugin for MySQL. It achieved up to 5x better throughput compared to InnoDB, which is MySQL's default storage engine. On the other hand, the approach of epoch-based synchronization affects both the performance characteristics and the consistency that the database provides to users. This paper also discusses these aspects.

This paper proposes a methodology for extracting MVP from User-Centered Requirements Engineering using goal-oriented requirements analysis by extending the Finnish method Ajatus Kartta. This methodology applies the kartta to goal-oriented requirements analysis that can clarify functional and non-functional requirements that contribute to goals. Even beginners of User-Centered Requirements Engineering can easily extract MVPs from the user's perspective by this. This paper discusses the procedure of this method and examines the reproducibility of this method and effectiveness of MVP extraction.

This paper gives an algorithm that solves the bi-abduction problem in symbolic-heap separation logic with arrays and lists. The logic is an assertion language of Hoare-style logic for program verification of pointer manipulating programs. The bi-abduction problem asks to find an additional assumption and an additional conclusion from a given assumption and a given conclusion such that the entailment becomes true. Bi-Abduction is indispensable for modular analysis and automatic verification with separation logic, since the condition at a call site that calls a function and the precondition of the called function are analyzed separately by modular analysis, and they both may contain spatial formulas of separation logic, and we have to guarantee they become the same by adding some spartial formulas to each of them. This paper shows the correctness of the bi-abduction algorithm with detailed proofs. A bi-abduction solver based on the algorithm has been implemented as a part of the authors' automatic program verifier, and experimental results of the bi-abduction solver with small inputs are also shown, which show the algorithm is usable.

Regular expression is widely known as a powerful and general-purpose text processing tool for programming. Though the regular expression is highly versatile, there are various difficulties in using them. One promising approach to reduce the burden of the pattern composition is reuse by referring to past usages. Still, several source code-specialized search engines have been proposed, they are not suitable for the scenario of reusing regular expression patterns. The purpose of this study is the efficient reuse of regular expression patterns. To achieve the purpose, we propose a usage retrieval system Resem specialized in regular expression patterns. Resem adopts two key features: search by semantics and collecting input/output examples. As an evaluation result, we confirm that Resem decreased the time required for describing patterns by 16%.

With the growth of the smartphone game market, many cases have been reported in which children have purchased charged items in games without restraint and have been charged large sums of money. In addition, there are users called "heavy payers" or "wasteful payers," who pay much more than the normal amount of money, and there are also reports of cases in which these users are disturbed in their daily lives. In this study, we actually play several representative smartphone games and analyze how developers induce users to pay for their games. As a result of extracting and tabulating the billing elements for each game, we found that many games employ a rich gacha system that makes consumers dependent on the game, similar to gambling, and that they also employ a number of game management strategies to make the amount charged look profitable and to stimulate user motivation to pay.

In autonomous mobile robots, localization is a fundamental process for realizing more advanced functions, and its accuracy greatly defines the range of possible applications. Typical localization is performed by integrating odometry from internal sensors and environmental information from external sensors. However, in some educational settings, expensive external sensors such as LiDAR may not be available, mainly due to cost issues. In this case, localization is performed using only odometry obtained from inexpensive internal sensors such as encoders of wheel drive motors or IMUs, but the error is large, limiting available experiments.
This paper proposes a method to improve the accuracy of localization using only wheel odometry information by using a recurrent neural network, mainly for the purpose of supporting the practice of autonomous mobility technology in an educational setting. The evaluation results show that the proposed method can reduce the orientation error by an average of about 82% and a maximum of about 90%, and the position error by an average of about 48% and a maximum of about 72%, and that it can be an effective solution to the problems of cumulative error and accuracy degradation during turning.

Sensing how a user holds a smartphone enables adaptive user interfaces, such as those that automatically change the displayed content and position of graphical user interface (GUI) components based on how the phone is held. We propose ReflecTouch, a novel method to detect how a smartphone is being held by capturing images of the smartphone screen reflected on the cornea with a built-in front camera. In these images, the areas where the user places their fingers on the screen appear as shadows, which makes it possible to estimate the grasp posture. Since most smartphones have a front camera, this method can be used regardless of the device model; in addition, no additional sensor or hardware is required. In this paper, we report the identification accuracy of the proposed method through laboratory experiments under controlled conditions and crowdsourcing experiments that are closer to the actual situations.