In order to prevent delivery delays in software development projects, a project manager often predicts the date when project will be completed. However, incorrect task information makes it harder to predict the milestone accurately. To solve this problem, we propose a method to automatically correct task information. Our approach guesses the correct task information by linking the task information to commit history data. We applied our approach to industrial embedded software development project data and confirmed that the error rate of the predicted project completion time improved from 28.6% to 8.6%, which satisfies a practical use level.

In open source software (OSS) development, not only coding but also various contributions such as developer assignment, debugging, function extension, code review, and question answering are required. In this paper, we define two metrics to distinguish contribution types. Through an analysis of many OSS projects using the proposed metrics, we identify 4 basic contribution types. Afterward, by using archetypal analysis to developers in each contribution type, this paper finally identifies 7 contribution types. These contribution types are useful in identifying required personnel in an individual OSS project.

In web application development, GUI testing is performed to verify that the screen elements such as buttons and links that appear on the screen are working correctly. In this testing, GUI testing automation tools which automatically operate the browser according to test scripts are widely used. GUI testing automation tools use locators to identify screen elements for testing, but if the software under test is modified or changed, existing locators may not be able to identify screen elements to operate. In such cases, locators in test scripts must be repaired so as to refer screen elements correctly. However, the repair work is not simple and is expensive. In this paper, we propose a method to repair locators based on three similarity indices based on Screen Element Sentences by distributed representation, numerical characters appeared in the Screen Element Sentences and XPath. We evaluate the method to show that our method can get candidates of locators at the same accuracy as the previous work that uses 19 indices.

Several techniques have been available in research studies on automated program repair. In test-based repair techniques among them, the preparation of test cases is key to the success of repair. Nevertheless, there is no sufficient knowledge based on the investigation of such test cases and programs to be fixed so far. We designed an experiment framework for investigation of how the characteristics of prepared test cases affect the number of generated patches, their correctness, and the consuming time of their generation. This paper shows results of an experiment with programs containing five kinds of well-known bugs. The results demonstrate that either the increasing or decreasing of the test cases improves the performance of patch generation, depending on the kinds of bugs.

Source code changes are essential for software evolution. However, a source code change also has a risk of introducing a new bug into the software. To predict such a bug by using features of the code change, researchers have studied various bug prediction models in the past. One of the most noteworthy methods is the personalized bug prediction, which builds a prediction model customized for each developer. Although the personalized bug prediction model is a promising one, it also has a challenge that we cannot make appropriate models for less experienced developers due to the lack of code change (commit) data. Toward a resolution of the challenge, this paper proposes a method for combining two or more personalized bug prediction models. The usefulness of the proposed method is proved by an empirical study with the data from five open-source software development projects.

One of the drawbacks of the touch input is limited input vocabulary. To address this problem, we designed a new touch gesture: bounded swipe. Bounded swipe is a swipe gesture that has start and end points inside a target object. We conducted three experiments to test the feasibility of this gesture. The experimental results showed that, when the user swiped normally on a target object, the probability of the accidental occurrence of a bounded swipe was 0.8%. Moreover, the accuracy of bounded swipe was 96.7%. We also investigated the accuracy when distinguishing between a tap and a bounded swipe and found that the two gestures could be distinguished accurately 97.4% of the time.

Research was conducted regarding a display that presents digital information using bubbles. Conventional bubble displays require moving parts, because it is common to use air taken from outside of the water to represent pixels. However, it is difficult to increase the number of pixels at a low cost. We propose a liquid-surface display using pixels of bubble clusters generated from electrolysis, and present the cup-type device BubBowl, which generates a 10×10 pixel dot matrix pattern on the surface of a beverage. Our technique requires neither a gas supply from the outside nor moving parts. Using the proposed electrolysis method, a higher-resolution display can easily be realized using a PCB with a higher density of matrix electrodes. Moreover, the method is simple and practical, and can be utilized in daily life, such as for presenting information using bubbles on the surface of coffee in a cup.