Ryota Nakamura

我々は受講生のプログラミング演習時におけるコーディング過程を記録し，可視化して講師に提示するシステムC3PVを提案する．本システムは，ウェブ上で動作するオンラインエディタとコーディング過程ビューから構成されている．オンラインエディタは受講生のコーディングプロセスにおける，文字入力，コンパイル，実行，提出といったすべての行動を記録する．コーディング過程ビューは課題の進み具合いや受講者の相対的な進捗遅れを可視化して講師に提示する．講師はあるエラーに関して長時間悩んでいる受講生や全体の進捗と比較して遅れている受講生をC3PVによって確認し，個別指導といった支援につなげることができる．本研究では実際にC3PVを学部1年生が受講するJavaプログラミング演習に適用し，C3PVによって可視化されたコーディング過程を利用した受講生対応を行った．その結果，コーディング過程ビューに基づいて対応した45件中38件（約84%）において，実際に受講生がサポートを必要としていたことが確認できた．In this paper, we propose a coding process visualizer for programming practice. Our system named "C3PV" consists of an online editor and a coding process viewer. The online editor can collect all coding process performed by each student. The coding process viewer can visualize whether the students are doing well about their programming exercises. This viewer enables a lecturer and teaching assistants to identify the students who are falling behind the rest of the class. We confirmed that the system could determine whether the student needed our help with 84 percent accuracy by some practical experiments.

視線測定器の性能および操作性が向上したことにともない，Webユーザビリティ評価に視線情報が用いられるようになってきた．Webページ上のユーザが注視した箇所や軌跡を分析することで，Webページの視認性や利便性を評価できるようになった．しかしながら従来の視線分析作業では，注視箇所を手作業でコード化ならびにタグ付けする必要があるために多くの労力と時間を必要としている．そこで本研究では，視線検出とWebページの構造解析を組み合わせた手法を考案し，視線の軌跡上にあるテキストや画像などのオブジェクトを自動的に取得する機能と，注視されたオブジェクトを自動的に強調表示する機能を有するツールを開発した．評価実験の結果，提案手法は従来手法よりも効率的に注視箇所を特定することが可能であることを確認した．Gaze data is now used for the Web usability evaluation thanks to the improvement of gazes measurement devices. A major problem when using eye movements to evaluate Web usability issues is that eye movement data is too large to analyze, and potentially difficult to interpret meaningfully. This paper presents a novel Web usability evaluation tool which makes it possible to get a gazed object such as a text or image. The tool includes a visualization function that marks up gazed objects. The paper presents the results of user experiments which compare our visualization technique with traditional methods and show our tool is better than existing Web evaluation tools.

Due to the large number of students, lectures in the auditorium are likely to be limited to delivering one-sided lecture and conducting final exam. However, it is apparently desirable for students to have opportunities to check their own progress or understanding level through frequent check-up. Some have tried to build information systems for online-test or taking attendance. However, paper could not be replaced by online system due to the inevitable trouble of students' laptop PC or computer network. For this reason, we developed an educational system where the merit of paper and electronic media were merged. The system automatically recognizes student's ID and check marks written in paper. The scanner reads each student's report annotated by the lecturer and creates PDF, and sends it back to each student via e-mail. For the personal information protection, the created PDF is entirely or partially encrypted through the public-key cryptography.

Recently, e-learning contents that combine the speaker video with supporting slides have been produced in educational institutions such as universities. They enabled people to study at any time and any place. However, there is a problem that those systems make learners become tired because produced contents are monotonous. We propose the system "MONTA" that can automatically conceal important phrases to attract attention. MONTA can automatically cover important phrases in slide, and can remove the label in synchronization with utterance of the speaker.

Recently, the digital video camcorder has become popular, so we have gotten an opportunity to videotape a daily life and we have showed our videos on the home page of a web site or the video-blogging. However it has difficult to edit videos for amateurs. Consequently, they have expectations of automatic video editing system. In this paper, we constructed an automatic video editing system using gaze. We related an eye-movement while we watch videos to camera works such as panning or zooming so that our System can add camera works to amateur's videos effectively. As a result of comparison amateur's videos with our videos, our videos were more attractive than other videos.

Recently, e-learning is widely used at Universities and offices. An on-demand lecture consists of the pre-recorded streamed video/audio and synchronized slides. You can review the on-demand lecture anytime and anywhere at your convenience. However, the progress speed of the lectures can't automatically adjust learner's pace. Therefore, we constructed the self-directed learning system "Ghost-Tutor" that can automatically control the lecture videos depending on each learner's learning pace. We focused automatically eye movement to distinguish the learning pace. As a result of our experiments, we confirmed the frequency of blinking, the time period of gaze and the direction of the gaze showed significant differences. We designed the method to distinguish learner's the degree of concentration quantitatively. As a result of the system evaluation, we confirmed that Ghost-Tutor could presume the difference of learner's learning situation, and suggest learning pace.

Recently, e-learning contents that combine the speaker video with supporting materials are produced in educational institutions such as universities. However, there is a problem that those systems make learners become tired because produced contents are monotonous. In this paper, we propose the system "MINO : Multimedia system an Instructor needs Not Operate" that can automatically edit the recorded speaker video and supporting materials. MINO allows users to automatically convert words in the supporting materials into conspicuous ones according to the utterance of the speaker. We used speech recognition to convert voice of lecturer into character string. Their data are matched up words into the supporting materials so that speaker video is synchronized with supporting materials. Through evaluations of the system, we verified the effectiveness of our system.

Recently, the spread of portable digital video cameras enabled us to easily record various family events. Although videotaping and video editing have got quite popular recently, the quality of the produced video has not been improved much. For instance, movies edited by nonprofessionals are likely to be a sequence of monotonous scenes. When editing home movies, users usually use video sources from their own single camcorder. This is one of the significant reasons why nonprofessionals can not create high-quality home movies. We developed MediaBlocks, a web-based video editing system for sharing personal video clips. In the system, video sources are collected from multi-users' camcorders via the Internet, shared among the users, and edited through each user's web browser. The system allows users to use other users' video clips, each of which is spontaneously videotaped by different digital video camcorder, and also enables them to create home movies according to the professionals' important editing rules. The system can extract the date and time information of video-recording from the digital video clips, and synchronize them automatically. Through evaluations of the system, we verified the effectiveness of our system.

At vocational schools and corporate in-service training facilities, greater emphasis is usually put on giving a practical training than giving a lecture. However, traditional e-learning systems that provide learners with lecturer's audio/voice with synchronized presentation slides can hardly support the practical training. Our study focused on the creation of distance training environment in which people can learn physical operations, e.g., machine operations or instrument operations, in a remote or asynchronous manner. As a part of the study, we developed MultiVideoPlayer (MVP for short), a software playing back multiple video clips synchronously, each of which is spontaneously videotaped by different digital video camcorder. The software can extract the date and time information of video-recording from the digital video clips, and synchronize them automatically. Since it is well-known that synchronizing multiple videotapes needs an enormous amount of time and effort, MVP can greatly con-tribute to the creation of multi-angle video contents. Through evaluations of the system, we verified the effectiveness of our system.

Most of the information systems for business use have been developed recently as web-based applications. Unfortunately the web technology has been considered not to be suitable for the basis of any real-time groupware due to the lack of real-time communication capabilities of HTTP. It, however, would be obviously meaningful to develop a real-time groupware which takes advantage of the web-server-side technology. We developed QuickBoard web server system which allows users to deliver presentation slides of any applications to multiple web browsers in real-time. In the development process of the system, we took the approach of user-centered iterative design. We first studied the existing problems by observing real educational scenes, and developed prototype-1 to solve these problems. We then re-designed the system and developed prototype-2 based on feedback from the users of prototype-1. We were also interested in how to reduce network traffic on web-based real-time groupware, and implemented several methods for this purpose. This paper describes the implementations and evaluations of both prototypes.