| 赤池 勇磨 | 1351001: M, 2回目発表 | 安本 慶一, | 加藤 博一, | 荒川 豊, | 玉井 森彦 |
|
title: Correlation analysis between online-log and offline-log and its application
abstract: In recent years, wearable health tracker become popular. Consecutive tracking is necessary not only for keeping a user's motivation but also for providing a proper health suggestion based on it. However, a user sometimes forget wearing the tracker. In addition, the tracker sometimes becomes out of battery. In this study, we propose a method for estimating a human activity as accurately as possible even in above mentioned cases. At that time, it is preferable not to request an additional task for complementing the missing data. Therefore, we focus on a user's online activities such as Twitter or LINE because these activities are implicitly recorded as an access-log. In this study, we clarify the existence of correlation between online activities (Facebook, Gmail, Twitter, LINE, etc.) and offline activities (Working, Running, etc.) and discuss about the possibility that online-log can be used for guessing the missing activity logs. In this experiment, two subjects put Fitbit Zip for collecting the users' offline activity. And LINE and Twitter are adopted as a representative online activities of both subjects. As a result, we confirm that a colocation between two logs exist and it is possible to estimate the user's offline activity by integrating several online activities. In addition, we propose the system to estimate user's busyness (free degree) as its application of this study. language of the presentation: Japanese 発表題目: オンラインログとオフラインログの相関関係分析とその応用 発表概要: 近年, 健康支援を目的としたウエアラブル活量計の利用が活発化している. 継続的なデータロギングは，ユーザのモチベーション維持だけではなく，そのデータを用いた健康行動支援においても重要となってくる．しかしながら，ユーザが装置を装着し忘れたり，装置の電池が切れることもある．本研究の目的は,こうしたセンシングデバイスの利用において, 装着し忘れや電池切れによって欠損したデータを補完することである．従来の統計的に補完するする手法に対して，本研究ではtwitterやLINEといったWebサービス上に残されたログ （以降，オンラインログ）を活用することを提案する．オンラインログは，サービスを使用した履歴として，暗黙的に残るログであることから，本提案にともなう追加作業が生じない点が特徴である．本研究では，ウエアラブル活量計によって記録される運動量（以降，オフラインログ） とオンラインログにどのような相関関係があるかどうかを明らかする. 被験者2 名から複数のオンラインログ (LINE, Twitter の利用履歴) とオフラインログ (Fitbit Zip によって取得された歩数データ) を収集し，分析した 結果, オンラインログと オフラインログの間には様々な視点から逆相関関係が見られ, 双方のデータを組み合わせて用いることで,行動推定がある程度実現可能であることが分かった．現在，その応用先として, オンラインログとオフラインログを組み合わせ，ユーザの忙しさ（暇さ）を推定するシステムを構築中である． | |||||
| 平部 裕子 | 1351091: M, 2回目発表 | 安本 慶一, | 加藤 博一, | 荒川 豊, | 玉井 森彦 |
|
title: The method for inferring internal contexts of user by utilizing user’s touch operation behavior on Android
abstract: In this study, we propose a method for inferring internal contexts of a human by analyzing a user’s touch operation behavior on Android. Although many researches on smartphone-based activity recognition (external contexts) have already proposed, the internal contexts will be more important to provide more appropriate ubiquitous services. The most interesting point of our proposal is to utilize a touch panel, that is generally used for input and output, as a sensor. We believe that some human internal contexts will be appeared on the user's touch behavior. Especially in case of static condition, an accelerometer does not work well for recognizing a human status such as "smoking". However, our proposal might recognize it because a user sometimes change a control hand to opposite hand. As a first trial, we develop a system for observing all the touch operation on Android. By using this system, we tried to recognize a user profile in terms of user operating type and handedness by analyzing the range of touch area and the arcs direction of swipe-curve of users. We experimented with four subjects to evaluate our system. As the result, the accuracy is beyond 60 percent. In the future, we continue to evaluate with more subjects. Futhermore, we will research about dynamic UI changing based on the user operating type. In addition, we will try to recognize user’s internal contexts, such as ‘smoking’, and ‘eating’, which are not obtained from accelerators. language of the presentation: Japanese 発表題目: タッチ操作の挙動に基づく内的コンテキスト推定手法の提案 発表概要: 近年，スマートフォンに搭載されたさまざまなセンサー用いた行動認識に関する研究が盛んに行われている． 行動認識の目的は，ユーザの行動に応じてあるいは予測して，最適なサービスを提供することであるが，より 最適なサービスを提供するためにはユーザのスキルや感情といった内面的な状態にまで踏み込む必要があると 考えている．加速度センサを用いた行動認識では，歩行や乗り物といったフィジカルな行動認識は可能である が，ユーザが動いていない場合は何も検知できないという問題がある．例えば，喫煙中というコンテキストは 加速度センサでは検知できない．そこで，本研究では，入力インターフェースであるタッチ操作自体をセンサ として用い，タッチ状況から内面的な状態を推定する手法を提案する．今回，内面的な状態推定に先立ち， 使用しているユーザのプロファイリングに取り組んだ．具体的には，タッチエリアの範囲やスワイプ時に現れ る曲線の弧の向きを分析し，これらの組み合わせから利き手を含めた7つの操作形態の推定手法を提案する． 4人の被験者による評価の結果，7つの操作形態を60%以上の確率で判定可能であることがわかった．今後は，操作 形態に応じた動的なUIの研究を行うとともに，喫煙中など加速度ではわからない状態の判別，さらにはストレ スなど内面的な状態の判別を行いたいと考えている． | |||||
| 尾上 紗野 | 1351025: M, 2回目発表 | 松本 健一, | 安本 慶一, | 門田 暁人, | 畑 秀明 |
|
title: The Current and the Future of OSS Development Communities, and Developer Classification Based on Developers' Activities in GitHub
abstract: Since human power is an essential resource, the contributors in a software development community is one of the health indicators of an open source software (OSS) project. For maintaining and increasing the populations in software development communities, both attracting new contributors and retaining existing contributors are important. Also, in OSS projects, there should be different types of developers, and knowing the structure of them are important. In this study, we research from the two point of view. language of the presentation: Japanese 発表題目: OSSプロジェクトにおける貢献者数の推移予測と活動履歴分析による貢献者の特徴分類 発表概要: オープンソースソフトウ（以下，OSS）プロジェクトの普及に伴い，OSSを導入する企業が増加している． しかし，ボランティアが融資で開発を行っているというOSSの性質上，有料ソフトウェアのような十分な保証が受けられないというリスクがある． 企業は健全性の高いプロジェクトを見極める必要があるが，それは必ずしも容易でない． そこで我々は，プロジェクトが貢献者の獲得を十分にできているかということと， プロジェクトが構成している貢献者の役割に偏りがないかという2つの人的リソースの観点から，OSSの健全性評価指標の提案を行う． | |||||
| 伊藤 冬樹 | 1351009: M, 2回目発表 | 杉本 謙二, | 安本 慶一, | 松原 崇充 | |
|
title: Data driven type feedforward learning control for non-linear systems abstract: This study proposes a technique for feedforward learning control with a data driven modeling method. A previous research has applied a scheme for data driven modeling to Feedback Error Learning control via a linear filter corresponding to a local model of inverse dynamics at each of the operating points. It uses a round-off method for accumulation, and Locally Weighted Regression for regression in a local model around the equilibrium point. We propose yet another scheme that uses an interpolated technique for accumulation and regression in local modeling, and also propose a parameter adapting law for a linear approximate model around each operating points, thereby improving accuracy of response shaping. language of the presentation: Japanese | |||||
| 程 飛 | 1361023: D, 中間発表 | 松本 裕治, | 中村 哲, | 新保 仁, | Kevin Duh |
|
title: Using SMT to build a multilingual dictionary
abstract: Multilingual dictionary plays an important role in many natural language processing tasks. For example, machine translation uses bilingual dictionary to reinforce word and phrase alignment. Although lots of bilingual dictionary resources exist, the enormous productivity of language leading to a large number of ad hoc words and corresponding translations. We introduce a methodology for using SMT method to translate a common word into another language. By using the lanugage link information of the Wikipedia, our system can generate more candidates from several source lanugages. Finally, our system adopts a ranking method to generate most possible results. language of the presentation: English | |||||
| 秋間 大輔 | 1351003: M, 2回目発表 | 松本 裕治, | 中村 哲, | 新保 仁, | 進藤 裕之 |
| 大村 舞 | 1351020: M, 2回目発表 | 松本 裕治, | 中村 哲, | 新保 仁, | 進藤 裕之 |
|
title: A Japanese predicate argument structure analysis considering multiple predicates
abstract: A predicate argument structure (PAS) give useful information to text processing tasks such as machine translation, information extraction. However, a Japanese PAS analysis is difficult task because pronouns abbreviated (zero pronouns) often occur when the argument has already been mentioned even between a sentence (intra-sentence). Threfore we forcus PAS analysis to one sentence and seek useful features for zero pronouns in a sentence. We use the relation between multiple predicates to improve the precision of PAS because zero pronouns in the intra-sentence oftern occur when multiple predicates occur. language of the presentation: Japanese 発表題目: 複数の述語関係を考慮した述語項構造解析 発表概要: 述語項構造解析は機械翻訳や述語項構造解析に有用であることが分かっている．しかし，日本語における 述語項構造解析は以前現れた項を省略することができるため，文内ゼロ項という述語と直接関係ない箇所に現れる項が発生する． そこで我々は，一文の述語項構造解析に焦点を当て，文内ゼロ項の精度構造に有用な素性を求める．文内ゼロ項は ，一文に焦点を当てた場合，複数の述語が現れたことによって発生してしまう現象である．そこで，文中の複数の述語の係り受け関係を予測モデルに組み込むことで，文内ゼロ項の述語項構造解析の精度向上に取り組んだ． | |||||
| AMORNPASHARA NUTTAPOOM | 1351115: M, 2回目発表 | 安本 慶一, | 中村 哲, | 荒川 豊, | 玉井 森彦 |
|
title: Phorec: Context-Aware Landscape Photography Support System with Special Camera Setting
abstract: DSLR camera becomes cheap and popular, it is not easy for beginners to take a good photograph by using it.This is because a user has to set various parameters such as ISO, shutter speed and aperture based on the user's situation or contexts. To solve this problem, I propose a context-aware photography support system, called "Phorec". In order to deal with various situations, Phorec utilizes huge number of photograph stocked on social photo services such as Flickr and 500px. In addition to the basic information included in EXIF, such as location, time, and camera setting, we include weather information for considering an obvious effect to the sunlight of weather condition. Furthermore, we add a function to pull out noteworthy parameters of each photo and implicitly inform user about the significance of each value based on statistical analysis. Field-experiments done in the real situation reflects that users can take better photos and understand more about the photography settings after using the system. I begin with background and importance of each photography settings followed by some related works. Then I introduce my system design and progress. Finally, I show the experimental result, plan for future work, and achievements. language of the presentation: English | |||||