Tetsuji Kuboyama

This paper proposes a novel method for analyzing PC usage logs aiming to find working patterns and behaviors of employees at work. The logs we analyze are recorded at individual PCs for employees in a company, and include active window transitions, our method consists of two levels of abstraction: (1) task summarization by HMM; (2) user behavior comparison by kernel Principle Component Analysis based on graph kernel. The experimental results show that our method reveals implicit user behavior at a high level of abstraction, and allows us to understand individual user behavior among groups, and over time.

本研究では、企業内のクライアント PC のログから個々のユーザの勤務形態や生産性について推定することを目標として、その要素技術について提案する．本稿では，実際に企業内 PC から約 60 ユーザーのマウスおよびキーボードの入力ログとアクティブウィンドウ遷移ログを取得し，クラスタリング手法と信号処理手法を用いて、部署ごとおよび個々のユーザーのアプリケーション利用パターンを分析した結果について報告する．また，PC 操作ログから，企業内における PC を利用した業務の基準的な利用パターンの抽出や，各個人の勤務形態を特徴付けるための要素技術について提案する．This paper proposes a simple method for analysing PC usage logs aiming to identify working patterns and behaviors of employees at work. The logs we analyse are recorded at individual PCs for employees in a company, and include active window transitions, the number of keystrokes, and mouse distance traveled per units of time. Through a simple method based on singular value decomposition (SVD) and clustering, we show that aggragate Application usage patterns of employees are extracted. We report results of analysis of individual users Application usage pattern from obtained And we propose a method for extracting basic pattern of working with PC and characterizing individual users working patterns from mouse, keyboard and active window transition logs.

This paper proposes a simple method for analysing PC usage logs aiming to identify working patterns and behaviors of employees at work. The logs we analyse are recorded at individual PCs for employees in a company, and include active window transitions, the number of keystrokes, and mouse distance traveled per units of time. Through a simple method based on singular value decomposition (SVD) and clustering, we show that aggragate Application usage patterns of employees are extracted. We report results of analysis of individual users Application usage pattern from obtained And we propose a method for extracting basic pattern of working with PC and characterizing individual users working patterns from mouse, keyboard and active window transition logs.

This paper proposes a simple method for analysing PC usage logs aiming to identify working patterns and behaviors of employees at work. The logs we analyse are recorded at individual PCs for employees in a company, and include active window transitions, the number of keystrokes, and mouse distance traveled per units of time. Through a simple method based on singular value decomposition (SVD), we show that aggragate PC usage patterns of employees are extracted.

金融工学数学のためのグラフィクス教材のデータベースのWEB サイトを立ち上げたので報告する。金融工学数学は難しいと言われるが，数学を理解させるために，グラフィクスの利用は有効であり，その利用により学生の理解度を高めることが可能である。我々は，大学および高校の数学授業において，グラフィクス教材をもっと多く活用してもらうためにこのサイトを開いた。そのため，数式処理ソフトウェアとしては，国内でも普及しているフリーソフトであるMAXIMA（マキシマ）を使った。本稿の目的は，グラフィクス教材作成に際して，ユーザー自身が，パラメータ値を自由に変更でき，ユーザー独自の金融工学数学グラフィクス教材を作成できるようにすることである。教材の内容としては，グラフィクスが効果的に生かせるテーマをとして，ポートフォリオの効率的フロンティア及びイミュニゼーションのテイラー展開描画，住宅ローン，年金の増加関数と減少関数の描画，債券価格の３次元的描画などを選んだ。本稿では，我々が研究中である，債券数学の学習支援システム構想にもふれ，このグラフィクス教材データベースが，その一部であることを説明する。

本稿では，大学における金融工学数学における確率の学習と，高校数学における確率の学習の関連を示す。金融工学数学の確率論の初歩が，数学Aの範囲の確率論の知識で十分理解可能であることを示す。その例としてオプションの理論における株価の二項モデルを示す。次に，数学Aでの確率論の教え方の改善について述べる。金融工学数学の中で，確率論は非常に重要である。しかしながら学生に確率の定義にある，同様に確からしい，の意味を教えることは容易ではない。順列の確率問題よりも，組合せの確率問題のほうが学生は困難を感じるようである。そこで，組合せの確率問題において，同様に確からしいサンプルの空間を，学生が容易にイメージできるようにする教授法を考案したので，本稿で提案する。サイコロや球を，識別子付きオブジェクトとして並べ，いずれのオブジェクトも同じ値域をもつようにモデル化する。この教授法をオブジェクト等確率モデル法と呼ぶ。本稿では，オブジェクト等確率モデル法により，組合せの確率問題の同様に確からしいサンプルの空間が，容易にイメージできることを示す。

本研究の目的は，金融工学数学教育のための学習支援システムの構築である．難度の高い金融関連文章題が与えられたときに，学生が演繹推論することを，支援するシステムを構築したい．この学習支援システムは，学生が解にいたるプロセスを演繹推論することを支援する．システムは知識ベースを検索し，必要な公式を探し，それを使った演繹推論の最適なパスを提示する．技術的課題としては，用語のユニークなセマンティクスのラベル付け，文章題日本語文からの Given Data と Unknown の抽出，公式のメタレベルでの記述，最適な解法プランの作成，等がある．The purpose of this research is construction of a computer-assisted learning system for financial mathematics. Given a hard bond mathematics word problem, the system helps a student conduct deductive reasoning to solve the problem. The system retrieves the knowledge database to obtain the formula that is required for the reasoning so that the system may construct and offer the solution path to the unknown. The problems we should overcome include systematic semantics labeling of bond mathematics terminology, extraction of Given Data list and Unknown from the given word problem, description of formula stored in the knowledge database, and construction of the optimal solution plan.

本研究の目的は，金融工学数学教育のための学習支援システムの構築である．難度の高い金融関連文章題が与えられたときに，学生が演繹推論することを，支援するシステムを構築したい．この学習支援システムは，学生が解にいたるプロセスを演繹推論することを支援する．システムは知識ベースを検索し，必要な公式を探し，それを使った演繹推論の最適なパスを提示する．技術的課題としては，用語のユニークなセマンティクスのラベル付け，文章題日本語文からの Given Data と Unknown の抽出，公式のメタレベルでの記述，最適な解法プランの作成，等がある．The purpose of this research is construction of a computer-assisted learning system for financial mathematics. Given a hard bond mathematics word problem, the system helps a student conduct deductive reasoning to solve the problem. The system retrieves the knowledge database to obtain the formula that is required for the reasoning so that the system may construct and offer the solution path to the unknown. The problems we should overcome include systematic semantics labeling of bond mathematics terminology, extraction of Given Data list and Unknown from the given word problem, description of formula stored in the knowledge database, and construction of the optimal solution plan.

A patent is an intellectual property which possesses legal uncertainty and it has the risk of patent invalidity at the stage of exercising the rights, e.g. license negotiation, demand for injunction, etc. In this study, we tried to search the factors which increase or decrease the risk, namely, the factors correlated to the validity of a patent. To be concrete, our analysis starts from selecting patents held valid or invalid by the Tokyo High Court (presently the IP High Court), terminated during the period of 2000-2006, and we set up an assumption if a patent is held valid in the court, the patent possesses high validity (difficult to be held invalid). Then, by using the method of machine learning for feature selection, we conducted a survey of the factors based on filing documents and the prosecution history. As a result, we found there existed several effective factors correlated to the validity of a patent in court level.

The growth of information and communication technology has changed the ways content is created and distributed. A new communication tool, such as Mobile Weblog (Moblog), enables users easily create and distribute information from anywhere in anytime using various terminals, such as PC and mobile phone. However, it becomes difficult to get information required and expresses emotion that facilitates communication with a receiver. This paper proposes SVM-based emotion extraction that extracts emotion of the person who writes Moblog text based on the correlation between text sentence and associated emoji using SVM.

The purpose of our research is to discover the creation process of users who do not already have high degrees of animation skills, and to provide a compatible authoring system that augments their process. It is very important that authoring systems have knowledge about a user's creation process. However, it is difficult to acquire this knowledge since it can vary depending on a user's preferred creation style. This paper proposes a new user model that considers both creation style and process, and explores an easy-to-use authoring interface design. We classify users with similar creation styles into clusters using a movie scenario structure of scenes and shots including events in the scenario text created by the users. Finally, an authoring interface based on the proposed user model is presented, and future research topics that must be addressed in order to achieve easy-to-use authoring are assessed.

Learning from tree-structured data has received increasing interest with the rapid growth of tree-encodable data in the World Wide Web, in biology, and in other areas. Our kernel function measures the similarity between two trees by counting the number of shared sub-patterns called tree q-grams, and runs, in effect, in linear time with respect to the number of tree nodes. We apply our kernel function with a support vector machine (SVM) to classify biological data, the glycans of several blood components. The experimental results show that our kernel function performs as well as one exclusively tailored to glycan properties.

The tree edit distance is one of the most prevailing frameworks to measure structural dissimilarity between two trees. It is well known that the effect of computing the tree edit distance is reduced to a structure called a tree mapping. The tree mapping gives us information about how similar two tree structures are; e.g., similarity in isomorphic subtrees, in topology, in hierarchical relation, and so forth. In this paper, we first give several important classes of tree mappings according to their structural similarities. Next, we give an efficient algorithm for checking alignability for amalgamating two unordered trees into one supertree in which these two trees are embedded. Through the checking process, the algorithm also identifies the class of tree mappings.

The web browsing has increased by the widespread diffusion of the Internet and the expansion of electronic commerce. Under these situations, it is assumed that the problem of mismatch between contents of web pages that the originator of the pages claims and that the recipient of the pages claims has been arising. The mismatch could be a serious issue for the financial interest. The established technique of the digital signature in the context of PKI(Public Key Infrastructure) can be imagined to be a technique to approach the problem since it supports the requirements of the integrity and the non-repudiation. However, the digital signature can only partially solve the problem. In fact, it can provide the recipient of web pages with the method to prove the existence of web pages that he or she has received only when they are signed by their originator. The originator can prove neither the existence of web pages that he or she has created in the past (since a malicious originator can sign web pages whenever it is advantageous for him or her) nor the non-existence of web pages tampered web pages. In the same way, the recipient cannot prove the non-existence of web pages the existence of which their originator claims. In this paper, we propose a moderation model for solving the mismatch problem. For that, we introduce Moderation Proxy. This model comprises of one or more proxy entities that cooperate with one another under a certain common policy. It can moderate the mismatch between contents of web pages with absolute equity. By this model, we can solve the current problem of mismatch.

The notion of the tree edit distance provides a unifying framework for measuring distance and finding approximate common patterns between two trees. A diversity of tree edit distance measures have been proposed to deal with tree related problems, such as minor containment, maximum common subtree isomorphism, maximum common embedded subtree, and alignment of trees. These classes of problems are characterized by the conditions of the tree mappings, which specify how to associate the nodes in one tree with the nodes in the other. In this paper, we study the declarative semantics of edit distance measures based on the tree mapping. In prior work, the edit distance measures have been not well-formalized. So the relationship among various algorithms based on the tree edit distance has hardly been studied. Our framework enables us to study the relationship. By using our framework, we reveal the declarative semantics of the alignment of trees, which has remained unknown in prior work.

We propose a novel general-purpose tree kernel and apply it to glycan structure analysis. Our kernel measures the similarity between two labeled trees by counting the number of common <I>q</I>-length substrings (<I>tree q-grams</I>) embedded in the trees for all possible lengths <I>q</I>. We apply our tree kernel using a support vector machine (SVM) to classification and specific feature extraction from glycan structure data. Our results show that our kernel outperforms the layered trimer kernel of Hizukuri <I>et al</I>.[9] which is well tailored to glycan data while we do not adjust our kernel to glycanspecific properties. In addition, we extract specific features from various types of glycan data using our trained SVM. The results show that our kernel is more flexible and capable of finding a wider variety of substructures from glycan data.

The web browsing has increased by the widespread diffusion of the Internet and the expansion of electronic commerce. Under these situations, it is assumed that the problem of mismatch between contents of web pages that the originator of the pages claims and that the recipient of the pages claims has been arising. The mismatch could be a serious issue for the financial interest. The established technique of the digital signature in the context of PKI (Public Key Infrastructure) can be imagined to be a technique to approach the problem since it supports the requirements of the integrity and the non-repudiation. However, the digital signature can only partially solve the problem. In this paper, we propose a moderation model for solving the mismatch problem. For that, we introduce moderator. This model comprises of one or more moderators that cooperate with one another under a certain common policy. It can moderate the mismatch between contents of web pages with absolute equity. By this model, we can solve the current problem of mismatch.