Дисертації з теми "007:681.3.06"

Tese de Doutoramento em Tecnologias e Sistemas de Informação - Área do Conhecimento Engenharia de Programação e dos Sistemas Informáticos
A visualização é a representação visual realística ou abstracta de um conjunto de dados que são gerados por modelos computacionais ou resultantes de medições físicas realizadas no mundo real. É fundamental para auxiliar as pessoas a compreenderem dados e processos complexos e pode ser classificada consoante os seus objectivos (nomeadamente a visualização científica e de informação). A correcta modelação e caracterização dos dados são partes fundamentais para a escolha de técnicas visuais eficazes e a produção de uma visualização válida. O grande desafio é exactamente o de identificar como a análise dos resultados pode e deve ser mostrada ao potencial utilizador de uma forma simultaneamente sucinta, coerente e útil. O conceito de modelação cartográfica ou álgebra de mapas foi desenvolvido por Dana Tomlin em 1983 com o Map Analysis Package1 [Sendra2000]. Um modelo cartográfico pode ser visualizado como uma colecção de mapas registados numa base cartográfica comum, em que cada mapa é uma variável sujeita a operações matemáticas tradicionais. A modelação é um processo que decorre de operações primitivas de pontos, vizinhança e regiões sobre diferentes mapas, numa lógica sequencial para interpretar e resolver problemas espaciais. Neste contexto, a sequência de operações é similar à solução algébrica de um conjunto de equações. A criação de ferramentas informáticas para a análise e visualização de dados relacionados com o corpo humano é uma área em forte expansão e de especial interesse. Apesar destas ferramentas serem muito úteis, sofrem bastante da limitação imposta pela arquitectura dos modelos utilizados para o seu desenvolvimento e consequente implementação. Isto ocorre porque estes modelos adoptam os mesmos princípios e ponderações que são aplicados a dados de natureza não humana ou biológica e tratando-os de forma independente e atómica. Por outro lado, a utilização de técnicas visuais pouco intuitivas no sentido de denotar a interdependência espacial inerente a este tipo de informação é outra limitação a salientar neste tipo de ferramentas. Os dados relacionados com o corpo humano apresentam uma forte componente espacial. Para que seja possível uma análise e investigação correctas é necessário ter isso sempre em consideração. Um bom exemplo desta situação é o diagnóstico médico. A combinação de informação oriunda de diferentes partes do corpo humano é normalmente necessária para que um médico possa diagnosticar a doença de um paciente. O acto de diagnosticar pode ser traduzido por um conjunto de operações de álgebra de mapas executadas sobre os dados relacionados com o corpo humano do paciente. Qualquer modelo que pretenda servir de base para o desenvolvimento e implementação de ferramentas informáticas orientadas para a medicina, e em especial, para a análise e visualização de dados relacionado com o corpo humano, deve incorporar os princípios fundamentais da modelação cartográfica. Desta maneira, é possível que os dados possam ser devidamente modelados e consequentemente extrapolada mais informação útil. Por outro lado, a utilização da visualização como instrumento de comunicação de resultados, com a inclusão de metáforas visuais cartográficas é outra mais-valia a ter em conta. O modelo CHUB (Cartographic Human Body), que é apresentado neste trabalho, pretende colmatar essa falha identificada no tratamento e visualização de dados relacionados com o corpo humano. Utiliza a modelação cartográfica como alicerce fundamental para a análise dos dados e a visualização científica e de informação como meio para a comunicação de resultados. Para ser possível a sua avaliação e validação foram considerados dois estudos de caso: diagnóstico da artrose no joelho e a análise de sessões de hidrocinesioterapia. Para estes dois estudos de caso foi implementado um protótipo que instancia o modelo CHUB nestes casos particulares, permitindo a sua utilização, avaliação e validação em dois domínios específicos. Os resultados obtidos após a utilização e avaliação do protótipo permitiram validar com sucesso o modelo CHUB proposto nesta tese de doutoramento.
Visualization is the realistic or abstract visual representation of a dataset that is generated by computer models or resulting from physical measurements of the real world. Visualization is fundamental to help people understand data and complexes processes and can be categorized according its goals (scientific or information). The correct data model and characterization are essential to the right choice of the visualization techniques and the production of useful visualizations. The great challenge lies in how to determine that the results are showed to the final users at the same time in a coherent, useful and simple way. The cartographic model concept was developed by Dana Tomlin in 1983 with the Map Analysis Package2 [Sendra2000]. A cartographic model can be seen as a collection of maps that are registered in a cartographic database, where each map is a “variable” that can be mathematically operated. These operations may involve primitives such as points or areas of different maps, for example, in a sequential order to interpret and solve spatial problems. In this context, the sequence of operations is similar to the algebraic solution of a group of equations. The creation of automatic tools for human’s body data analysis and visualization is a field in expansion and of great interest. However these tools are very valuable, they suffer from a common limitation that is imposed by their basis architectural model. In general, they rarely represent in a suitable way biological, morphological and/or biomedical data spatial interdependency. These models treat data in an almost total focused and independent way. The human body systems and organs work as a complex machine, where each part depends strongly on the others. This dependency might be stronger or weaker to the system or organ importance on the overall patient condition. The doctor diagnoses an illness by comparing and analyzing information not only directly related to the mostly affected organ, but also to the body as a whole. In fact the doctor performs a subtle spatial analysis, and therefore, executes a typical algebraic map operation in his/her mind, when diagnosing a patient. An illness might arouse different symptoms and physiological changes in systems/organs that are not directly related to the spatial location of it. CHUB is a model that was developed taking into consideration the main principles of cartographic modelling. It structures data according to different layers of information. Each layer is associated to a specific organ and/or system, and might contain geometric data or attributes that are “human-referenced”. CHUB has not been developed as a dynamic model. It is considered that dynamic issues related to human’s body data, such as body movement, blood flow or heartbeat (besides others) will be accomplished by other models that should be used as a specialized extension to CHUB. In order to validate CHUB two cases of study were considered – osteoarthritis knee diagnosis and hydrokinetic therapy sessions analysis, proposed two strategies for its validation and a prototype implemented. This prototype allowed its utilization, evaluation and validation in two different domains. The results achieved after its utilization and test lead to a complete CHUB validation.

Магістерська дисертація: 185 с., 84 рис., 40 табл., 20 джерел, 2 додатки. Актуальність теми: прогнозування сонячної активності є дуже важливою задачею в сучасному світі, адже космічна погода дуже сильно впливає як на навколоземні супутники та космічні станції, так і на різні процеси на Землі Об'єктом дослідження є процеси сонячної активності, представлені статистичними даними. Предмет дослідження: математичні моделі для формального опису процесів сонячної активності, методи фільтрації, згладжування і прогнозування на основі статистичних даних. Мета дослідження полягає у побудові та аналізі характеристик адаптивного фільтра Калмана і гранулярного фільтра, а також їх використання для прогнозування сонячної активності, у тому числі: виконання теоретичних досліджень та створення програмного продукту, що реалізовує вказані методи. Методи дослідження: оптимальний адаптивний фільтр Калмана, гранулярний фільтр, експоненційне згладжування, згладжування за Калманом; програмна реалізація методів виконана в середовищі Matlab R2017b, Python3 та C#. Отримані результати: розроблено метод для прогнозування сонячної активності на основі адаптивного оптимального та гранулярного фільтрів. Зроблено порівняння отриманих результатів. Наукова новизна: новий метод і моделі для фільтрації та прогнозування нестаціонарних часових рядів на основі оптимального адаптивного фільтра Калмана та гранулярного фільтра.
The theme: Method and models for solar activity forecasting A master dissertation contains 185 pages, 84 drawings, 40 tables, 20 sources, 2 appendixes. Actuality of topic: forecasting of solar activity is a very important issue in a modern world because space weather has a great impact on Earth observation satellites and space stations as well as on different processes on Earth. The research object: processes of solar activity represented by statistical data. The research subject: mathematical models for the formal description of solar activity processes, methods of filtration, smoothing and forecasting on the basis of statistical data. The purpose of the research is to construct and analyze the characteristics of an adaptive Kalman and particle filter, as well as their use for prediction of solar activity, including: performing theoretical studies and creating a software product that implements these methods. Research methods: optimal adaptive Kalman filtering techniques, particle filter, exponential smoothing, Kalman smoothing; the software implementation of the methods mentioned in the environment of Matlab R2017b, Python3 and C#. Results: the method for forecasting of solar activity on the basis of adaptive optimal and particle filters is developed. A comparison of the experimental results was performed. What is new: new method and models for filtering and forecasting nonstationary time series on the basis of optimal adaptive Kalman filter and particle filter.

Il marketing cinematografico effettuato nel corso degli anni sulla serie di film di 007. La saga rappresenta una parte importante di storia del cinema che al difuori della sala ha avuto influenze importanti sulla società, sulla moda e sul costume di diverse generazioni. Dai selling elements, al merchandising, al product placement e alle campagne pubblicitarie, vengono analizzati gli aspetti che hanno permesso il successo del personaggio di James Bond e lo hanno reso un fenomeno commerciale, portandolo a rappresentare un vero e proprio brand.

Mobility is a key element of the future Internet. The location privacy problem is one of the problems involved in mobility. A great benefit of universally available access to the Internet might bring a risk such that a user’s location is traceable by others. Most of the mobility protocols define a mechanism of informing the correspondents of location change, in order to realize mobility. Therefore, the correspondents and eavesdroppers on the path will notice a movement and its destination. The problem changes its situation, according to the node from which the location of a mobile node should be concealed. We classify the nodes into two types: correspondents or onlookers on the path. In addition, we often assume there are some trustworthy nodes on the path. There are some existing researches on this problem, e.g. HIP Location Privacy Framework by Matos et al. and BLIND by Ylitalo and Nikander. They showed that it is possible to conceal location from a correspondents and a part of onlookers by introducing a trustworthy helper node, and especially when mobility is not needed at all, from all the onlookers, too. In this research, we have proposed a new framework using Host Identity Protocol (HIP), and with it we showed that it is also possible to protect location privacy from all other nodes in IP communication with mobility. We take advantage of the notable feature of HIP that public keys are used as host identifiers, so that our framework gives a way to separate IDs for mobility from those for end-to-end communication. We constructed an extensional mobility management protocol of BLIND, and discussed a trade-off in terms of efficiency and operational cost.
モビリティ技術は今後のインターネットの発展を考える上で重要な位置を占める要素である．モビリティと深く関連する問題のひとつとして，ロケーションプライバシ問題がある．あらゆる場所からインターネットへアクセスが可能となる利便性の裏には，ユーザの位置を第三者に追跡されるリスクがつきまとう．多くの場合，モビリティプロトコルでは通信相手に自身の移動を知らせることによってモビリティを実現する．そのため通信相手や盗聴者が移動ノードの位置の変化およびその移動先を知ることになる．この問題はユーザの位置情報を誰に対して秘匿するかによって状況が異なる．秘匿対象として通信相手と通信経路上の第三者の二種類を考え，さらに後者の一部に信頼できるノードを仮定する場合がある．この問題に対する従来研究としてMatos らによるHIP Location PrivacyFramework やYlitalo らによるBLIND などがあり，信頼できる補助ノードを導入することで通信相手や一部の盗聴者に対する位置の秘匿が可能であることや，モビリティを考慮しない状況に限れば，通信相手および通信経路上の全ノードに対する位置の秘匿が可能であることが知られている．本研究において我々はHost Identity Protocol (HIP) を使った新たな手法を提案し，ネットワーク間の移動を伴うIP 通信においても，すべての対象に対するロケーションプライバシの保護が可能であることを示した．我々の手法では，公開鍵がホストの識別子として使われるというHIP の特徴を活用し，移動用のID と通信用のID とを分離する．これに基づいてBLINDに対してモビリティ管理を行うための拡張プロトコルを構成し，モビリティとロケーションプライバシの両立に伴う通信効率や運用コストとのトレードオフについて考察した．
Kyoto University (京都大学)
0048
修士
修士(情報学)

本研究では参加者5, 6 人がプレゼンテーションとフリーディスカッションを交えて行うミーティングを対象として，その様子を映像と音声で記録するとともに，参加者の視線集中，写真撮影と電子メモの記入，プレゼンテーションスライドの操作などを検出してインデクスを生成することで，後から振り返るときのための利便性を確保したミーティングキャプチャシステムを構築する．このインデクスは映像の自動推薦を行う際に利用することを想定しており，過去をうまく想起させるような部分や，複数映像の中のふさわしい部分を選択するための指標となるようにする．収録したデータを自発的に振り返るときにもこのインデクスが手がかりとして提示され，見る人に対してどこから見ればよいのか，どのあたりが重要なシーンであるかが視覚的にわかるようにする．視線集中の獲得は，参加者の顔方向および視線方向をミーティングスペース中央の全方位カメラによって観測し，東西南北のうち特定の方向に対して過半数の集中が起きたときを意味のある重要なシーンが起こったと推測する．この際，顔方向と視線方向は独立に判定される．顔方向集中と視線方向集中では顔方向集中のほうがより印象的な会話イベントが起こったことが示唆され，その特徴としては笑いやジャスチャを伴うことや，会話が質問応答の形式になっていることが挙げられる．また参加者は捉えておきたい場面に対して，カメラのシャッターを押したりメモを書いたりする行為によって実時間タグと呼ぶインデクスを意識的に付与することが可能である．実時間タグのインデクスでは画像が生成されてから最後にメモ書きが起こるまでの活動時間に注目することにより，話題の継続性への手がかりとすることができる．この両者は同時刻にはあまり発生せず，両者のインデクスは性質が異なる相補的なものである．そのため両者のインデクスは目的によって使い分ける必要がある．
In this paper, I construct a capture system for meetings, in which 4 to 5 participants make presentations and/or free discussions. It enables participants and none-participants to look over captured meetings. This system captures paging of presentation slides, pictures by PhotoChat and notes as indices in addition to videos and audios. Besides captured indexes, gathering of gazes are calculated and used as index. I suppose these indexes are good marks to recall meetings and to select a video from multi view point and they will be variable in auto-recommendation. They are also used as markers to show where to start to look and where to look for their impotance when users want to look back. A spherical vision camera set up at the center of meeting table captures meeting participant’s faces in order to recognize that their gaze points gather at the same area. When the majority of them gather their gaze in any area of north, south, east, and west, the system guesses that a significant event occurs and generates the index. The precision ratio of the judgment of gaze gathering is about 65Some features are found in the conversations around the time that gaze gathering occurs. Participants also can give real-time tags to the scene for which they want to leave the indices intentionally by pressing the shutter button and writing the memos. The length of time from to the memo book daily life and happening of real time at the end can be used as a clue to the continuance of the topic. The length of the time between the generation of the image of real-time tag and the last memo can be used as a clue to the continuance of the topic. Both are not so generated at the same time, and the indices of both is the complementary one another. Since the indices of both differ, it is necessary to use them properly by the purposes.
Kyoto University (京都大学)
0048
修士
修士(情報学)

In this paper we design a certified e-mail exchange system based on simultaneous secret exchange protocol proposed by Okamoto and Ohta. We selected their protocol because it is superior in efficiency, especially at the view point of the number of sessions. At first, we designed the whole procedure mainly based on Okamoto and Ohta's protocols, adding processes to confirm the contents each party have received. In this system, we want to realize good properties often referred in certified mail researches with TTPs, for example 'send-and-forget'. So we assume two players, an MTA (Mail Transfer Agent, e.g. SMTP Server) and an MUA ( Mail User Agent, e.g. the client the sender or the receiver use) in each party, and designed construction which enables 'send-and-forget' for the sender when the sender can control an MTA, or enables to exchange messages for the sender with the receiver directly when cannot. Then, we made message formats in the form of XML to send each transactions. We selected the method of attaching transaction XML to e-mail to communicate messages to realize reachability. With this method, it takes rather long time to complete exchanging secret keys and get contents of the mail or receipt when the mail delay is large. Then we proposed a method to reduce the number of exchanging transactions. This method doubles the number of bits sent at once after some bits have been sent regularly. The 'fairness' is defined as the maximum of the difference of time needed to calculate the remaining bits of keys at any time, where the difference is taken between two players. This amount is 'at most twice' in a normal simultaneous secret exchange protocols, while it is 'at most constant difference' in our method.
本論文で我々は証明付き電子メール配達システムを同時秘密交換プロトコルに基づいて設計した. 我々は効率, 特にセッションの回数の観点から優れていることから岡本・太田らの提案したプロトコルを選択した.岡本らのプロトコルを主軸として, 我々は双方が受け取った内容に相異がないかの最終確認のプロセスなどを加えて, メッセージと領収書を交換するための手続きを設計した. システムを設計するにあたって我々は, TTP を用いて構成される証明付きメールの研究で頻繁に言及されるSend-and-forget など良い性質を実現させたかった. そこで, 我々はそれぞれのパーティ内にMTA (Mail TransferAgent, SMTP サーバなど) , MUA (Mail User Agent, 送信者, 受信者が用いるメールクライアント) の2 者が存在することを仮定した. これにより送信者がMTAを制御下に置ける場合にSend-and-forget をMUA のユーザレベルでは実現でき,そうでない場合には受信者と直接メッセージ交換を行える構成を設計することができた. そして, それぞれのトランザクションを伝送するためのメッセージフォーマットをXML としてまとめた.我々は到達可能性を考慮してメッセージの伝達手段としてメールにトランザクションのXML を添付するという形式を選んだが, この方法ではメールの一往復あたり遅延が大きい場合, 必要な秘密鍵の交換を完了してメールや領収書の内容を入手するまでにかなり長い時間がかかってしまう. そこでセッションの往復回数を減らすための手段として, ある数のビットを普通にに送った後には1度に送信する数を増加させていくという手法を提案した. 同時秘密交換プロトコルでは両者が鍵を計算して解こうとしたときにかかる時間の公平性が求められ, それは通常は「かかる時間割合が二倍以内である」という条件により保障されている. 我々の手法においては, 「かかる時間差が大きくなることはない」ことにより保障される.
Kyoto University (京都大学)
0048
修士
修士(情報学)

調理の様子を観測した映像アーカイブの各シーンに対し，そのシーンで使用されている食材名を注釈として付与するためには，調理者が扱っている食材を常時特定している必要がある．調理中の食材は，切断・混合・加熱などの操作によって外見が変化するので，食材を継続的に追跡することによって個体を特定するアプローチをとる．操作時に容器を用いる混合・加熱の操作では，容器に取り付けたセンサなどを用いて食材を追跡することが可能である．一方，手で直接行う操作である切断では，センサ情報による追跡が不可能である．そのため本研究では切断などの直接手で操作を施している最中の食材を追跡することを目的とする．食材は切断されると，多数の断片になったりひとまとめにされたりするため，映像中での食材領域は，多数の分裂と統合を繰り返す．追跡を実現するためには，各フレームで分裂・統合した領域をフレーム間で対応付ける必要がある．しかし，分裂・統合した領域を対応付けるとき，その候補は膨大になりうるし，各領域の画像特徴は切断とともに大きく変化しうる．そのため，考えうるすべての候補から画像特徴のみを用いて正しい対応付けを求めるのは困難である．従来手法では，各時刻で分裂・統合しうる食材領域が1種類のみであるという制約のもと，追跡を実現していた．対応付け候補が多数あっても，画像特徴が変化する食材領域が１種類であることがわかっていれば，対応付けは容易である．しかしながら，この制約は，実際の調理では実現性が低い．本研究では実際の調理での有効性を考慮して上述の制約を緩和し，同色の食材については分裂・統合しうる食材は1種類である，という制約のもと，追跡を行う手法を提案する．制約の緩和により，複数種類の食材についての対応付けが一意に定まらなくなる．そこで，食材領域の画像特徴変化モデルを導入することで，食材領域の対応付け候補について尤度を求め，尤度が高い候補を採用することで，対応付けを定める．以上で述べた手法の有効性の確認のため，画像特徴変化モデルの妥当性を評価する実験を行った．実際の調理を観測した映像から，食材領域の画像特徴変化モデルにより，異色の食材からなる領域間の対応付け候補を棄却できることを示した．また，提案制約が実際の調理でどの程度満たされるかについて，調査を行った．多数のレシピをもとに，従来手法と提案手法が導入している制約が成立する割合を検証し，実際の調理における有効性が向上することを確認した．
Annotation of food stuff to a cooking video is convenient for searching scenes in which the food stuff is used. In order to annotate food stuff in the video, identification of the food stuff is required in the whole video. Food stuff undergoes various changes through cooking processes: cut, mix, boil, etc. Since it is difficult to recognize food stuff in a single captured image, we track the food stuff from the beginning of the cooking. In a scene of mixing or boiling food stuff, the cook puts the food stuff in some container: bowl, pan, etc. We can track easily the food stuff by a sensor attached to the container. In a scene of cutting food stuff, the cook manipulates the food stuff directly by the hands. We can not attach sensors directly to the food stuff. Thus we aim to track food stuff handled directly by cook’s hands in a video. The cooking processes mentioned above will split the food stuff into pieces, or gather up pieces of food stuff. This is observed in the video as split and merge of regions of the food stuff. Tracking the food stuff in the video requires finding correspondence of split and merge of regions between frames. The number of candidates of the correspondence in the whole video can be vast and the appearance of the regions of each correspondence can be different between beginning and end. Therefore, it is difficult to determine the correct correspondences from all the candidates by their appearance between beginning and end. Previous work introduces the constraint that the cook processes only one kind of food stuff. Under the constraint, the number of species of food stuff is one that causes split and merge of the regions simultaneously between frames. Since, under the constraint, a number of candidates in a frame originate from only one species of food stuff, finding correspondence is easily obtained. However, the constraint is often unsatisfied. In this work, we relax the constraint to one that is easily satisfied. We introduce the constraint that the cook processes separately the food stuff that has different appearance. For reducing ambiguity in finding correspondences caused by the relaxation, we introduce an appearance transition model of food stuff. We adopt the candidate with the highest likelihood among the candidates, and thus tracking food stuff is obtained. We implemented the method mentioned above, and evaluated the capability of calculating the likelihood and determining the correct correspondence. In addition, we assessed the possibility of the proposed constraint satisfied. The results showed that the method can increase the feasibility in real activities.
Kyoto University (京都大学)
0048
修士
修士(情報学)

近年，ウェブページが爆発的に増加しており，我々は検索エンジンを用いることにより多種多様な情報を得ることができる．しかし，ウェブページの約40%が類似ページといわれており，検索結果に類似ページが含まれるという問題がある．本研究では1 億ページという大規模なウェブコレクションを対象として，類似ページ検出を行なう．本研究では類似ページを，文字列をある程度共有する2つのページと定義し，ミラーページなどの同一ページ，引用ページ，盗作ページなどが含まれる．本手法はまず，各ページから長い低頻度の文を抽出する．これは，文長が長く，また，ウェブ全体での頻度が低い文を2 ページで共有すればこれらのページは関連性が高いといえるためである．また，各ページにおいてコンテンツ領域を抽出し，コンテンツ領域にある文のみを類似ページ検出の手がかりとする．これは非コンテンツ領域にある文を共有しても2 つのページに関連性が低いからである．以上の処理によって得られた文を共有するページペアを類似ページとみなす．次に，類似ページを同一ページ，引用ページ，盗作ページなどに自動分類する．分類は，ページに対する類似文字列の割合である重複率，インリンク/アウトリンクの有無，URLの類似度などの様々な情報を用いて行なう．類似ページ検出の実験を行なったところ，単純なURLの正規化ではわからないミラーページや，引用ページ，様々なサイトから記事をはりあわせたようなスパムページを発見することができた．
The recent explosive increase of Web pages has made it possible for us to obtain a variety of information with a search engine. However, by some estimates, as many as 40% of the pages on the Web are duplicates of the other pages. Thus, there is a problem that some search results contain the duplicate pages. This thesis proposes a method for detecting similar pages from a huge amount of Web pages: hundred million Japanese Web pages. Similar pages are defined as two pages that share some sentences, and are classified into mirror pages, citation pages and plagiaristic pages, etc. First, from each page, relatively long sentences are extracted. This is because two pages tend to be relevant when they share relatively long sentences. A pair of pages that has the identical sentences is regarded as similar pages. Next, similar pages are classified based on several information such as an overlap ratio, the number of inlinks/outlinks, and contents region extraction. We conducted the similar page detection and classification on the large scale Japanese Web page collection, and can find some mirror pages that we cannot find by the simple URL normalization, citation pages, and plagiaristic pages.
Kyoto University (京都大学)
0048
修士
修士(情報学)

バーチャルスタジオを教育映像の撮影に利用すれば，CGによる特殊効果が加わった三次元の仮想物体を教材として利用できるため，視認性の高い教育映像を作成できる．このとき，教師である演者による説明に迫真性をもたせるためには，演者が仮想物体の説明箇所を直接見て判断し，その説明箇所を視聴者に見せるための並進・回転という直接操作を交えた主体的な説明が要求される．演者が三次元の仮想物体を直接操作できるようにするため，仮想物体と同形の現実物体を導入し，実写映像中の現実物体を仮想物体に置き換えるという手法がある．この手法では，演者が特定色の現実物体を直接操作する様子をカメラで撮影し，実写映像の特定色領域に仮想物体を重畳することで目的を達成している．しかし，現実物体は単色であることから，演者は仮想物体の説明箇所を直接見ることができないので，演者は主体的に説明することができない．本研究では，演者が仮想物体の説明箇所を直接見られるようにするため，現実物体に仮想物体のテクスチャ（演技支援情報）を投影する．但し，演技支援情報が実写映像に映り込むと現実物体領域が特定色ではなくなるため，過不足なく仮想物体を重畳できない．また，現実物体に対する演者の視線方向と投影方向が異なると，演者には演技支援情報が投影されない部分が見えてしまう．本研究では，演者が物体を説明対象として説明する場面において物体に対する演者とカメラの視線方向は異なるという一般的事実に基づき，再帰性反射材を貼付した現実物体に演者の側から演技支援情報を投影することで，演技支援情報の実写映像への映り込みを回避する．また，カメラ・演者・物体の位置関係，再帰性反射材の反射輝度分布，演者の手の動きを基に，演者の身体による遮蔽が発生しない演者後方の投影位置を特定することで，演者の視線方向と投影方向を一致させる．提案手法の有効性を検証するため，演者の視点位置の近傍に設置したカメラとスタジオカメラの2台のカメラで現実物体を観測し，それぞれのカメラから得られる映像を目視で定性的に評価した．その結果，演技支援情報が演者には視認可能でスタジオカメラには撮影不可能となっていることを確認した．
Kyoto University (京都大学)
0048
修士
修士(情報学)

As computers and their networks continue to be developed, our day-to-day lives are being surrounded by increasingly more complex instruments, and we often have to ask questions about using them. At the same time, large collections of texts to answer these questions are being gathered. Therefore, there are potential answers to many of our questions that exist as texts somewhere. However, there are various gaps between our various questions and the texts, and these prevent us from accessing appropriate texts to answer our questions. The gaps are mainly composed of both expression and vagueness gaps. When we seek texts for answers using conventional keyword-based text retrieval systems, we often have trouble locating them. In contrast, when we ask experts on instruments or operators of call centers, they can resolve the various gaps, by interpreting our questions flexibly, and by producing some ask-backs. The problem with experts and call centers is that they are not always available. Two approaches have been studied to resolve the various gaps: the extension of keyword-based text retrieval systems, and the application of artificial intelligence techniques. However, these approaches have their respective limitations. The former uses texts or keywords as methods for ask-back questions, but these methods are not always suitable. The latter requires a specialized knowledge base described in formal languages, so it cannot be applied to existing collections with large amount of texts. This thesis targets real-world the large text collections provided by Microsoft Corporation, and addresses a novel methodology to resolve the gaps between various user questions and the texts. The methodology consists of two key solutions: precise and flexible methods of matching user questions with texts based on NLP (natural language processing) techniques, and ask-back methods using the matching methods. First, the matching methods, including sentence structure analysis and expression gap resolution, are described. In addition, these methods are extended into matching through metonymy, which is frequently observed in natural languages. After that, a solution to make ask backs based on these matching methods, by using two kinds of ask-backs that complement each other, is proposed. Both ask-backs navigate users from vague questions to specific answers. Finally, our methodology is evaluated through the real-world operation of a dialog system, Dialog Navigator, in which all the proposed methods are implemented. Chapter 1 discusses issues on information retrieval, and present which issues are to be solved. That is, it examines the question logs from a real-world natural-language-based text retrieval system, and organizes types and factors of the gaps. The examination indicates that some gaps between user questions and texts cannot be resolved well by methods used in previous studies, and suggests that both interactions with users and applicability to real-world text collections are needed. Based on the discussion, a solution to deal with these gaps is proposed, by advancing an approach employed in open-domain question-answering systems, i.e., utilization of recent NLP techniques, into resolving the various gaps. Chapter 2 proposes several methods of matching user questions with texts, based on the NLP techniques. Of these techniques, sentence structure analysis through fullparsing is essential for two reasons: first, it enables expression gaps to be resolved beyond the keyword level; second, it is indispensable in resolving vagueness gaps by providing ask-backs. Our methods include: sentence structure analysis using a Japanese parser KNP, expression-gap resolution based on two kinds of dictionaries, text-collection selection through question-type estimates, and score calculations based on sentence structures. An experimental evaluation on testsets shows significant improvements of performance by our methods. Chapter 3 proposes a novel method of processing metonymy, as an extension of the matching methods proposed in Chapter 2. Metonymy is a figure of speech in which the name of one thing is substituted for that of something else to which it is related, and this frequently occurs in both user questions and texts. Namely, this chapter addresses the automatic acquisition of pairs of metonymic expressions and their interpretative expressions from large corpora, and applies the acquired pairs to resolving structural gaps caused by metonymy. Unlike previous studies on metonymy, the method targets both recognition and interpretation process of metonymy. The method acquired 1, 126 pairs from corpora, and over 80% of the pairs were correct as interpretations of metonymy. Furthermore, an experimental evaluation on the testsets demonstrated that introducing the acquired pairs significantly improves matching. Chapter 4 presents a strategy of navigating users from vague questions to specific texts based on the previously discussed matching methods. Of course, it is necessary to make some use of ask-backs to achieve this, and this strategy involves two approaches: description extraction as a bottom-up approach, and dialog cards as a top-down approach. The former extracts the neighborhoods of the part that matches the user question in each text through matching methods. Such neighborhoods are mostly suitable for ask-backs that clarify vague user questions. However, if a user’s question is too vague, this approach often fails. The latter covers vague questions based on the know-how of the call center; dialog cards systematize procedures for ask-backs to clarify frequently asked questions that are vague. Matching methods are also applied to match user questions with the cards. Finally, a comparison of the approaches with those used in other related work demonstrates the novelty of the approaches. Chapter 5 describes the architecture for Dialog Navigator, a dialog system in which all the proposed methods are implemented. The system uses the real-world large text collections provided by Microsoft Corporation, and it has been open to the public on a website from April 2002. The methods were evaluated based on the real-world operational results of the system, because the various gaps to be resolved should reflect those in the real-world. The evaluation proved the effectiveness of the methods: more than 70% of all user questions were answered with relevant texts, the behaviors of both users and the system were reasonable with most dialogs, and most of the extracted descriptions for ask-backs were suitably matched. Chapter 6 concludes the thesis.
Kyoto University (京都大学)
0048
新制・課程博士
博士(情報学)
甲第11209号
情博第135号
新制||情||31(附属図書館)
UT51-2004-T178
京都大学大学院情報学研究科知能情報学専攻
(主査)教授 松山 隆司, 教授 河原 達也, 助教授 佐藤 理史
学位規則第4条第1項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・課程博士
博士(情報学)
甲第9052号
情博第23号
新制||情||6(附属図書館)
UT51-2001-F382
京都大学大学院情報学研究科社会情報学専攻
(主査)教授 上林 彌彦, 教授 石田 亨, 教授 茨木 俊秀
学位規則第4条第1項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・課程博士
博士(情報学)
甲第9058号
情博第29号
新制||情||7(附属図書館)
UT51-2001-F388
京都大学大学院情報学研究科通信情報システム専攻
(主査)教授 小野寺 秀俊, 教授 吉田 進, 教授 佐藤 亨
学位規則第4条第1項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・課程博士
博士(情報学)
甲第9064号
情博第35号
新制||情||8(附属図書館)
UT51-2001-F394
京都大学大学院情報学研究科社会情報学専攻
(主査)教授 石田 亨, 教授 林 春男, 教授 酒井 徹朗
学位規則第4条第1項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・課程博士
博士(情報学)
甲第9067号
情博第38号
新制||情||8(附属図書館)
UT51-2001-F397
京都大学大学院情報学研究科数理工学専攻
(主査)教授 宗像 豊哲, 教授 藤坂 博一, 講師 青柳 富誌生
学位規則第4条第1項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・論文博士
博士(情報学)
乙第10698号
論情博第16号
新制||情||9(附属図書館)
UT51-2001-F496
(主査)教授 池田 克夫, 教授 佐藤 雅彦, 教授 美濃 導彦
学位規則第4条第2項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・課程博士
博士(情報学)
甲第8603号
情博第21号
新制||情||4(附属図書館)
UT51-2000-P656
京都大学大学院情報学研究科知能情報学専攻
(主査)教授 美濃 導彦, 教授 石田 亨, 教授 佐藤 雅彦
学位規則第4条第1項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・論文博士
博士(情報学)
乙第10519号
論情博第6号
新制||情||4(附属図書館)
UT51-2000-P686
(主査)教授 山本 裕, 教授 磯 祐介, 教授 片山 徹
学位規則第4条第2項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・論文博士
博士(情報学)
乙第10600号
論情博第12号
新制||情||6(附属図書館)
UT51-2001-A792
(主査)教授 茨木 俊秀, 教授 高橋 豊, 助教授 滝根 哲哉
学位規則第4条第2項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・論文博士
博士(情報学)
乙第10601号
論情博第13号
新制||情||6(附属図書館)
UT51-2001-A793
(主査)教授 石田 亨, 教授 酒井 徹朗, 教授 美濃 導彦
学位規則第4条第2項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・課程博士
博士(情報学)
甲第8480号
情博第6号
新制||情||1(附属図書館)
UT51-2000-F384
京都大学大学院情報学研究科知能情報学専攻
(主査)教授 乾 敏郎, 教授 池田 克夫, 教授 松山 隆司
学位規則第4条第1項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・課程博士
博士(情報学)
甲第8482号
情博第8号
新制||情||1(附属図書館)
UT51-2000-F386
京都大学大学院情報学研究科数理工学専攻
(主査)教授 宗像 豊哲, 教授 酒井 英昭, 教授 藤坂 博一
学位規則第4条第1項該当

本文データは平成22年度国立国会図書館の学位論文(博士)のデジタル化実施により作成された画像ファイルを基にpdf変換したものである
Kyoto University (京都大学)
0048
新制・課程博士
博士(情報学)
甲第8486号
情博第12号
新制||情||2(附属図書館)
UT51-2000-F390
京都大学大学院情報学研究科知能情報学専攻
(主査)教授 乾 敏郎, 教授 小林 茂夫, 教授 松山 隆司
学位規則第4条第1項該当