博士学位論文

(1)

博

士

学

位

論

文

Doctoral Dissertation

内容の要旨及び審査結果の要旨

Dissertation Abstract and

Summary of the Dissertation Review Result

第

33

号

The Thirty-Third Issue

2019

年

3

月

March, 2019

The University of Aizu

(2)

はしがき

博士の学位を授与したので、学位規則（昭和２８年４月１日文部省令第９号）第８条の規定に基づき、その論文の内容の要旨及び論文審査の結果の要旨をここに公表する。

学位記番号に付した｢甲｣は学位規則第４条第１項（いわゆる課程博士）によるものであることを示す。また、「乙」は学位規則第４条第２項（いわゆる論文博士）によるものであることを示す。

Preface

On granting the Doctoral Degree to the individuals mentioned below, abstracts of their theses and the theses review results are herewith publicly announced, in according to the provisions provided for in Article 8 of the Ruling of Degrees (Ministry Of Education Ordinance No.9, enacted on April 1, 1953)

The Chinese character, “甲”, at the beginning of the diploma number represents that an

individual has been granted the degree in accordance with the provisions provided for in

Paragraph 4-1 of the Ruling Of Degrees (what is called “Katei Hakase” or the Doctoral

Degree granted by the University at which the grantee was enrolled.). The Chinese

character,

“乙”, at the beginning of the diploma number represents that an individual has

been granted the degree in accordance with the provisions provided for in Paragraph 4 -2 of

the Ruling Of Degrees (what is called

“Ronpaku”).

(3)

- 1 -

Associate

Prof. JING, L 会津大学教授程子学（主査）

会津大学教授サバシュバーラ会津大学准教授王軍波会津大学准教授荊雷

(5)

- 3 -

Abstract

Big Data is one of the most actively researched field in computer science. Many researchers, companies, research institutions etc., are studying how to analyze and take advantage of it in their fields. One challenge in big data analysis is to observe events of public concern with a good accuracy, because the demand for observations of what happens in the real world through big data in real time is strongly increasing. Although there is a number of social networking services, many researchers consider that Twitter is one of the most suitable data resources for event detection, because it encourages its users to feel free to post tweets frequently by allowing them to start posting tweets only with very simple registration, and by limiting the number of characters in a tweet to 140 characters.

Therefore, there is a number of studies in Twitter-based event detection, but the studies have limitations and a problem. They are limitations of kinds of detectable events, limitations of provided information quantity and a problem of uncertain credibility of Twitter-based event detection results.

This thesis removes the limitations, and then solve the problem. This thesis consists of two research.

The first research is Twitter-based Real-Time Event Detection without Primary Existing Limitations.

To this end, we have to remove both of the two limitations: the limitations of kinds of detectable events and the limitations of provided information quantity. That is because Twitter-based event detection with the first limitations is just pre-defined specific event detection. Additionally, although we can detect every kind of events if the first limitations are removed and information related to events are updated to Twitter, the event detection cannot provide enough amount of information about detected events because amount of information which tweets have is limited. Consequently, Twitter-based event detection is never completed if both limitations are removed. The first limitations are caused by necessity of pre-training and necessity of preparing keyword lists in advance. To remove the limitations, we propose a novel Twitter-based real-time event detection scheme which does not need pre-training and preparing keyword lists in advance. Additionally, Extended Hybrid TF-IDF (EHTF-IDF) and Remarkable Word Detecting Method (RWDM) are also proposed to realize our proposed scheme. EHTF-IDF can calculate how important a word is to a set of tweets in numerical form. Although the degrees are represented in numerical form well by EHTF-IDF, they are no more than numerical values, so that the threshold is required to detect events. RWDM can provide the threshold dynamically. We have performed a comprehensive experiment to evaluate EHTF-IDF, RWDM, and the proposed scheme. The experimental result shows EHTF-IDF strongly improves event detection accuracy, RWDM strongly improves event detection efficiency, and the proposed scheme strongly improves both event detection accuracy and efficiency. The second limitations mean, specifically, Twitter-based event detection systems cannot provide an enough amount of information on detected events. It causes the limitations that the number of characters in a tweet is limited to 140 characters. One possible solution is to retrieve additional information, which is related to a Twitter-based event detection result, from heterogeneous data resources such as articles, Web Pages, blog posts etc. If the related information is retrieved, it can be used to provide as further information to increase provided information quantity drastically. However, properly retrieving related contents from heterogeneous data resources is not easy because of different types of data. To remove the limitations

(6)

- 4 -

of provided information quantity, we propose Additional Information Retrieving Method. This method can retrieve additional information related to a set of tweets, which are related to a detected event, from heterogeneous data resources based on similarity (distance) between them. The similarity is calculated with Normalized Compression Distance (NCD). We have performed a comprehensive experiment to evaluate the proposed method, and the result show that it has high anti-noise capability and performs well even in practical situation.

The second research is Twitter-based Event Detection Result Credibility Evaluation. The first research makes Twitter-based event detection possible to detect every kind of events and provide enough amount of information about detected events. However, the first research cannot give any guarantee of credibility in terms of their detection results as with other Twitter-based event detection systems. We name it the problem of uncertain credibility of Twitter-based event detection results. To solve this problem, rumor detection has been studied in these years to enable credible event detection.

Nevertheless, this problem has not been solved yet because most of existing studies only focus on information on Twitter as with the Twitter-based event detection systems. The existing studies detect rumor by identifying and checking special features of incredible information on Twitter. However, values of the identified features can be faked easily, so that it is important to execute mixing analysis of both twitter and external data obtained from credible data resources to solve this problem. Difficulty, in this approach, is how harmoniously to analyze heterogeneous data together, since they have different types of data format, generating time, and so on. To overcome the difficulty, we propose Twitter-Based Event Detection Result Credibility Evaluating Method. This method evaluates credibility of Twitter-based event detection results based on two heterogeneous data resources, i.e.

tweets, and articles. We have performed a comprehensive experiment with real data, and the experimental result shows that the proposed method can give correctly detected events high credibility and the others low credibility.

Summary of the Dissertation Review Result

The dissertation realizes Twitter-based real-time credible event detection. However, it is not easy, because there are three serious problems which must be solved to realize it: limitations of kinds of detectable events, limitations of provided information quantity, uncertain credibility of Twitter-based event detection results. His research contribution is to solve these problems. Specifically, this dissertation is divided into two researches. In the first one, he proposes Twitter-based real-time event detection scheme, Extended Hybrid TF-IDF, Remarkable Word Detecting Method and Additional Information Retrieving Method to solve the first and second problem, and also he confirms that his proposals work well as the solutions through practical experiments. In the second one, he proposes Twitter-based Event Detection Result Credibility Evaluating Method to solve the third problem, and he confirms that it evaluates credibility of event detection results accurately and can be the solution through practical experiments.

He clearly presented these proposals, experiments, research contribution and the response to

(7)

- 5 -

questions raised after the preliminary review in detail in his doctoral dissertation final review meeting.

After his presentation, we asked questions and discussed about his research actively about 50 minutes.

The committee reviewed the submitted dissertation and the response studiously, and also evaluated whether not only his research, but also his presentation skill, fundamental scholastic aptitude and English language proficiency are enough to obtain a Ph.D. degree. As a result of the review and evaluation, all of the committee members unanimously agreed to confirm the significance of the dissertation for a Ph.D. degree.

(8)

- 6 - Name

氏名

PURGINA, Marina

（プルギナマリナ）

甲CI博第67号

論文題目

Mobile technology for gamification of natural language grammar acquisition

自然言語の文法習得というゲーミフィケーションのためのモバイル技術

The University of Aizu, Associate Prof. MOZGOVOY, M.

(Chief Referee)

The University of Aizu, Prof. KLYUEV, V.

The University of Aizu, Prof. BRINE, J.

The University of Aizu, Associate Prof. PAIK, I.

会津大学准教授マキシムモズゴボイ（主査）

会津大学教授ヴィタリークリュエフ会津大学教授ジョンブライン会津大学教授白寅天

(9)

- 7 -

Abstract

Recent years are marked with rising interest to technologies of gamification, defined as the use of game design elements in non-gaming contexts. The basic premise of gamification is that the principles making computer games attractive can also increase attractiveness of other activities, such as learning.

The interest to gamification technologies is triggered with widespread use of smartphones in general audience, and the growing popularity of casual mobile games, designed for wide range of people.

Therefore, application developers can rely on unprecedented reach of their products and expect acceptance of game-like elements by the users. There is also an active discussion on what exactly constitutes “game-like elements”, suitable for the use in educational applications without harm for their primary educational objectives.

In the present work, we discuss a particular case of gamification of language learning via mobile system WordBricks, created at the University of Aizu. Most present systems of language learning are based on traditional learning activities, such as reading, listening, translating, and solving quizzes.

WordBricks focuses specifically on the task of natural language grammar acquisition, and implements a concept of more user-centric lab-style experimental activities. The primary purpose of WordBricks is to give the users the capability to construct sentences according to predefined grammatical rules, and thus understand the basics of grammar system of a particular natural language. The app is based on the concept of visual grammar formalism, aimed to encode the rules of grammar in intuitive and user friendly way.

WordBricks was evaluated in three different use scenarios: 1) as a learning aid at English language classes for computer science students at the University of Aizu, Japan; 2) as a teacher’s demonstrational tool for the students of the same background; 3) as a supplementary learning material at Irish language classes for junior students of a public school in Dublin, Ireland. Our experiments demonstrate the feasibility of chosen approach on the basis of user feedback and numerical evidence showing that WordBricks can be as efficient as traditional learning materials, but providing more immersion and user enjoyment.

We also explore the possibility to automate the process of authoring WordBricks exercises with natural language processing modules. A significant part of this work includes manual annotation of grammatical attributes of words and word-word relationships, which can be also done with current language processing algorithms. The resulting markup can be corrected if necessary. In addition, automation of text processing allowed us to implement a procedure of converting arbitrary sentences into structured visualizations. This functionality helps students to understand the structure of sentences, not covered in WordBricks exercises.

As a result of experiments, we outline further directions for subsequent improvement of our technology. It includes introducing additional game-like elements, designing more learning materials, and making the application easily customizable by the educators. We also discuss principal difficulties faced by computer-assisted language learning technology experts due to inherent complexity of natural language, and challenging issues for our system.

(10)

- 8 -

Summary of the Dissertation Review Result

The dissertation addresses the problem of gamification of natural language grammar acquisition and proposes a novel and efficient way of using modern natural language processing technologies to create a mobile application that can be used by independent learners and foreign language teachers alike. The author used dependency grammar theory to design a representation of natural language grammar constructions in linearized linked-brick form, and provided a practical implementation of the concept for English and Irish languages. In addition, existing natural language processing modules were used to implement automatic transformation of sentences into brick structures, which helps the learners to understand grammatical constructions. The effectiveness of the proposed approach was confirmed in three independent studies, reflecting distinct use scenarios of the proposed software. The committee considers the contributions of the dissertation significant to the fields of computer science and computer-assisted language education. The goal of developing gamification technologies for natural language learning acquisition is relevant to the challenges of modern education, and addressed properly in the work. The details of the technological side of the contribution are also formulated clearly. The author has demonstrated sufficient ability to perform research activities as well as do practical technological work. Her dissertation is written according to established practices, and follows generally accepted outline. The contribution is relevant to the field, and convincing. The issues raised during the preliminary review were properly addressed in the final version of the dissertation. All members of the committee approved the submitted work.

(11)

- 9 - Name

氏名

THENUWARA HANNADIGE, Akila Sanjaya Siriweera

（テヌワラハンナディゲアキラサンジャヤシリウィラ） The relevant degree

学位の種類

甲CI博第68号

論文題目

Architecture for Intelligent Big Data Analysis based on Automatic Service Composition

自動サービス合成による知的ビックデータ分析のためのアーキテクチャー

(Chief Referee)

The University of Aizu, Prof. ZHAO, Q.

The University of Aizu, Prof. VAZHENIN, A.

The University of Aizu, Prof. NARUSE, K.

会津大学教授白寅天（主査）

会津大学教授趙強福

会津大学教授アレクサンダーヴァジェニン会津大学教授成瀬継太郎

(12)

- 10 -

Abstract

Big data analytics (BDA) is the preferred approach to managing high volumes of highly varied data generated at high velocity (3V-data), which is becoming prevalent in the data sciences. BDA is evolving for many V’s (mV-data) such as high variability and veracity. It is an extreme challenge to store and process the mV-data. Moreover the BDA process which consumes mV-data is raising an extreme concerns. Understanding data, addressing the data quality, dealing with outliers, modeling for analysis and displaying meaningful results are considered as concerns occurred perspective to the data science. This implies, that the BDA process is diverse stepped, heavily resource and time consuming job. Therefore, these two limitations are hampering the meaningful adoption of the BDA across the research and industry domains. Therefore, automating the BDA is a cognitive approach for the industry, which is suffering most. Besides that dramatic expansion of services related to data analytic shows bright prospect in the data science. Then service composition becoming the preferred platform for the BDA.Therefore we proposed a novel architectural design process to automate the BDA process based on automatic service composition (ASC), with the cross-industry standard process for data mining being used as the data science underpinning BDA. Proposed ASC comprised five main stages, which are, planning, discovery, selection, verification-refinement and execution. Each stages are comprised with respective stage-specific one or more combinations of concerns, such as constraint-awareness, NP-complete and domain-specific concerns. Then it is an essential to address respective concerns in adoptable to constraints, heuristic and ready to comply with domain concerns.

Consequently, we proposed dedicated solutions to the each stages of the ASC process, and we employed AI techniques to the concerns, which are encountered during the process. Our experiments demonstrate that the proposed solutions are well behaved and efficiently facilitate to accomplish to satisfy the overall architecture to automate the BDA process

Summary of the Dissertation Review Result

The key contribution of the dissertation is automate the Big Data Analysis (BDA) process based on the automatic service composition. CRISP-DM process used as the data mining process behind BDA.

In addition to that, proposed ASC process consists with main five stages, which are planning, discovery, selection, verification-refinement and execution. The research proposed the entire intelligent Big Data Analysis Architecture based on CRISP-DM and ASC. Contributions at each stage are as follows.

First, in the planning stage, for the problem that data mining process of the BDA is heavily constraints aware, a method to generate the workflow for the analytics considering the constraint awareness of data mining process using Graphplan technique was proposed.

Second, in the discovery stage, the issue that service in the BDA are highly domain and context aware, which cause to reduce the accuracy has been addressed. To satisfy the concerns, a domain context aware service discovery and linked social service network based service discovery was proposed.

(13)

- 11 -

Third, in the selection stage, long processing time and highly resource consuming BDA tasks tend to have highly riskier transactions. To avoid the risk of transaction, a customizable transaction aware selection based on genetic algorithm was proposed.

Forth, in the verification and refinement stage, tasks in the BDA is satisfying by the composite web services (CWS). Existing atomic service based partial order planners loose benefits of the CWS and disrupts the automation, and CWS based partial order planner was proposed.

Finally, in the execution stage, composed verified and refined workflow for the analysis and successfully output the expected BDA result.

For the first issue two major conference papers, for the second and fourth issue one conference paper each, and for the third, one major journal and one major conference paper have been published. The candidate is the first author of all the papers.

In the final review, the candidate successfully improved his presentation and dissertation accordingly to the review committee's comments.

As a conclusion, the candidate has fulfilled all the requirements for the doctoral degree, and the review committee approved it.

(14)

- 12 - Name

氏名

RUPASINGHA ARACHCHILAGE, Hiruni Madhusha Rupasingha

（ルパシンハアラチチラゲヒルニマドゥーシャルパシンハ）

甲CI博第69号

論文題目

Improving Web Service Clustering and Recommendation by Specificity-Aware Ontology Generation

特異性を考慮したオントロジー生成によるウェブサービスクラスタリング及びレコメンデーションの改善

(Chief Referee)

The University of Aizu, Prof. KLYUEV, V.

Senior Associate Prof. TRUONG, C. T.

The University of Aizu, Associate Prof. OFUJI, K.

会津大学教授白寅天（主査）

会津大学教授ヴィタリークリュエフ会津大学上級准教授コンタンチョオン会津大学准教授大藤建太

(15)

- 13 -

Abstract

The rapid development of the Internet in recent years has led to a vast increase in the numbers of Web services, which challenges the process of clustering and users’ capability to find their favorite services quickly and accurately. Clustering Web services based on their functional features to different domains have started to play a major role in several service management tasks such as efficient Web service discovery and recommendations. In this thesis, we present solutions for Web services clustering and recommendations.

In this thesis, first we present a Web service clustering approach that uses novel ontology learning and a similarity calculation method based on the specificity of an ontology in a domain with respect to information theory. Instead of using traditional methods, we generate the ontology using a novel method that considers the specificity and similarity of terms. The specificity of a term describes the amount of domain-specific information contained in that term. Although general terms contain little domain-specific information, specific terms may contain much more domain-related information. The generated ontology is used in the similarity calculations. New logic-based filters are introduced for the similarity-calculation procedure. If similarity calculations using the specified filters fail, then Information-Retrieval (IR)-based methods are applied to the similarity calculations. Finally, an agglomerative clustering algorithm, based on the calculated similarity values, is used for the clustering.

As a second step we propose a recommendation approach. Among the service recommendation algorithms, Collaborative Filtering (CF) gives credence to user inputs by comparing user’s correlations. Although the CF technique is one of the most successful recommendation system technologies, it suffers from data sparsity and cold-start problems, which make the incomplete and inadequate information to analyze a user predicament on Web services. This thesis proposes a CF-based recommendation approach that first alleviates the sparsity problem using a proposed ontology-based clustering. This clustering approach can easily and effectively increase the data density of the user-service dataset by assuming blank user preferences according to the history of user-favored domain(s). Then, we propose a trustbased user rating prediction by determining the trust value between users by calculating the correlation of users. Finally, recommendation was based on these predictions.

We achieved highly efficient and accurate results with this clustering approach than other existing clustering approaches. And the experimental results of recommendation approach indicate that the proposed approach can effectively alleviate the data sparsity and cold-start problems with lower prediction error with the best recommendation performance.

Summary of the Dissertation Review Result

In this research, there are two main tributions: improving performance of (1) Web service clustering and (2) its recommendation using the clustering.

First, the new Web service clustering approach is based on a novel ontology generation method using

(16)

- 14 -

information theory to calculate self-specificity and context specificity. The result shows considerable improvement comparing existing work.

Second, the proposed novel clustering approach is used to overcome the sparsity and cold-start problem by alleviating the sparsity of web service-user ratings and that result is used to purposed a new Web service recommendation approach.

According to the experiments both clustering and recommendation approached show the best performance than existing approaches.

For the first issue two major conference papers and one major journal, for the second two major conference papers have been published. The candidate is the first author of all the papers.

In the final review, the candidate successfully improved his presentation and dissertation accordingly to the review committee's comments.

As a conclusion, the candidate has fulfilled all the requirements for the doctoral degree, and the review committee approved it.

(17)

- 15 - Name

氏名

PHAM VAN, Thanh

（ファムヴァンタイン）

甲CI博第70号

論文題目

Design Framework for Multi-Cell Multi-User Visible Light Communications (VLC) Networks

マルチセル・マルチユーザ可視光通信（VLC）ネットワークのための設計フレームワーク

The University of Aizu, Prof. PHAM, A. (Chief Referee)

The University of Aizu, Prof. TEI, S.

The University of Aizu, Prof. MIYAZAKI, T.

Senior Associate Prof. TRUONG, C. T.

Niigata University, Prof. HAYASHI. T.

会津大学教授ファントゥアンアン（主査）

会津大学教授程子学会津大学教授宮崎敏明

会津大学上級准教授コンタンチョオン新潟大学教授林隆史

(18)

- 16 -

Abstract

With the explosive growth of mobile devices, it is forecasted that the global data traffic from 2016 to 2021 will increase from 7 exabytes to 49 exabytes per month. To address this tremendous demand given the spectrum scarcity problem in radio- frequency (RF) communications, there has been a great deal of interest in research and development of optical wireless communications (OWC), which can be an alternative or complementary to the existing wireless technologies. In particular, for indoor applications, visible light communications (VLC), which is a subset of OWC, has been gaining a lot of attention from both academia and industry over the last decade.

One on hand, due to the broadcast nature of the visible light, VLC can be categorized as broadcast networks, which are capable of serving multiple users. On the other hand, this raises concerns in security and privacy as any users located within the illuminated area can gain accessibility of the transmitted signals making eavesdropping is a possible threat. In addition to security at network and transport layers, security at physical layer (also known as physical layer security) has been emerging as a completely new measure against eavesdropping. In this dissertation, our first objective is to design efficient methods to enhance physical layer security in multi-user VLC systems.

To support even a larger number of users, it is conceivable that the multi- cell configuration is a natural progression in the development of VLC. We introduce in this thesis a concept of multi-cell VLC networks, where each cell composes of multiple light-emitting diode (LED) transmitters to support multiple users by means of precoding techniques. In such multi-cell multi-user networks, the presence of intra- cell and inter-cell interferences, which can severely degrade the system performance, is inevitable due to the overlapping of multiple signals at each user. Therefore, the second objective of the thesis is to design cell coordination/cooperation strategies and their corresponding coordinated/cooperative precoding to alleviate, or possibly, to cancel out the intra-cell and inter-cell interferences. The performances of the proposed strategies in terms of users’

sum-capacity are extensively evaluated and compared.

Summary of the Dissertation Review Result

The dissertation has two main contributions. Firstly, it is the study on the security fundamental limits of MU-MIMO VLC systems within the context of physical layer security. Specifically, it is investigation of the secrecy capacity with respect two different approaches: precoding and artificial noise (AN). As for the precoding approach, to ensure the confidentiality among legitimate users, zero-forcing (ZF) precoding is adopted. Then, optimal ZF precoding designs to maximize users’

secrecy sum-rate are studied for two scenarios: known and unknown eavesdropper’s channel state information (CSI) at the transmitter. To design ZF precoders with respect to user’s secrecy capacity, a tractable formula for the VLC channel capacity should be known. Nevertheless, it is known that the input signal of VLC channels is amplitude constrained, for which no simple expressions for the channel capacity are available. To tackle this issue, the candidate derives closed-form expressions for

(19)

- 17 -

the lower and upper bounds of the VLC channel capacity. It is revealed that at high signal-to-noise ratio (SNR) region (the condition that is usually available in VLC), the lower bound, which is simpler, is sufficient in characterizing the channel capacity due to its negligible gap to the exact one. The optimal ZF precoders are then designed with respect to the lower bound. For the AN approach, the objective to guarantee max-min fairness of user’s signal-to-interference-plus-noise ratios (SINR). In addition to that, the AN design also needs to meet specific requirements depending on the availability of eavesdropper’s CSI at the transmitter. In case the eavesdropper’s CSI is unknown, the AN is designed in such a way that it does not interfere legitimate users but possibly degrades the eavesdropper’s channel. In case the eavesdropper’s CSI is known, the AN design aims to limit the eavesdropper’s SINRs below a predefined threshold.

Secondly, it is the study on multi-cell multi-user VLC networks supporting multiple users by means of precoding techniques. To reduce the impact of intra-cell and inter- cell interferences, the candidate studies three different cell cooperation/coordination strategies and their corresponding cooperative/coordinated precoding designs, namely: per-cell coordinated precoding, coordinated precoding and cooperative precoding with partial data sharing. In the considered multi-cell network, the VLC channels are subject to interferences, which are also amplitude constrained. Similar to the case with only amplitude constraint on the input signal mentioned previously, there are no closed-form expression for the channel capacity of such interference channels. Thus, the candidate derives a lower and upper bound on the channel capacity of a scalar Gaussian interference channel in which both input signal and interference are amplitude constrained. The optimal cooperative/coordinated precoders are then designed based on the derived bounds. Finally, the performances of these strategies are evaluated and compared in terms of users’ sum-capacity.

The candidate has an excellent scholastic aptitude which is reflected in his strong record of publications and achievements. The candidate also has an excellent English command. Both review sessions went smoothly. Having carefully evaluated the submitted dissertation by the candidate, the committee unanimously agrees that the contribution of the dissertation is significant to the field of communication networking. In overall, the candidate is fully qualified for the conferment of PhD degree considering the contributions of the dissertation, his publication record and scholastic ability.

(20)

- 18 -

博士学位論文 Doctoral Dissertation

内容の要旨及び審査結果の要旨 Dissertation Abstract

and

Summary of the Dissertation Review Result

第33号

The Thirty-Third Issue

2019年3月 March, 2019

発行会津大学

〒965-8580 福島県会津若松市一箕町鶴賀 TEL: 0242-37-2600

FAX: 0242-37-2526 THE UNIVERSITY OF AIZU Tsuruga, Ikki-machi Aizu-Wakamatsu City

Fukushima, 965-8580 Japan

博 士 学 位 論 文

Doctoral Dissertation

Dissertation Abstract and

Summary of the Dissertation Review Result

33

The Thirty-Third Issue

2019

3

March, 2019

The University of Aizu

Preface

The Chinese character, “甲”, at the beginning of the diploma number represents that an

individual has been granted the degree in accordance with the provisions provided for in

Paragraph 4-1 of the Ruling Of Degrees (what is called “Katei Hakase” or the Doctoral

Degree granted by the University at which the grantee was enrolled.). The Chinese

character,

been granted the degree in accordance with the provisions provided for in Paragraph 4 -2 of

the Ruling Of Degrees (what is called

Contents

Associate

Abstract

Summary of the Dissertation Review Result

Abstract

Summary of the Dissertation Review Result

Abstract

Summary of the Dissertation Review Result

Abstract

Summary of the Dissertation Review Result

Abstract

Summary of the Dissertation Review Result

博士学位論文