Ayako Hiramatsu

This paper is a survey report about how users accept new social networking services (SNS). Questionnaires based on Technology Acceptance Model (TAM) with some extended factors were given to students for three years. With these questionnaire results, the changes of users' intention to use SNS are described, influences between factors in TAM are compared by structural equation modeling (SEM).

This paper shows an analysis why students use online video hosting services. Based on the Technology Acceptance Model (TAM), a hypothesis model for video hosting website users was designed. The hypothesis model incorporates social influence, flow experience, comfortable communication, and advertisement interference into the basic TAM. A questionnaire was given to about 350 students. Behaviors of online video hosting website users were analyzed with correlation analysis and structural equation modeling. © Springer-Verlag Berlin Heidelberg 2013.

The presentation of travel time information to drivers is one of the important roles of information systems concerned with road traffic. This paper proposes it method for predicting the travel time in urban districts using travel time data front the small sections which make up a route. The forecasting model consists of past time series data oil the travel times in the small sections. The travel time for the route in the near future and its accuracy are estimated by using a state equation of the forecasting model. The validity of the method is verified using real data. Validation showed that the prediction error was approximately 2.5 minutes. Generally, for 50 minutes of travel time, it is desirable that the prediction error be within 5 minutes. Therefore, this result appears satisfactory for practical use. (C) 2009 Wiley Periodicals, Inc. Electron Comm Jpn, 92(7): 1-11, 2009; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecj.10089

This paper describes a consumer behavior model for online shopping, especially online music services, because they are the most Popular online shopping service in Japan. Based on Howard's consumer decision model, questionnaires about decision making for online music services were given to 282 Students. The questionnaire results show that almost 90% of respondents have downloaded from online Music services, and high school students use Such services slightly more than university students. This paper analyzes the questionnaire results by structure equation modeling and examines the relationships between the factors in Howard's model.

The presentation of the travel time information to drivers is one of the important roles of the information system concerned with road traffic. This paper proposes a prediction method of the travel time in urban district using the travel time data of the small section which constitutes a route. The forecasting model consists of past time series data about the travel time of the small section. The travel time and accuracy of a near future of the route are estimated using State Equation of the forecasting model. The validity of the method is verified using real data. As the results of examinations, the prediction error became almost 2.5 minutes. Generally, to fifty minutes of travel time, it is desirable that prediction error is within five minutes. Therefore, it is conceivable that this result is fit to sufficient practical use. © 2008 The Institute of Electrical Engineers of Japan.

This paper describes a survey of consumer behavior for online shopping, especially online music services, because they are the most popular online shopping service in Japan. Based on Howard's consumer decision model, questionnaires about decision making for online music services were given to 282 students. The questionnaire results show that almost 90% of these students have downloaded from online music services, and high school students use such services slightly more than university students.

Various guidelines exist for explaining product information in e-shopping web sites that have rapidly increased in recent years. However, in Japan, it remains doubtful whether they are actually helpful. This survey report shows what kind of explanations about gourmet products are presented in Japanese retail e-shopping web sites.

本研究では、モバイルコンテンツ配信におけるユーザ数の予測を行う。ユーザの退会率は、継続月数、ステージ、ランキングなどユーザの属性により統計的に決定できる。しかし、月ごとにユーザ属性が流動的に変化してしまうため、ユーザ属性の動的変化に対応した予測方法が必要となる。ユーザの属性を元にユーザをセグメント分けし、各セグメントにおける退会の傾向に従って、各ユーザの退会率を定める。ユーザ属性の流動性に対応するために、各セグメントの退会率を各ユーザに割り当て、その退会率を用いて時間軸に沿って確率的にユーザの退会を決定する。

本研究では、ニューラルネットワークを利用した環境音の識別方法を提案する。ニューラルネットワークでは試行錯誤的なパラメータ調整が必要となる。そこで、パラメータ調整の負荷軽減のために、2つの集団学習方法（Bagging, AdaBoost）の適用を試みた。この2つの集団学習方法と、単純なニューラルネットワークとの比較実験を行い、その有効性を検証した。

本論文ではモバイルコンテンツを対象としたプロモーション案比較検討のためのシミュレーションツールを提案する。改善案はすべてのユーザにとってプラスに働くことはない。そこで、本ツールでは改善案ごとに、どのようなユーザにとって有利な案であるかをシミュレーションに取り入れために、属性ごとに退会率の変化予測値を入力し、数ヵ月後までのユーザ数の変化を算出する。本ツールを実際の改善案比較に利用し、その効果を確認した。

モバイルコンテンツを対象とし、ユーザの契約月数長期化を目指した改善方法の評価をする。ユーザに対して行ったアンケート結果をもとに、構造方程式モデリング（SEM）によりユーザの退会意思に影響を与える要因を明らかにしたところ、インセンティブ提供方式が最も退会意思との因果関係が深いことがわかった。そこで、新しいインセンティブ提供方式の一つとしてポイント制を取り上げ、その効果を検討するため、ポイント制に対するアンケートを実施、SEMにより分析した。

本研究では、携帯電話などから入力された日本語の構文を成していない非文法的な文を多く含む自由回答形式アンケートデータから、意外性のある意見を抽出する手法を提案する。非文法的な文を対象とするため、単語単位で文の意図を据えるアプローチをとる。形態素解析を用いて回答文の単語リスト生成し、予め設定した単語の組合せと比較することにより既知の意見であるか判定する。 しかし、単語の組合せが回答文中に含まれているだけでは、その組合せが示す意図を回答文中で正確に表していない可能性がある。また、複数の内容が記入される意見では文章を分割して考える必要があるが、単純に文章を分割できない。そこで、文中の単語の出現順序の差を単語間の距離ととらえ、一定の距離以内に出現した単語の組合せが意味を成すものとし、出現した場所を新たな意味を成す文の始まりとすることで文を分割する。分割されたそれぞれの文を判定し、

本研究では、モバイルコンテンツ退会時にユーザが回答する自由記述形式のアンケート文の分析支援システムを提案する。回答には予想外の回答とよくありがちな回答がある。これらのはそれぞれ分析方法が異なるため、本システムでは分類後、異なるユーザインタフェースで提供する。予想外の回答はすべて読む必要があるため、どのような内容の回答があるか視覚的に訴えるため、キーワードを二次元的に表示する。よくある回答は読む必要がなく、単に何件あるか、また件数の変動どうなっているかを表示するためグラフとして提供する。

This paper presents a method for atypical opinion extraction from answers to open-ended questions supplied by users of a mobile game content when they unsubscribe from the service. The proposed system excludes typical opinions and extracts only atypical opinions. To cope with incomplete syntax of texts due to the input by mobile phone, the system treats the opinions as the sets of keywords. The combination of words are established beforehand in a typical word database. Based on the ratio of typical word combinations in sentences of an opinion, the system classifies the opinion typical or atypical. When typical word combinations are sought in an opinion, the system considers the word order and the distance of difference between the positions of words to exclude unnecessary combinations. Furthermore, when an opinion includes meanings the system divides the opinion into phrases at each typical word combination. With an experiment, the extraction accuracy of the proposed system was confirmed.

Retention of customers is a critical challenge for mobile content providers. This paper presents a customer analysis result aiming at prolonging users' subscription. Based on users' score information, the access log, and so on, customers who are likely to unsubscribe in near future are identified. Regularity is discovered from the unsubscribed users' data using the C4.5 algorithm and a module library is prepared to easily formulate the character sets for the prediction from the users' access log. By using various character sets, the prediction of the users' subscription period was tested by using 6,000 real users' data. We predicted whether a user would unsubscribe within two months, with results showing 72% recall and 65% precision.

This paper proposes a support system for analyzing answers to open-ended questions supplied by users of the mobile game content reply when they unsubscribe from the services. Because useful, unexpected opinions (the atypical opinions) are only 5% of all answers, it is inefficient to read them all to find a useful one. Therefore, we propose a support system for analysis of unsubscribing questionnaire data. The main function of the support system excludes typical opinions and extracts only atypical opinions. The combination of words is established beforehand in a typical word database. Based on the ratio of opical word combinations in the sentences of an opinion, the system classifies the opinion opical or atypical.

This paper proposes a support system for the information extraction from BBS (Bulletin Board System) On BBS in EC (Electronic Commerce) sites, participants who are consumers write frank opinions easily. This produces active conversation as enormous text data and includes unexpected opinions and consumers' requirements. In this system, opinions on BBS are classified by topics and arranged on a 3-dimensional space so that users (for example product designers) can easily understand how topics are interwoven into conversations or how topics are related to each other. Furthermore, two methods using the feature of the conversation structure on BBS are proposed for correcting the recognition of opinions. © 2003 by Springer Science+Business Media New York.

This paper proposes a support system for the information extraction from BBS (Bulletin Board System). On BBS in EC (Electronic Commerce) sites, participants who are consumers write frank opinions easily. This produces active conversation as enormous text data and includes unexpected opinions and consumers' requirements. In this system, opinions on BBS are classified by topics and arranged on a 3-dimensional space so that users (for example product designers) can easily understand how topics are interwoven into conversations or how topics are related to each other. Furthermore, two methods using the feature of the conversation structure on BBS are proposed for correcting the recognition of opinions.

We have developed a location-aware information service system using Bluetooth in an exhibition hall like a gallery and a museum. With this system, the visitors who have their mobile devices such as a cellular phone or PDA can automatically get the information about the exhibits nearby them, based on the estimated visitors' locations. In this paper, we propose several operation support features for smoothly maintaining contents information and the Bluetooth communication area setup. The proposed operation support was evaluated with time to input. With the support, the operator took 183 min for data input and it was one-third of input time without the support system.

In this paper, the numerical solution method of an on-ramp traffic control problem is discussed from the viewpoint of the application of evolutionary computation. The problem is formulated as a discrete-time nonlinear optimal control problem. The optimal control problem is then transferred to an optimization problem with successive constraints. A parallel-adaptive genetic algorithm is proposed to obtain the numerical solution of the problem. A real-time practical control method is also proposed. Numerical simulations for assessing the performance of the proposed methods are performed using observed data of Hanshin Expressway. As a result, the proposed method obtains the solution at comparatively small generations. This means shortening the calculation time, and is useful for the real-time control. Moreover, it is shown that the proposed real-time control is effective to reduce heavy congestion on the expressway.

This paper presents an information compressing 3-dimensional user interface on the support system of requirement extraction from BBS (Bulletin Board System). To correspond to users' viewpoints, the Fish-Eye is applied. With the Fish-Eye, a size of an object on the output screen is decided according to user's attention. This function is realized in topic groups, opinion cards and text in cards respectively. A way of applying Fish-Eye in regards to the topic groups and opinion cards is to show the objects on the inclined coordinates from viewpoints. To realize the Fish-Eye for letters, the priority is set on each sentence in the opinion. In this paper, the priority is decided from the character of conversational structure, that has a tree structure. With this structure, the attention of the user is specified and is reflect in selecting which sentences should be written in the opinion cards.

This paper proposes a new type of cooperative virtual mall system in which the customers can join the interested groups and can propose prodlucts through discussions with bulletin board system，chat, or e-mail．This system is based on acooperative mall’s basic concept：“Consumer To Business”．Only members can access the site of this system for shopping and joining auctions．Furthermore，the members can join various groups and propose products that they desired through virtual group meetings．The ICtag store a group home page URL，which is used as an invitation tool for the group．The ICtags are sent with purchased products or other catalogs．The members can simply access from any PC to the intended page using the ICtag．

This paper provides a method that calculates the optimal control of on-ramp traffic density from the basis of the dynamic estimate of traffic density. The on-ramp traffic control problem is formulated as follows. The problem is being able to find the optimal inflow of each on-ramp, based on the evaluation of an on-ramp and main line traffic density. We consider the on-ramp traffic density as the decision variable, and treat this control problem as the nonlinear optimization problem that maximizes the evaluation function under restrictions. To maximize the evaluation function under the restrictions, we propose an adaptive Genetic Algorithm. The solution candidates for on-ramp traffic density have inequality restrictions. Therefore, we consider gene values as ratio of the inflow traffic density within the limit of inequality restrictions. Moreover, in the proposed method, genetic operations are not performed in series, but they are performed in parallel and are adaptive. As numerical examples, we adapt statistical data (the on-ramp traffic density) and examine whether the proposed method can control the traffic jam.

In regard to the design of client server systems (CSS), rapid progress in technology prevents an individual system engineer (SE) from utilizing effective cases. The numerous combinations of hardware and software and complexity in requirement analysis make it difficult for an SE to acquire CSS design technique and it is therefore necessary to support CSS design. In this paper, we propose a CSS configuration design support system that consists of the following four functions: a conversational requirement analysis function, a case based configuration planning function, a plan analysis-explanation function, and an adaptable knowledge learning function. In particular, plan analysis-explanation gives educational support to SEs for CSS design. This explanation is written in natural scenario language which shows how the proposed CSS is designed and how the designed CSS satisfies requirements. The scenario is generated based on structural models that show the processes of requirement analysis and design assumptions. Effective information is selected from nodes in the models in order to generate concise scenarios. (C) Scripta Technica.

The scenario generator for qualitative simulation systems has been developed for explaining the influence flow to the concerned node in the structural model. This paper presents the method through which companies evaluate business to business electronic commerce (e-commerce) systems by using the scenario generator. The method consists of the following processes: I) The adjustment of the structural model to specific companies 2) The qualitative simulation and the scenario generation of the new systems 3) The evaluation of the scenario. This method will be useful for companies to evaluate e-commerce systems.

In the requirement analysis part of CIDAC(Case based Integrated Design Assist system for CSS) for SE support, inputed ambiguous requirement data are transformed into concrete requirement data which are forwarded to the design generation function. One type of concrete requirement data is a set of function trees. This tree represents business system structure and helps SE to decide a framework for the target computer system. This paper presents a method that can generate function trees using several already generated function trees. In this method, suitable parts that are chosen by the SE from other cases are automatically merged in order to complete the set of target function trees. Adding the partial cases to the foundation case, a set of target function trees can be combined without duplicate nodes.