研究者詳細 - 門田　暁人

Program Encryption Based on the Execution Time

Hideshi Sakaguchi, Yuichiro Kanzaki, Akito Monden

Proc. International Symposium on Technology for Sustainability (ISTS2012) 188 - 191 2012年

A Heuristic Rule Reduction Approach to Software Fault-proneness Prediction

Akito Monden, Jacky Keung, Shuji Morisaki, Yasutaka Kamei, Ken-ichi Matsumoto

2012 19TH ASIA-PACIFIC SOFTWARE ENGINEERING CONFERENCE (APSEC), VOL 1 838 - 847 2012年

記述言語：英語出版者・発行元：IEEE COMPUTER SOC

Background: Association rules are more comprehensive and understandable than fault-prone module predictors (such as logistic regression model, random forest and support vector machine). One of the challenges is that there are usually too many similar rules to be extracted by the rule mining.
Aim: This paper proposes a rule reduction technique that can eliminate complex (long) and/or similar rules without sacrificing the prediction performance as much as possible.
Method: The notion of the method is to removing long and similar rules unless their confidence level as a heuristic is high enough than shorter rules. For example, it starts with selecting rules with shortest length (length=1), and then it continues through the 2nd shortest rules selection (length=2) based on the current confidence level, this process is repeated on the selection for longer rules until no rules are worth included.
Result: An empirical experiment has been conducted with the Mylyn and Eclipse PDE datasets. The result of the Mylyn dataset showed the proposed method was able to reduce the number of rules from 1347 down to 13, while the delta of the prediction performance was only .015 (from .757 down to .742) in terms of the F1 prediction criteria. In the experiment with Eclipsed PDE dataset, the proposed method reduced the number of rules from 398 to 12, while the prediction performance even improved (from .426 to .441.)
Conclusion: The novel technique introduced resolves the rule explosion problem in association rule mining for software proneness prediction, which is significant and provides better understanding of the causes of faulty modules.

DOI： 10.1109/APSEC.2012.103

Incorporating Expert Judgment into Regression Models of Software Effort Estimation

Masateru Tsunoda, Akito Monden, Jacky Keung, Kenichi Matsumoto

2012 19TH ASIA-PACIFIC SOFTWARE ENGINEERING CONFERENCE (APSEC), VOL 1 374 - 379 2012年

記述言語：英語出版者・発行元：IEEE COMPUTER SOC

One of the common problems in building an effort estimation model is that not all the effort factors are suitable as predictor variables. As a supplement of missing information in estimation models, this paper explores the project manager's knowledge about the target project. We assume that the experts can judge the target project's productivity level based on his/her own expert knowledge about the project. We also assume that this judgment can be further improved, because using the expert's judgment solely could incur subjective perception. This paper proposes a regression model building/selection method to address this challenge. In the proposed method, a fit dataset for model building is divided into two or three subsets by project productivity, and an estimation model is built on each data subset. The expert judges the productivity level of the target project and selects one of the models to be used. In the experiment, we used three datasets to evaluate the produced effort estimation models. In the experiment, we adjusted the error rate of the judgment and analyzed the relationship between the error rate and the estimation accuracy. As a result, the judgment-incorporating models produced significantly higher estimation accuracy than the conventional linear regression model, where the expert's error rate is less than 37%.

DOI： 10.1109/APSEC.2012.58

An Investigation on Software Bug-Fix Prediction for Open Source Software Projects-A Case Study on the Eclipse Project

Akinori Ihara, Yasutaka Kamei, Akito Monden, Masao Ohira, Jacky Wai Keung, Naoyasu Ubayashi, Ken-ichi Matsumoto

2012 19TH ASIA-PACIFIC SOFTWARE ENGINEERING CONFERENCE WORKSHOPS (APSECW), VOL. 2 2 112 - 119 2012年

記述言語：英語出版者・発行元：IEEE COMPUTER SOC

Open source software projects (OSS) receive a large number of bug reports from various contributors and developers alike, where many planned to be fixed by OSS developers. Given the next release cycle information, OSS users can be more effective and flexible in planning and to fix the bugs that are not to be fixed in the next release. It is therefore vital for OSS users to learn which bugs the OSS developers will fix, unfortunately such information may not be readily available, nor there is a prediction framework exists to serve such an important purpose. In this study, we would like to answer the question "Will this bug be fixed by the next release?", this is addressed by building a bug fixing prediction model based on the characteristics of a bug-related metric and by incorporating the progress of bug fixing measures such as status, period and developer metrics to provide aggregated information for the OSS users. The proposed model calculates the deviance of each variable to analyze the most important metrics, and it has been experimented using a case study with Eclipse platform. Result shows a bug fixing prediction model using both base metrics and state metrics provide significantly better performance in precision (139%) and recall (114%) than the standard model using only base metrics.

DOI： 10.1109/APSEC.2012.86

生産性に基づくソフトウェア開発工数予測モデル

門田暁人, 松本健一, 大岩佐和子, 押野智樹

経済調査研究レビュー 11 ( 11 ) 32 - 37 2012年

Handling Categorical Variables in Effort Estimation

Masateru Tsunoda, Sousuke Amasaki, Akito Monden

PROCEEDINGS OF THE ACM-IEEE INTERNATIONAL SYMPOSIUM ON EMPIRICAL SOFTWARE ENGINEERING AND MEASUREMENT (ESEM'12) 99 - 102 2012年

記述言語：英語出版者・発行元：IEEE

Background: Accurate effort estimation is the basis of the software development project management. The linear regression model is one of the widely-used methods for the purpose. A dataset used to build a model often includes categorical variables denoting such as programming languages. Categorical variables are usually handled with two methods: the stratification and dummy variables. Those methods have a positive effect on accuracy but have shortcomings. The other handing method, the interaction and the hierarchical linear model (HLM), might be able to compensate for them. However, the two methods have not been examined in the research area. Aim: giving useful suggestions for handling categorical variables with the stratification, transforming dummy variables, the interaction, or HLM, when building an estimation model. Method: We built estimation models with the four handling methods on ISBSG, NASA, and Desharnais datasets, and compared accuracy of the methods with each other. Results: The most effective method was different for datasets, and the difference was statistically significant on both mean balanced relative error (MBRE) and mean magnitude of relative error (MMRE). The interaction and HLM were effective in a certain case. Conclusions: The stratification and transforming dummy variables should be tried at least, for obtaining an accurate model. In addition, we suggest that the application of the interaction and HLM should be considered when building the estimation model.

DOI： 10.1145/2372251.2372267

Analysis of attributes relating to custom software price

Masateru Tsunoda, Akito Monden, Kenichi Matsumoto, Sawako Ohiwa, Tomoki Oshino

Proceedings - 2012 4th International Workshop on Empirical Software Engineering in Practice, IWESEP 2012 16 - 22 2012年

記述言語：英語

Price of custom software is very important for a user (end user company). However, there is very little information which helps the user judge the validity of the custom software price. So, one of our research goals is building price estimation model and showing its accuracy for the user to judge the validity of the custom software price. The other goal is how to get value for money custom software. In the analysis, we used 31 projects collected from Japanese organizations. First, we analyzed relationships of unit price of effort, unit price of FP (function point), and productivity. The analysis result showed productivity is more important variable than unit price of effort for the custom software price estimation. Next, relationships of other variables were analyzed to identify important variables for the price estimation. The result suggested some variables such as system architecture are important for that. At the end, the price was estimated based on FP and effort. In the analysis, when the price was estimated based on FP, the median balanced relative error (BRE) was 86.6%, and when it was estimated based on effort, median BRE was 20.2%. © 2012 IEEE.

DOI： 10.1109/IWESEP.2012.19

Scopus

An ensemble approach of simple regression models to cross-project fault prediction

Satoshi Uchigaki, Shinji Uchida, Koji Toda, Akito Monden

Proceedings - 13th ACIS International Conference on Software Engineering, Artificial Intelligence, Networking, and Parallel/Distributed Computing, SNPD 2012 476 - 481 2012年

記述言語：英語

In software development, prediction of fault-prone modules is an important challenge for effective software testing. However, high prediction accuracy may not be achieved in cross-project prediction, since there is a large difference in distribution of predictor variables between the base project and the target project.@In this paper we propose an prediction technique called gan ensemble of simple regression modelsh to improve the prediction accuracy of cross-project prediction. The proposed method uses weighted sum of outputs of simple logistic regression models to improve the generalization ability of logistic models. To evaluate the performance of the proposed method, we conducted cross-project prediction using datasets of projects from NASA IV&amp
V Facility Metrics Data Program. As a result, the proposed method outperformed conventional logistic regression models in terms of AUC of the Alberg diagram. © 2012 IEEE.

DOI： 10.1109/SNPD.2012.34

Scopus

Fault-proneモジュール判別のための相関ルールの絞り込み

ソフトウェア開発プロジェクトをまたがるfault-proneモジュール判別の試み - 18プロジェクトの実験から得た教訓

藏本達也, 亀井靖高, 門田暁人, 松本健一

電子情報通信学会論文誌D J95-D ( 3 ) 425 - 436 2012年

類似性に基づくソフトウェア開発工数見積もりにおける外れ値除去法の比較

角田雅照, 門田暁人, 渡邊瑞穂, 柿元健, 松本健一

電子情報通信学会論文誌 D J95-D ( 4 ) 895 - 908 2012年

An Empirical Evaluation of Outlier Deletion Methods for Analogy-Based Cost Estimation 査読

Masateru Tsunoda, Akito Monden, Takeshi Kakimoto, Ken-ichi Matsumoto

In Proceedings of International Conference on Predictive Models in Software Engineering 17 1 - 10 2011年9月

西川朋希, 門田暁人, 森崎修司, 松本健一

研究報告ソフトウェア工学（SE） 2011 ( 20 ) 1 - 6 2011年3月

記述言語：日本語出版者・発行元：情報処理学会

ソフトウェア開発において，限られた開発期間で十分な品質を確保することの重要性が高まってきている．近年，モデルベース手法である線形判別分析やロジスティック回帰分析を始めとするfault-proneモジュール（バグを含む確率の高いモジュール）判別手法が提案されている．しかし，モデルベース手法は，数式を見ても理解しにくいため，開発現場に受け入れられにくい．そこで本稿では，ルールベース手法である相関ルール分析を用いたfault-proneモジュール判別に着目する．相関ルール分析で作成されるルールは，バグを含む条件が明確なので，現場に受け入れられやすいという利点がある．ただし，ルールが大量に生成されすぎてしまい，どのルールに着目すればいいのか分からないという点が問題である．そこで，本稿では，相関ルールを絞り込むアルゴリズムを提案し，その効果を実験により評価する．While model based software fault predictors, e.g. logistic regression model, Random Forest and Support Vector Machine, are too complex to understand the causes of faults, association rules are much more understandable since rules are described in a simple intuitive form (cond. ⇒ faulty or cond. ⇒ not faulty). However, although each association rule is enough comprehensive, usually too many (similar) rules are extracted by the association rule mining. This paper proposes a rule reduction method that can eliminate complex (long) rules as much as possible without reducing the prediction performance.

デスクワークにおけるアウェアネス向上のための作業計測

その他リンク： http://id.nii.ac.jp/1001/00074116/

D-3-4 ランダムフォレスト法によるソフトウェア開発工数予測の試み(D-3.ソフトウェアサイエンス,一般セッション)

小西文章, 内田眞司, 戸田航史, 門田暁人

電子情報通信学会総合大会講演論文集 2011 ( 1 ) 2011年2月

Fault-prone モジュール判別モデルを用いた検証項目数削減に関する一検討

林卓磨, 信田祥司, 白井久美子, 吉田順一, 門田暁人, 松本健一

2011年電子情報通信学会ソサイエティ大会通信講演論文集 2 406 - 406 2011年

Applying Outlier Deletion to Analosgy Based Cost Estimation 査読

Masateru Tsunoda, Akito Monden, Mizuho Watanabe, Takeshi Kakimoto, Ken-ichi Matsumoto

In Proceedings of International Workshop on Empirical Software Engineering in Practice 13 - 18 2010年12月

下村洋介, 藏本達也, 門田暁人, 松本健一

研究報告ソフトウェア工学（SE） 2010 ( 1 ) 1 - 8 2010年3月

記述言語：日本語出版者・発行元：情報処理学会

デスクワークにおける生産性を向上させるためには，作業の実態を把握し，作業の無駄や改善点に気づくこと（アウェアネスの向上）が重要と考えられる．本研究では，デスクワークのうち情報科学分野の研究活動を対象として，パソコン上の作業を自動計測することでアウェアネスが得られることを実験的に確かめることを目的とする．実験の実施にあたって，研究活動を 11 のタスクに分類し，各タスクとパソコン上のアプリケーションとのひも付けを行うことで，タスク単位での作業時間の自動計測を可能とした．実験では，3 人の被験者が 7 週間にわたって実施した研究活動を計測した．そして，被験者が推定した 1 営業日当たりの各タスクの作業時間と，計測された作業時間とを比較した．また，インタビューにより，推定結果と計測結果に対するコメントを収集した．実験の結果，推定した作業時間と計測された作業時間との間には各被験者でそれぞれ平均 28%，62%，86% の誤差が見られ，作業の実態のアウェアネスを向上するために計測が必要であることが確認できた．また，インタビューの結果から，計測結果を参照することで作業の改善点を発見できることが分かった．To increase productivity of any deskwork, it is important to increase "awareness" of working (e.g. knowing the actual time spent for each task) so that ideas to improve work processes can be discovered. This thesis focuses on "research" in information science area as a deskwork, and empirically evaluate the effectiveness of automatical measurement of research tasks on computers. Before conducting an experiment, this thesis classfied the research activity into 11 tasks, and connected them to software applications so that automatical measurement of tasks becomes feasible. In the experiment, three subjects conducted research tasks for 7 weeks. After measuring the time spent for each task, we collected estimated time for each task by the subjects. We also conducted interviews to subjects to collect comments on measured time and estimated time. The result showed that average difference between measured time and estimated time of each subject was 28, 62 and 86 percent respectively. This indicates the necessity of automatical measurement to increase the awareness of working. Also, the result of interviews showed that measured time and estimated time are useful to discover the ideas to improve work processes.

コーディングスタイルの特徴量とソースコード盗用との関係の分析

その他リンク： http://id.nii.ac.jp/1001/00068312/

武田隆之, 牛窓朋義, 山内寛己, 門田暁人, 松本健一

研究報告ソフトウェア工学（SE） 2010 ( 8 ) 1 - 8 2010年3月

記述言語：日本語出版者・発行元：情報処理学会

本稿では，学生の演習課題のような小規模なソースコードを対象とした盗用の検出を目的とする．インデント，演算子などのコーディングスタイルに着目し，59 項目の特徴量として抽出し，盗用の発見に用いる．盗用関係にあるソースコード間において 59 項目の特徴量の差分を測定し，盗用関係にないソースコード間における特徴量の差分と比較したところ，28 項目の特徴量が盗用検出に有効であること，28 項目のうち 8 項目の特徴量はプログラムの内容によらず盗用検出に有効であること，8 項目のうち 3 項目の特徴量はソースコード整形ツールによるインデント整形に対して耐性を持つことが分かった．The goal of this paper is to detect software plagiarism in small-size source code like exercise assignments at school. This paper focused on coding style elements, such as indents and operators, and computed 59 quantitative measures from these elements. To evalute the usefulness of measures for plagiarism detection, we compared measures of suspected pairs (of plagiarism) and nonsuspected pairs. As a result, we found that 28 measures were effective to detect plagiarisms. Especially, 8 of 28 measures were effective for different program specifications, and 3 out of 8 measures were effective even after source code indentation tools were applied.

その他リンク： http://id.nii.ac.jp/1001/00068319/

データ分布の偏りがコスト超過判別予測に及ぼす影響

柴田淳一郎, 角田雅照, 門田暁人, 松本健一

研究報告ソフトウェア工学（SE） 2010 ( 36 ) 1 - 6 2010年3月

記述言語：日本語出版者・発行元：情報処理学会

プロジェクトのコスト超過を防ぐためには，コスト超過が発生しそうなプロジェクトを早期に発見し，対応策を実施する必要がある．コスト超過プロジェクトの発見には，線形判別分析，ロジスティック回帰，マハラノビス・タグチ法などの判別予測法が用いられる．ソフトウェアプロセスが改善されると，データセットに占めるコスト超過プロジェクトとコスト非超過プロジェクトの割合が変化し，割合がアンバランスとなり，これらの判別予測法の精度に影響を与える可能性がある．本論文では，コスト超過プロジェクトが減少し，コスト非超過プロジェクトの割合が増加した場合における，各種判別予測方法の精度変化を分析する．実験を行った結果，コスト超過プロジェクトとコスト非超過プロジェクトの割合に差が少ない場合は，線形判別分析が適しているが，コスト超過プロジェクトが少ない（35% 以下）場合，マハラノビス・タグチ法が予測に最も適しているといえた．To prevent cost overrun of software projects, it is necessary that project managers discriminate projects which has risk of cost overrun early, and they perform countermeasures. Linear discriminant analysis, logistic regression analysis, and Mahalanobis-Tagichi method are used to find cost overrun projects. When software process is improved, balance of cost overrun projects and cost non-overrun projects become unbalance on dataset, and it may affects accuracy of discriminant prediction. In this paper, we analysis accuracy of various discriminant prediction methods, changing ratio of cost overrun projects and cost non-overrun projects. As a result, when cost overrun projects and cost non-overrun projects are balanced, linear discriminant analysis showed highest accuracy. Also, Mahalanobis-Tagichi method showed highest accuracy when ratio of cost overrun projects is smaller than 35%.