FSE2011-FoundationsofSoftwareEngineering2011年论文集资源-CSDN文库

共37个文件

pdf：37个

软件工程

2011

5星 · 超过95%的资源需积分: 10 156 浏览量 2012-01-19 20:33:02 上传评论收藏 22.3MB RAR 举报

资源推荐

资源详情

资源评论

收起资源包目录

FSE 2011.rar （37个子文件）

Defects

Micro interaction metrics for defect prediction.pdf 599KB

High-impact defects _ a study of breakage and surprise defects.pdf 416KB

BugCache for inspections _ hit or miss.pdf 901KB

Analysis I

Managing performance vs. accuracy trade-offs with loop perforation.pdf 539KB

Checking conformance of a producer and a consumer.pdf 616KB

Proving programs robust.pdf 726KB

Configurations

Version-consistent dynamic reconfiguration of component-based distributed systems.pdf 898KB

Taming uncertainty in self-adaptive software.pdf 830KB

On software component co-installability.pdf 804KB

Models and requirements

Modeling the HTML DOM and browser API in static analysis of JavaScript web applications.pdf 787KB

CSSL_ a logic for specifying conditional scenarios.pdf 647KB

Using an SMT solver for interactive requirements prioritization.pdf 609KB

Empirical studies

The onion patch_ migration in open source ecosystems.pdf 527KB

Effective communication of software development knowledge through community portals.pdf 1.31MB

Does adding manpower also affect quality_ an empirical, longitudinal analysis.pdf 652KB

Bugs and changes

ReLink_ recovering links between bugs and changes.pdf 854KB

Don't touch my code_ examining the effects of ownership on software quality.pdf 777KB

How do fixes become bugs.pdf 665KB

Keynote addresses

ELI-ALPS_ the ultrafast challenges in Hungary.pdf 331KB

Building advanced mechatronic systems.pdf 359KB

Software architecture_ reflections on an evolving discipline.pdf 372KB

Testing

Improved multithreaded unit testing.pdf 523KB

Strong higher order mutation-based test data generation.pdf 566KB

Testing software in age of data privacy _ a balancing act.pdf 901KB

Analysis III

Boosting the performance of flow-sensitive points-to analysis using value flow.pdf 664KB

Inferring data polymorphism in systems code.pdf 415KB

Collaboration

ADDiff _ semantic differencing for activity diagrams.pdf 831KB

Proactive detection of collaboration conflicts.pdf 1.29MB

Semistructured merge _ rethinking merge in revision control systems.pdf 510KB

Mining

On the congruence of modularity and code coupling.pdf 490KB

Fuzzy set and cache-based approach for bug triaging.pdf 805KB

Analysis II

Synthesizing data structure manipulations from storyboards.pdf 977KB

Path exploration based on symbolic output.pdf 492KB

Leveraging existing instrumentation to automatically infer invariant-constrained models.pdf 665KB

Debugging

Fault localization for data-centric programs.pdf 515KB

Partial replay of long-running applications.pdf 610KB

Mitigating the confounding effects of program dependences for effective fault localization.pdf 613KB

Effective Communication of Software Development

Knowledge Through Community Portals

Christoph Treude, Margaret-Anne Storey

Dept. of Computer Science, University of Victoria

ctreude@uvic.ca, mstorey@uvic.ca

ABSTRACT

Knowledge management plays an important role in many

software organizations. Knowledge can be captured and dis-

tributed using a variety of media, including traditional help

ﬁles and manuals, videos, technical articles, wikis, and blogs.

In recent years, web-based community portals have emerged

as an important mechanism for combining various commu-

nication channels. However, there is little advice on how

they can be eﬀectively deployed in a software project.

In this paper, we present a ﬁrst study of a community por-

tal used by a closed source software project. Using grounded

theory, we develop a model that characterizes documen-

tation artifacts along several dimensions, such as content

type, intended audience, feedback options, and review mech-

anisms. Our ﬁndings lead to actionable advice for indus-

try by articulating the beneﬁts and possible shortcomings of

the various communication channels in a knowledge-sharing

portal. We conclude by suggesting future research on the

increasing adoption of community portals in software engi-

neering projects.

Categories and Subject Descriptors

D.2.9 [Software Engineering]: Management

General Terms

Documentation, Human Factors

Keywords

Community Portal, Knowledge, Documentation

1. INTRODUCTION AND MOTIVATION

Software development is knowledge-intensive [25], and the

eﬀective management and exchange of knowledge is key

in every software organization. Knowledge is distributed

through various artifacts and media forms, from formal

documentation and technical articles, to blogs and wikis.

Permission to make digital or hard copies of all or part of this work for

personal or classroom use is granted without fee provided that copies are

not made or distributed for proﬁt or commercial advantage and that copies

bear this notice and the full citation on the ﬁrst page. To copy otherwise, to

republish, to post on servers or to redistribute to lists, requires prior speciﬁc

permission and/or a fee.

ESEC/FSE’11, September 5–9, 2011, Szeged, Hungary.

However, many organizations struggle with the eﬀective ex-

change and dissemination of knowledge across the commu-

nity [15, 28].

Although various forms of documentation exist, software

projects encounter many knowledge management challenges

[9, 24]: How should knowledge be distributed? How should

knowledge be kept up to date? How should feedback be

solicited? How should knowledge be organized for easy ac-

cess? There is no roadmap for what kind of information is

best presented in a given artifact, and new forms of docu-

mentation, such as wikis and blogs, have evolved [20]. Unlike

more formal mechanisms, wikis and blogs are easy to create

and maintain. However, they do not oﬀer the same author-

itativeness that comes with traditional documentation, and

they can become outdated and less concise over time [5].

While the informality of a wiki page or blog is sometimes

enough, users often expect reviewed technical articles.

One mechanism that has emerged recently and brings var-

ious communication channels together is the use of web or

community portals. Such portals are not just used in soft-

ware communities, but are essential to companies, such as

Amazon.com, eBay or TripAdvisor, where they enable the

development of communities around products. Similarly,

many of today’s software projects wish to solicit feedback

and input from a broader community of users, beta-testers,

and stakeholders. While several projects use community

portals, such as Microsoft’s MSDN

or IBM’s Jazz

, the use

of portals for software development projects has not been

studied yet. It is unclear how portals can play an eﬀective

role in software development, and how diﬀerent communi-

cation channels and artifacts should be utilized.

This paper presents a ﬁrst study of a successful software

project – IBM’s Rational Team Concert – and its use of a

community portal. Our methodology follows a grounded

theory approach, using data from 13 semi-structured in-

terviews with developers, ethnographic ﬁeld notes gathered

during observations on-site, and quantitative data on the

content of the community portal. Based on our ﬁndings,

we develop a model of artifacts in a community portal that

characterizes the artifacts along diﬀerent dimensions, such

as intended audience, content type, feedback options, and

review mechanisms. We provide actionable advice on how a

community portal can provide beneﬁts to a software project,

and we discuss how the diﬀerent channels and artifacts avail-

able in a portal can be used eﬀectively. We also identify

the shortcomings of diﬀerent artifacts, and suggest improve-

http://msdn.microsoft.com/en-us/

https://jazz.net/pub/index.jsp

ments to current tools and processes. With this research,

we aim to assist managers and developers in their decisions

about using community portals.

The remainder of this paper is structured as follows. Re-

lated work is summarized in Section 2. In Section 3, we

describe the community portal case study, and in Section

4, we introduce our methodology. In Section 5, we report

our ﬁndings and describe a knowledge artifact model. We

discuss our ﬁndings in Section 6, and we report limitations

of our work in Section 7. Section 8 concludes the paper and

outlines future work.

2. BACKGROUND AND RELATED WORK

We discuss three areas that are key to our research: knowl-

edge management, documentation in software development,

and the use of community portals to share information.

2.1 Knowledge in Software Development

Knowledge management has long been recognized as es-

sential to software development [14]. Rowley [26] deﬁnes

knowledge management as “concerned with the exploitation

and development of the knowledge assets of an organization

with a view to furthering the organization’s objectives. The

knowledge to be managed includes both explicit, documented

knowledge, and tacit, subjective knowledge.” How knowl-

edge is managed in an organization depends on the particu-

lar style of software development. Plan-based or traditional

methods usually rely on the management of explicit knowl-

edge, while Agile methods rely on the management of tacit

knowledge [22].

The number of studies on knowledge management in soft-

ware development projects is limited. An overview of knowl-

edge management in software engineering was conducted by

Rus et al. [27]. The focus of their review is on motivations for

knowledge management, diﬀerent approaches to knowledge

management, and factors that are important when imple-

menting knowledge management strategies in companies.

Dingsøyr and Conradi [8] report on a literature survey

of the lessons learned on the actions taken by companies,

the eﬀects and beneﬁts of these actions, and descriptions

of the strategies for knowledge management. Bjørnson and

Dingsøyr [2] also identiﬁed the following main ﬁnding across

several papers: there is a need not to focus exclusively on

explicit knowledge but also on tacit knowledge.

Within knowledge management, the transfer of knowledge

is particularly challenging. As described by Komi-Sirvi

o et

al. [17], sharing knowledge is diﬃcult in the light of short-

term goals and companies often fall back to needs-based

approaches for knowledge transfer. Knowledge transfer is

essential when integrating new developers into a team. In a

recent study using grounded theory, Dagenais et al. [4] iden-

tify early experimentation, internalization of structures and

cultures, and progress validation as the main factors for the

integration of newcomers. They also found that documen-

tation was often not helpful and outdated.

Our research focuses on the knowledge captured in a com-

munity portal and on how to present it eﬀectively.

2.2 Documentation in Software Development

Documentation has long been prominent in the list of rec-

ommended software engineering practices [6]. In 1986, Par-

nas and Clements [23] described a design process in which

documentation plays a major role. They argue that many

developers regard documentation as a necessary evil, writ-

ten only as an afterthought to adhere to bureaucratic regu-

lations. Therefore, documentation ends up being incomplete

and inaccurate. Parnas and Clements identiﬁed four orga-

nizational issues that lead to those problems: poor organi-

zation of the documents which makes it harder to maintain

them, boring prose that leads to inattentive reading, con-

fusing and inconsistent terminology, and myopia caused by

authors who know the documented system too well to take

a comprehensive point of view.

In a survey of software professionals, Forward and Leth-

bridge [10] found that content, up-to-dateness, availability,

and the use of examples are the most important document

attributes. In a related study, Lethbridge et al. [19] ex-

plored how software engineers use and maintain documen-

tation. They found that most software engineers do not up-

date documentation in a timely manner, with the exception

being highly-structured, easily-maintained forms of docu-

mentation, such as test cases and in-line comments.

Kajko-Mattsson [15] reports on a study in which she found

poor documentation practices, even though her interviewees

understood the necessity of documentation. Her partici-

pants identiﬁed the lack of a detailed documentation model

as the main problem. Similar results were found in a study

by Visconti and Cook [28]. They discovered a maturity gap

between policies and the adherence to those policies.

Dagenais and Robillard [5] conducted a study to under-

stand how documentation in open source projects is created

and maintained. Among other ﬁndings, they report that

the use of wikis for documentation has several shortcom-

ings, such as a lack of authoritativeness, and that the use of

a separate documentation team results in naming inconsis-

tencies and out-of-date documentation.

Recent trends, such as agile documentation [1], suggest

that executable products should be preferred over static doc-

umentation. From the agile point-of-view, documentation

should be just good enough and it should only be updated

“when it hurts” [1]. The rationale behind this is that the fun-

damental issue is communication, not documentation, and

that comprehensive documentation does not ensure project

success. While the developers in our study did not use agile

documentation, only some of the documentation was for-

mally organized, whereas other parts of the community por-

tal evolved without strict rules.

2.3 Research on Community Portals

Community portals have become a platform used by on-

line communities for the dissemination of information on

the Web [13]. Members of a community can ﬁnd relevant

information on portals and can often contribute their own

content. These community portals attract a disproportion-

ately large amount of web traﬃc and engage visitors with

free content [7]. Kondatova and Goldfarb [18] identify three

objectives of such portals and the communities surrounding

them: to supply content to users, to encourage members

to contribute, and to facilitate communication and interac-

tion. Portals aim to meet these objectives by oﬀering var-

ious forums, wikis, mailing lists, and other communication

mechanisms [18]. As Millen [21] points out, the archive of

conversation in an online community is often a valuable in-

formation resource.

To date, research on community portals has focused on

areas such as education and e-Government. Katz [16] exam-

ines how the use of portals allows educational institutions to

integrate information and services used by their communi-

ties, and aims to guide institutions in their decisions about

using portals. Gant and Gant [12] found the responsibili-

ties of simultaneously providing breadth and depth in online

content to be a challenge for governments using portals.

As mentioned earlier, the use of community portals by

software development projects has not been studied yet.

Several open source projects, such as Eclipse

, PHP

or the

Apache HTTP Server

, oﬀer portals that bring together a

variety of artifacts, such as bug trackers, mailing lists, fo-

rums, wikis, and events. More recently, commercial prod-

ucts, such as Microsoft’s Visual Studio or IBM’s Jazz, have

started using community portals. For example, the MSDN

portal

features a library with API references, code samples

and tutorials, a learning section, forums, and a blog. Next,

we describe the IBM Jazz community portal in more detail.

3. THE COMMUNITY PORTAL

IBM’s Jazz is a technology platform for collaborative soft-

ware delivery and the development of products on top of the

platform is made transparent through the community portal

jazz.net. The ﬁrst product developed on the Jazz platform

is Rational Team Concert (RTC), a software development

team collaboration tool that was ﬁrst released in 2008. Re-

cent additions to the Jazz platform include products such as

Rational Insight and Rational Build Forge. In our study, we

focus on RTC as it is the most mature of the Jazz products.

Jazz.net hosts the entire development process including

work items, developer mailing lists, the development wiki,

forums, the team blog, and a library. The library features

technical articles, podcasts, presentations, videos, and of-

ﬁcial product documentation, with tags organizing all arti-

facts in the library. All artifacts on jazz.net contain informa-

tion that indicates their associated product, with some ar-

tifacts belonging to more than one product. Figure 1 shows

a screenshot of one of the portal pages. All content is acces-

sible to users after a free registration.

Figure 1: The IBM Jazz community portal

Technical articles are written by developers and product

experts, and explore tasks, use cases, solutions, and con-

cepts in depth. Podcasts stem from various sources, includ-

ing news and updates about Jazz products, and often feature

Jazz developers and managers. The presentations have been

given by developers and product managers at conferences,

http://www.eclipse.org/

http://www.php.net/

http://httpd.apache.org/

http://msdn.microsoft.com/en-us/

or they are simply informational. Videos are produced by

Jazz developers or experts, and typically demonstrate the

use of Jazz products or introduce enhancements. The oﬃ-

cial product documentation consists of HTML content that

is part of the product releases. The content is produced by a

separate documentation team and is also accessible through

the product help menus. Table 1 shows the diﬀerent RTC-

related artifacts that were available on jazz.net at the time

of our study. Developers started to contribute to the com-

munity portal in the summer of 2007.

Table 1: RTC-related artifacts on jazz.net

type amount

forum 8,413 threads

wiki 3,334 pages

mailing list 3,292 messages

blog 77 posts

oﬃcial product documentation 4 manuals

library 93 articles

58 videos

30 presentations

7 podcasts

4. RESEARCH METHODOLOGY

Here, we outline the research methodology used to under-

stand how software developer knowledge is communicated

through the Jazz portal. We cover the setting of our study,

as well as the data collection and analysis methods used.

4.1 Research Setting

Our study took place with the development team of Ra-

tional Team Concert at IBM in the summer of 2010. The

team consists of approximately 150 contributors in about 30

functional teams, with some teams acting as sub-teams of

larger teams, and with many contributors assigned to multi-

ple teams. The team members are located at approximately

15 locations, primarily in North America and Europe. The

development follows the “Eclipse Way” development process

[11]. Created by the Eclipse Development team, this process

is an agile, iteration-based process with a focus on consis-

tent, on-time delivery of software through continuous inte-

gration, testing, and incremental planning.

4.2 Data Collection

Our methodology followed a mixed methods approach,

collecting both quantitative and qualitative data to allow for

triangulation. To gather quantitative data on jazz.net, we

used web scraping to download the artifacts shown in Table

1, along with their tags. At the time of our data extraction,

the RTC team was working on milestones towards the 3.0

release. The quantitative data was collected to gain initial

insights into the kinds of artifacts available on the commu-

nity portal. These insights were used to guide our qualitative

data collection that was done through a series of 13 semi-

structured interviews. Seven interviews were conducted in

person at an IBM location, and the remaining six interviews

were conducted by phone. We interviewed ﬁve developers,

four component leads, the RTC development manager, the

RTC project administrator, and two client developers that

work on Jazz-related projects inside IBM. For the remainder

of this paper, we use D1 to D5 to refer to the developers, L1

to L4 to refer to the component leads, M1 and M2 to refer

to the individuals with project-wide tasks, and C1 and C2

to refer to the client developers.

The initial interview script contained about 40 questions

on participants’ use of jazz.net

. As expected with grounded

theory, new questions emerged which led us to reﬁne the in-

terview questions used during the course of the study. We

asked our interviewees how they learn about Jazz function-

alities, what kind of questions they ask using jazz.net, and

how they use the published artifacts. We also investigated

whether they had ever contributed content, and if so, who

or what had triggered that contribution. We asked if there

were perceived gaps in the documentation, and about po-

tential tool and process enhancements. We used follow-up

questions for clariﬁcations and additional details, trying to

understand the scenarios that had led to the creation and

use of diﬀerent artifacts. In addition, one of our researchers

spent three months on-site, frequently engaging in informal

discussions with developers regarding their use of jazz.net,

and facilitating the member checking of our ﬁndings. These

observations were recorded using ethnographic ﬁeld notes

and allowed for insights into the internal processes that were

not revealed in interviews, in particular the processes that

would ultimately lead to the creation of new artifacts.

4.3 Data Analysis

Our data analysis followed the grounded theory approach

as described by Corbin and Strauss [3]. Grounded theory

implies that data collection and analysis are interrelated

processes, and that concepts are the basic units of analysis.

These concepts are obtained using“open coding”in which the

collected data is conceptualized line by line and concepts are

only created when they are present repeatedly. We applied

open coding to the transcripts of our interviews, to the data

downloaded from jazz.net, and to the ﬁeld notes collected

on site. Based on the concepts, more abstract categories are

developed and related. Each category is developed in terms

of its properties, dimensions, conditions, and consequences.

In the next step, called “axial coding”, data is put together

in new ways, thus making explicit connections between cat-

egories and sub-categories. Sampling in grounded theory is

done on theoretical grounds where incidents and events are

sampled rather than subjects or data sources. In the ﬁnal

step of “selective coding”, the core category is identiﬁed and

systematically related to other categories. Since all ﬁndings

in a grounded theory study are linked to speciﬁc evidence,

we are able to attribute all the ﬁndings we reported to inter-

view quotes, ﬁeld notes, or speciﬁc content on jazz.net. We

considered all quantitative and qualitative data collected in

our study during the analysis.

5. MODEL OF KNOWLEDGE ARTIFACTS

The “core category” identiﬁed in our grounded theory

study is a set of key characteristics of diﬀerent artifacts in

a community portal. We can distinguish the artifacts along

eight diﬀerent dimensions that emerged from the axial and

selective coding as part of our data analysis. The dimensions

underline the particular role of each artifact.

The set of questions used in the interviews is available on-

line at http://tinyurl.com/jazz-net.

Content: the type of content typically presented in the

artifact.

Audience: the audience for which the artifact is intended.

Trigger: the motivation that triggers the creation of a new

artifact.

Collaboration: the extent of collaboration during the cre-

ation of a new artifact.

Review: the extent to which new artifacts are reviewed

before publication.

Feedback: the extent to which readers can give feedback.

Fanfare: the amount of fanfare with which a new artifact

is released.

Time Sensitivity: the time sensitivity of information in

the artifact.

For the oﬃcial product documentation, technical articles,

blog posts, and the developer wiki, we analyzed these di-

mensions in detail. Figures 2 and 3 summarize the ﬁndings

and show an example for each kind of artifact. We focus

on these particular kinds of artifacts because they are also

common in other community portals and in software devel-

opment in general. The ﬁndings can be traced back to quotes

from the interviews we conducted, and the interviewees are

noted in the ﬁgures and text using subscript. Where rele-

vant, we also discuss the artifacts and the ethnographic ﬁeld

notes that we analyzed related to the dimensions. The fol-

lowing sections report on each of the dimensions in detail. In

addition, Section 5.9 highlights several explicit and implicit

connections between the artifacts.

5.1 What content is communicated?

Content is the ﬁrst dimension along which the diﬀerent

artifacts of the community portal can be distinguished. The

oﬃcial product documentation is based on features and

does not cover scenarios: “I would try to cover all the corner

cases, and I found those were often omitted in the help doc

because the eﬀort to get into those corner cases were lengthy

things. You would have to have a whole section of how to put

yourself into a corner, and then a whole other section of how

to get yourself out of this corner.”

However, documenting

scenarios is important: “It’s always good to document a wid-

get, but it’s more important in many cases to document a

process and being able to follow the process of how you do

an upgrade or how you do a plan and all the widgets that

you touch during that. [...] It’s the context of how you use

the widget that’s much more important.”

This need is addressed by technical articles that feature

scenarios and oﬀer more depth than the oﬃcial documenta-

tion. D4 described the reason to write an article: “Because I

wanted to put in a lot of screenshots, and also explain what’s

happening. [...] You had to write some text around it. Why

you’re doing something and why is that needed.” Technical

articles also diﬀer in their style of writing: “Articles spice it

up a little. There’s more pictures and more personal I guess.

People have their own style.”

Tags also shed light on the content of diﬀerent kinds of

artifacts. The tags are assigned by a single person

. Table

2 shows the most-used tags per type. Videos are typically

used for high-level overviews or demos

L3,D2

, whereas pre-

sentations are related to conferences.

Blogs add a personal note to the artifacts on the com-

munity portal: “a personal view on something and not re-

ally documentation. [...] You want to make people aware

of something or tell them about something, but more like a

评论收藏

内容反馈

yangdemon

2012-09-24

很全，谢谢楼主分享。省了好多功夫
若水风尘

2012-10-27

没有软件体系结构方面的，不过其他方面挺全~
tulian

2013-06-27

楼主提供的资料都很有价值，赞！
ocean1231

2012-10-20

不错内容很全感谢分享！
践行者

2012-10-20

有的论文不去全，只有摘要

前往

页

蛐蛐蛐

粉丝: 668
资源: 65

FSE 2011-Foundations of Software Engineering 2011年论文集

最新资源

FSE 2011-Foundations of Software Engineering 2011年论文集

FSE 2012-International Symposium on the Foundations of Software Engineering 2012

ToSEM 2013-ACM Transactions on Software Engineering and Methodology 2013年论文集

ISSRE 2010-International Symposium on Software Reliability Engineering 2010论文集

Engineering Better Software at Microsoft - Jason Yang, Slides - 30th Nov 2011 (25-sal)-计算机科学

International Conference on Software Engineering 2010年论文集

Foundations of Software Engineering 2010年论文集

CMA.zip_CMA算法_FSE CMA_FSE-CMA_FSE-CMA算法的Verilog实现_cma

Python库 | python_fse-0.3.3-cp35-cp35m-manylinux1_x86_64.whl

ASE 2013-International Conference on Automated Software Engineering 2013

ICSE 2011-International Conference on Software Engineering

Sommerville,.Software.Engineering,.9ed,.Addison-Wesley,.2011(ED2000.COM) PPT

ICSE 2011-International Conference on Software Engineering论文集合2

ICSE 2011-International Conference on Software Engineering论文集合1

Foundations of Software and System Performance Engineering Process, 无水印原版pdf

PyPI 官网下载 | python_fse-0.3.3-cp35-cp35m-manylinux1_x86_64.whl

FSE-S15-A1-SSNoC

FSE-2011-Crystal---old

fse18-artifact-183:阿尔卑斯山

基于FPGA的FSE-CMA盲均衡器设计与实现.pdf

Foundations of Software and System Performance Engineering 无水印原版pdf

FSE-Service-Mobility

9.her.ipynb

ct-fse-repo

fse-F14-SA5-SSNoC-Java-REST:fse-F14-SA5-SSNoC-Java-REST

Liber 5_information_FSE_源码

FSE - FutraSiteEngine-开源

pydriller：用于分析Git存储库的Python框架

FSE-Chat_Application

最新资源