A Collaborative Platform for Software Evolution Visualization

Leveraging Meta-model Driven Measurements with Big Data Strengths

João Carlos Caldeira

ISCTE - University Institute of Lisbon, Avenida das Forças Armadas, Lisbon, Portugal

1 STAGE OF THE RESEARCH

This document describes a preliminary PhD thesis

proposal that will hopefully lead to a collaborative

framework and platform for Software Evolution

Visualization (SEV).

Currently, most of the time is being dedicated to

review related works and deeply analysing gaps for

further research.

We sustain our decision to follow this research area

by evaluating recent works which have shown that

there is a need for multi-metrics, multi-perspective

and multi-strategy approaches to SEV as

summarized by (Novais, et al., 2013). The authors

identify some research niches such as missing case

studies, tool comparisons and experiments with the

aim of predicting defects, improve software quality

and development processes. Another missing aspect

relates to the presentation of real scalable

visualization and dependency impact among

projects. It is also recognized that there is little

formal validation and collaboration in this area, most

likely because the data is scarce, dispersed and not

widely shared by each individual researcher. The

lack of empirical studies is a real constraint to allow

the community to perform benchmarking and

compare methodologies and results. In other words,

the SEV community has failed to provide sound

evidences, through empirical validation studies, of

the impact of using the technology they created. In

fact SEV research deliverables provide visual

insights that are expected to help understand

complex software artefacts and ultimately contribute

to improve their quality and the maintenance process

itself. Failing to provide adequate justification, may

explain the reduced adoption of SEV tools in

industry as evidenced by a small percentage of job

offerings in industry related to software

visualization, when compared to software analysis,

and software metrics as represented in Table 1.

Table 1: Job offerings per area.

Software

metrics

Software

visualization

Software

analysis

www.dice.com

2085 427 13217

www.monster.com

1000+ 432 1000+

www.careerbuilder.com

4349 441 22846

2 OUTLINE OF OBJECTIVES

The goals of our research consist on proposing a

structured approach to (i) collect data from public

domain software repositories, (ii) extract complexity

and quality metrics using a meta-model driven

measurement approach (M2DM), (iii) store and

eventually transform those metrics by adopting big

data technologies for scalability sake, (iv) visualize

software evolution, along the corresponding metrics,

in a collaborative fashion, allowing to identify

patterns and trends. The aforementioned approach is

expected to scaffold exploratory activities on top of

the collected data, allowing the community to do

benchmarking, evaluate software engineering best

practices and assess software engineering research

questions by means of empirical studies (Goulão, et

al., 2012).

While (Sakamoto, et al., 2012) present a service

oriented framework to visualize software evolution

using Google charts and (Gonzalez-Torres, et al.,

2011) develops a tool for providing insights into

software evolution based on Eclipse, our work will

adopt some concepts from both works but will be

more oriented to understanding metric relationships

and impacts between components over time, and

also on producing consolidated predictions,

presumably using time series analysis. In addition,

we expect to explore an agile time navigation

paradigm based on the non-seasonal delivery of

software versions. Besides, we plan to use rendering

techniques to combine sequential static snapshots to

produce dynamic software visualization.

Carlos Caldeira J..

A Collaborative Platform for Software Evolution Visualization - Leveraging Meta-model Driven Measurements with Big Data Strengths.

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

3 RESEARCH PROBLEM

There are several strategies, approaches and ways to

visualize software evolution (Alam & Dugerdil, s.d.)

(Lanza & Ducasse, 2002) (Balzer, et al., 2005)

(Beyer & Hassan, 2006) (Burch, et al., 2005). Most

of the methods used to achieve this are based on

proprietary applications and/or plugins developed

mainly for Eclipse. Some plugins render data related

with source code management activities like check-

ins, check-outs and conflict detections, whilst some

others plot information related with source code

metrics such as number of lines of code, number of

classes and interfaces, complexity and quality

related metrics, amongst others (Hanakawa, 2007).

Although Eclipse is a very flexible and powerful

IDE, and great achievements have been obtained

with it to progress SEV techniques, most of the

developments and results are limited to a standalone

environment with lack of collaboration between

community peers. Data sharing and teamwork is

indeed needed for results comparison, software

quality analysis and finally to trigger improvement

processes. Like in any other improvement procedure

a critical resource required is to have data on which

to apply basic statistics methods, operations research

algorithms or simulation techniques.

When we leverage the statistics domain, a huge

challenge faced by researchers is the lack of enough

and/or accurate data in order to perform the desired

experiences and related studies. This is also true in

the experimental software engineering and mainly in

SEV where data collection is critical for a

quantitative and qualitative analysis to produce valid

results. Meanwhile, development teams and

individual programmers aim to have near real-time

understanding of software quality metrics about the

progress of their projects (Holten, et al., s.d.), (Lanza

& Ducasse, 2002).

Based on this, the main research problems in

software visualization are, (i) lack of clear

identification of metrics and uncertainty about the

format on how to store the information, (ii) no

public domain repository exists on where to easily

store, represent and share the metrics and, (iii) there

are some tools in the community, but in their

essence, they are standalone components with lack

of integration, collaboration and visualization

functionality, which makes it hard to perform

exhaustive analysis related with the software

development lifecycle.

Some of the existing limitations cause

tremendous constraints in using SEV as an

integrated approach to combine quantitative source

code analysis (SCA), source code management

(SCM) activities and bug tracking (BT).

4 STATE OF THE ART

This section describes the important topics to be

addressed within the whole research and also the

gaps that this research can fulfil within SEV.

4.1 Software Engineering

Software engineering (SE) is the application of a

systematic, disciplined, quantifiable approach to the

design, development, operation, and maintenance of

software, and the study of these approaches; that is,

the application of engineering to software.

4.1.1 Software Evolution

Software evolution sometimes also identified as

maintenance refers to maintaining software

components by finding and fixing defects and

introducing new functionalities to an existing

system. The perception is that more time is spent in

bug fixing rather than in adding new features. This

turns out to be incorrect as most of reported defects

are related with the need for functionality

enhancements. This points us to the fact that

software development is an evolutionary process

rather than a contained and well defined set of

components and functionalities. As all evolutionary

systems, the complexity and maintainability efforts

increase over time and the need to understand

software through the use of visual resources is of

much help mainly when the development is made in

a collaborative way, by diverse people in disjointed

geographies. Therefore, visualizing software

evolution related data in a project context and timely

oriented is a fundamental goal for software

development teams and individuals.

4.1.2 Visualization

In the context of software development, software

visualization is used to understand the components

being develop through the use of visual resources.

There are several proposals in the community to

address this need and the majority of them render

software related information either in specific

applications or as views mainly within the Eclipse

platform. Our goal is to decouple the visualization

layer from the IDE platform but allowing a cross

reference link between both. This approach gives

ACollaborativePlatformforSoftwareEvolutionVisualization-LeveragingMeta-modelDrivenMeasurementswithBig

DataStrengths

more flexibility to the developers as they are not

dependent of one specific environment to be able to

visualize software evolution.

4.1.3 Meta-model Driven Measurement

Several approaches have been used as the basis for

the research on software measurement. This has

resulted in the design of measurement techniques

and software metrics to assess software quality. The

term meta-model driven measurement (M2DM) was

initially described by (Abreu, 2001) and is used by

applying meta-models to detail both the components

to be measured and the metrics by which they are

measured with.

Several Eclipse plugins existed but they were not

leveraging the MD2M approach. Recently (Coimbra,

2013) made available a plugin to mine software

metrics using a M2DM concept and we plan to use it

to perform the initial extraction of code related

metrics.

4.1.4 Source Code Management

Another important source of data to SEV is code

repositories, mainly Git, SVN and Mercurial. These

repositories track the changes made by the

contributors to the projects they belong to. Those

changes are related with packages, classes, files and

code. It includes the act of adding new files or

folders, modifying existing ones and the removal of

others. These activities might alter tremendously the

architecture of a software project during its lifecycle.

Information related with SCM is intended to be used

as a source of data for the platform and for further

analysis.

4.2 Big Data

The term Big Data is currently extremely popular

within the IT market (Simmhan, et al., 2013). It is

also becoming a hot topic within the academic

institutions (Liu, et al., 2010). It refers to the ability

of collecting, storing, analysing and visualizing large

amounts of data. Usually this data is so large and

sometimes so complex that traditional tools and data

processing applications are no longer able to process

it within a tolerable elapsed time. It is common to

accept a system as a candidate for a big data use case

when it falls under these three basic dimensions

(3Vs): volume, variety and velocity. They

correspond to the amount of data being captured, the

number of different types of data and to the speed it

has to be processed in order to provide relevant and

timely results to the stakeholders. Usually this data

is a relevant factor when to perform analytical

functions within a specific area of business (Zhang

& Xie, 2012), (Zhang, et al., 2012). We plan to use

big data technologies, for collecting data and as a

repository for our platform as it does a perfect match

to address the requirements spawned by software

evolution visualization. It also provides the

scalability mechanisms that the collection of large

volumes of data might require in a short to medium

term.

4.3 The Platform

Figure 1: Collaborative SEV Platform.

The platform will divided in layers. Each layer will

have its own tasks and gateway points. The ETL

layer will be responsible for all the extraction data

from software repositories. The Parsing Layer will

be the point where metrics generation and extraction

take place. The Persistence Layer will store and

archive all the data produced. The Analytical

Service Layer will deal with all the services related

with analytical functions and the disposal of the data

for further consume by the Visualization Layer.

5 METHODOLOGY

The methodology comprises three main phases

aligned with the research goals which are: (i) the

setup phase for the initial research and global

planning, (ii) the execution phase for the

instantiation of the platform we want to develop and

to produce and publish the first results, (iii) the

validation phase where we plan to instantiate our

platform with data and perform exploratory analysis.

MODELSWARD2014-DoctoralConsortium

5.1 Setup

In this phase we expect to identify the problem, the

contributions we may add and the analysis of the

related work. Additionally it is the phase where we

setup our calendar and define the milestones in the

research.

5.2 Execution

It is related with the development of the platform

and the first experiments and hypothesis testing. As

we plan to have feedback early in the process from

international peers we will try to publish the first

results in the most relevant conferences, journals and

papers.

5.3 Validation

This is when we plan to perform exhaustive

hypothesis testing and start to extract some

conclusions that might help the software

maintenance process and software quality in general.

It is also the moment when we start to consolidate

all the knowledge and produce the final PhD

dissertation.

6 EXPECTED OUTCOME

The main goal is definitely to contribute to the

process of understanding how the software evolves

during its lifecycle and how the quality and/or

complexity is affected (or not) by all the changes

being made by the contributors. At the end of this

work software development teams and the academic

community should expect to have a common

platform for software evolution visualization. This

platform is intended to have near real time

integration with the most used software repositories

like: Git, SVN and Mercurial in order to extract the

metrics and metadata about code changes. Due to the

fact that Eclipse is one of the most used IDEs and

that Eclipse Orion is a recent platform for Cloud

based development, our goal is also to have plugins

for these environments. Using them, developers and

team managers will be able, in real time, to commit

their code changes to the repository and to analyse

software metrics and facts of their projects, compare

them with some other projects and even investigate

within the development groups the individuals who

are more active or build better quality software. For

the platform to be really useful, it requires to store

enormous amounts of data, which, once collected

and stored can then be used for pattern searching,

trend analysis, metric correlation and finally to build

prediction models and what-if scenarios.

6.1 Benefits

Each community will extract their own benefits

based on their interests, inputs and analysis

performed in the platform. Based on this

assumption, it’s important to highlight the potential

gains obtained by each of them.

6.1.1 Software Industry

To be able to link software quality metrics with

software changes and incidents, to correlate them, to

understand their impact, and at the same time to

assure near real time visualization of software

complexity which can be of great help in allocating

the right resources to specific activities.

6.1.2 Software Engineering Research

Community

Perform benchmark studies within a specific group

of students or compare results within different

classes. One may not exclude the possibility to

analyse and compare results between different

institutions and geographical zones, predict

behaviours and simulate scenarios.

6.1.3 Individual Developers

Improve development skills and adopt best practices

by keeping real time track of their development

performance and compare it with the community

peers.

REFERENCES

Abreu, F. B., 2001. Using OCL to formalize object

oriented metrics definitions, Lisbon: Technical Report

ES007/2001.

Alam, S. & Dugerdil, P., s.d. EvoSpaces Visualization

Tool: Exploring Software Architecture in 3D, Geneva,

Switzerland: s.n.

Bacchelli, A., Rigotti, F., Hattori, L. & Lanza, M., s.d.

Manhattan— 3D City Visualizations in Eclipse,

Switzerland: University of Lugano.

Balzer, M., Deussen, O. & Lewerentz, C., 2005. Voronoi

Treemaps for the Visualization of Software Metrics,

s.l.: s.n.

ACollaborativePlatformforSoftwareEvolutionVisualization-LeveragingMeta-modelDrivenMeasurementswithBig

DataStrengths

Beyer, D. & Hassan, A. E., 2006. Animated Visualization

of Software History using Evolution Storyboards. s.l.,

IEEE.

Beyer, D. & Hassan, A. E., 2006. Evolution Storyboards:

Visualization of Software Structure Dynamics. s.l.,

IEEE.

Breivold, H. P., Crnkovic, I. & Larsson, M., 2011. A

systematic review of software architecture evolution

research. Information and Software Technology.

Burch, M., Diehl, S. & WeiB3gerber, P., 2005. EPOSee -

A Tool For Visualizing Software Evolution,

Eichstatt,Germany: s.n.

Coimbra, P. J., 2013. An Eclipse Plug-in for Metamodel

Driven Measurement, Lisbon: ISCTE-IUL.

D’Ambros, M., Lanza, M. & Lungu, M., 2006. The

Evolution Radar: Visualizing Integrated Logical

Coupling Information. Shanghai, China, ACM.

Erra, U., Scanniello, G. & Capece, N., 2012. Visualizing

the Evolution of Software Systems using the Forest

Metaphor. s.l., s.n.

Gonzalez-Torres, A. et al., 2011. Maleku: an evolutionary

visual software analytics tool for providing insights

into software evolution. Williamsburg VA, USA,

IEEE.

Goulão, M., Fonte, N., Wermelinger, M. & Abreu, F. B.

e., 2012. Software Evolution Prediction using Sea-

sonal Time Analysis: A Comparative Study. s.l., s.n.

Hanakawa, N., 2007. Visualization for software evolution

based on logical coupling and module coupling. s.l.,

IEEE.

Holten, D., Vliegen, R. & Wijk, J. J. v., s.d. Visualization

of Software Metrics using Computer Graphics

Techniques, Eindhoven, The Netherlands: s.n.

Holt, R. & Pak, J. Y., 1996. GASE: Visualizing Software

Evolution-in-the-Large, Toronto, Canada: s.n.

Hong, Q., Kim, S., Cheung, S. & Bird, C., 2011.

Understanding a Developer Social Network and its

Evolution. s.l., IEEE.

Langelier, G., Sahraoui, H. & Poulin, P., 2008. Exploring

the Evolution of Software Quality with Animated

Visualization. Montréal, Canada, IEEE.

Lanza, M. et al., 2013. Manhattan: Supporting Real-Time

Visual Team Activity Awareness. San Francisco, USA,

IEEE.

Lanza, M. & Ducasse, S., 2002. Understanding software

evolution using a Combination of Software

Visualization and Software Metrics, Berne,

Switzerland: s.n.

Lanza, M., Gall, H. & Dugerdil, P., 2009. EvoSpaces:

Multi-dimensional Navigation Spaces for Software

Evolution. s.l., IEEE.

Liu, G., Zhang, M. & Yan, F., 2010. Large-Scale Social

Network Analysis based on MapReduce. s.l., IEEE.

Novais, R. L., 2013.

Visualizando Evolução de Software

Em Detalhes, Salvador: s.n.

Novais, R. L. et al., 2011. An Interactive Differential and

Temporal Approach to Visually Analyze Software

Evolution, s.l.: IEEE.

Novais, R. et al., 2012. On the Proactive and Interactive

Visualization for Feature Evolution Comprehension:

An Industrial Investigation. Zurich, Switzerland, s.n.

Pérez, J., Deshayes, R., Goeminne, M. & Mens, T., 2012.

SECONDA: Software Ecosystem Analysis Dashboard.

s.l., IEEE.

Renato Lima Novais, A. T. T. S. M. M. N. Z., 2013.

Software evolution visualization: A systematic

mapping study. Information and Software Technology,

31 May.

Ripley, R. M., Sarma, A. & Hoek, A. v. d., s.d. A

Visualization for Software Project Awareness and

Evolution, Irvine, CA 92697-3425 USA: Donald Bren

School of Information and Computer Sciences.

Sakamoto, Y., Matsumoto, S. & Nakamura, M., 2012.

Integrating Service Oriented MSR Framework and

Google Chart Tools for Visualizing Software

Evolution. s.l., IEEE.

Servant, F. & Jones, J. A., s.d. Chronos: Visualizing Slices

of Source-Code History, Irvine, USA: s.n.

Shollo, A. & Pandazo, K., 2008. Improving presentations

of software metrics indicators using visualization

techniques, Göteborg, Sweden: s.n.

Simmhan, Y. et al., 2013. Cloud-based Software Platform

For Big Data Analytics In Smart Grids. s.l., IEEE.

Souza, R., Silva, B., Mendes, T. & Mendonça, M., s.d.

SkyscrapAR: An Augmented Reality Visualization for

Software Evolution. s.l., s.n.

Sun, X., Gao, B., Fan, L. & An, W., 2012. s.l., IEEE.

Ulges, A., 2005. Visualizing Software Evolution,

Kaiserslautern: s.n.

Vasa, R., 2010. Growth and Change Dynamics in Open

Source Software Systems, Melbourne, Australia: s.n.

Wu, J., Holt, R. C. & Hassan, A. E., 2004. Exploring

Software Evolution Using Spectrographs, s.l.: s.n.

Zhang, D., 2012. Software Analytics in Practice –

Approaches and Experiences. Zurich, Switzerland,

IEEE.

Zhang, D., Dang, Y. & Han, S., 2012. Teaching and

Training for Software Analytics. s.l., IEEE.

Zhang, D. & Xie, T., 2012. Software Analytics in

Practice:Mini Tutorial. Zurich, Switzerland, IEEE.

MODELSWARD2014-DoctoralConsortium