Software Engineering Projects Analysis using Interactive Multimodal

Graph Explorer – IMiGEr

Lukas Holy, Petr Picha, Richard Lipka and Premek Brada

NTIS – New Technologies for the Information Society, European Centre of Excellence, Faculty of Applied Sciences,

University of West Bohemia, Univerzitni 8, Pilsen, Czech Republic

Keywords:

Graph Visualization, Visual Clutter, Large Graphs, Software Engineering.

Abstract:

This paper describes a visualization technique designed to help work with complex diagrams containing mul-

tiple types of nodes and edges, by using a combination of visual clutter reduction and graph exploration

techniques. We show its application, including preliminary evaluation, on software engineering projects data

gathered from various tools and repositories used for software development. An online tool implementing the

technique and plans for its extension by a connected view of time perspective data are brieﬂy presented.

1 INTRODUCTION

Software development projects become more com-

plex and their content includes more types of dedi-

cated roles, artifacts and activities. Thus, getting in-

sights and understanding the relations among these el-

ements is more and more difﬁcult. Often the informa-

tion is tracked in multiple unconnected or loosely con-

nected systems. There are several challenges when it

comes to full understanding of a project such as being

able to easily ﬁnd out:

• status of activities and artifacts,

• overall difference of artifacts since last check,

• health of the project from the process point of

view (e.g., if there are any anti-patterns occur-

ring),

• balance of effort spent by various team members,

• ﬁnd out the artifacts related to a group of items

(e.g., tickets a in sprint)

• each team member performing activities appropri-

ate to role(s) assigned to him/her,

The above-mentioned topics are just a few of dif-

ferent phenomena one could need to ﬁnd out about

the project as there are many anti-patterns identiﬁed.

Besides that there are many problems remaining un-

covered.

We address the above-mentioned challenges by

the new approach using software process anti-pattern

detector and novel interactive visualization. This pa-

per mainly focuses on the visualization part of the ap-

proach. Its novel visualization techniques will be used

for understanding the software development process.

We use the software engineering project data repre-

sented as large graphs consisting of large number of

nodes and edges. These are in its nature of various

types. It is difﬁcult to explore the structure of such

graphs, create a mental model of the whole graph and

ﬁnd its relevant parts.

The approach presented in this paper attempts

to reconcile the contradictory requirements to show

large graphs while being able to see enough details

needed for understanding. Advanced visualization

techniques used in our approach helps the process of

understanding. It describes possible ways of address-

ing some of the challenges by providing the ability

of fast understanding element relations and mental

model creation. Understanding such relations could

lead to lowering of time wasted on its analysis and

improvements design. In this paper, we present gen-

eralized concepts of node-edge graph exploration on

domain of software engineering project elements (and

their relations). We are designing the solution to be

able to work with data from other domains as well.

Currently, we implemented part of the overall

concept of project exploration. Currently researched

parts are described in Section 5. Such concept could

then help understanding various aspects by interac-

tive visualization of static relations visualization and

interconnected Time-line.

330

Holy, L., Picha, P., Lipka, R. and Brada, P.

Software Engineering Projects Analysis using Interactive Multimodal Graph Explorer – IMiGEr.

DOI: 10.5220/0007579803300337

In Proceedings of the 14th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2019), pages 330-337

ISBN: 978-989-758-354-4

1.1 Structure of the Paper

The remainder of the paper is structured into follow-

ing sections. Section 2 provides overview of the re-

lated work. Section 2.1 describes the SPADe tool

used for repositories mining and analysis. Section 3

describes our experimental notation, the IMiGEr tool

and relevant parts for visualization. Technical infor-

mation related to implementation and availability are

described in Section 4. Section 5 presents the work

in progress and improvements planned. Finally, Sec-

tion 6 concludes the paper with the summary of ﬁnd-

ings.

2 RELATED WORK

For the related work, we are focusing on visualization

domain as we mainly present such approach in this

paper. Our focus is on various techniques to reduce

the complexity of visualized information.

Visual clutter in large node-edge diagrams

can be reduced by many techniques, such as

bundling (Holten and Van Wijk, 2009), sam-

pling (Raﬁei, 2005), clustering (Chen and Liu, 2003),

etc. From the well-developed taxonomy of these tech-

niques described by Ellis and Dix (Ellis and Dix,

2007) the following ones are primarily relevant for

our work. An example of such clutter is illustrated

in Figure 1 that shows relations between software en-

gineering artifacts in 5 months software development

projects done by 4 members teams.

Firstly, visual clutter caused by the lines is often

reduced by edge bundling (Holten, 2006). Although

this approach reduces the clutter, it makes it difﬁcult

to trace the dependencies between connected nodes

leading through the edge bundles. Secondly, the clus-

tering of components (nodes) so that multiple compo-

nents are visually represented by one node can also

reduce visual clutter. Thus, the number of nodes in

the whole diagram is reduced, though the connections

among components are usually still present. Clus-

ters can either be marked manually, in an automated

way (Chiricota et al., 2003; Mancoridis et al., 1998)

or by a combination of those approaches. Lastly, the

chosen layout algorithm is an inﬂuencing factor since

it can ease orientation in both clustered graphs (Feng,

1997) and the non-clustered ones (Hachul and Jnger,

2007).

Previously, we veriﬁed some of the concepts de-

scribed in this paper on a user-study (Holy et al.,

2015) , but limiting the approach only to one type of

node and one type of edge visualizing only the do-

main of the Java software call graphs. There are many

available tools for graph visualization such as Gephi

or NodeXL

. Currently visualized data are in its na-

ture multimodal graphs (Ghani et al., 2012), (Srini-

vasan and Stasko, 2018) because the nodes and re-

lations belong to different types. For nodes, we are

showing for instance authors, commits, wiki pages

etc. Edges then captures various types of relations

among nodes, such as authorship, commit to ticket

connections etc.

2.1 SPADe

To be able to address above-mentioned challenges

we need a way of collecting the software engineer-

ing data from various tools. Software Process Anti-

pattern Detector (SPADe) is an experimental tool col-

lecting software project development data from repos-

itories in various Application Lifecycle Management

(ALM) tools. The overall architecture of SPADe is

shown in Figure 2. SPADe stores data in a metamodel

(Picha and Brada, 2016) capable of capturing project

data irrespective of the source tool and process model

used by any particular project. The main purpose of

SPADe is to detect software process and project man-

agement bad practices, a.k.a. anti-patterns. Other

uses include compliance check of the project and a

given process model or analyses speciﬁcally aimed

on a certain aspect of the project, for example, the

role of an architect (Picha et al., 2017) . The research

topic attached to SPADe then focuses on investigating

the relation (if there is any) between these phenomena

and projects success and/or product quality.

Along with storing data in the database, SPADe

produces several outputs to be used in other tools for

specialized analysis. One of those is a JSON ﬁle

with relevant project data to be visualized in IMiGEr.

These include people active in the project, artifacts

(ﬁles, wiki pages, etc.), tickets from issue-tracking

systems, repository commits and general changes per-

formed on the tickets or artifacts. The SPADe JSON

ﬁle also includes relations between these elements,

for instance authorship, ticket assignments, explicit

ticket relations, commit-ticket traceability links, arti-

facts attached to tickets and/or wiki pages, etc. When

visualized in IMiGEr, this data provides an insight

into the structure of the projects and personnel activi-

ties.

https://gephi.org/

https://archive.codeplex.com/

Software Engineering Projects Analysis using Interactive Multimodal Graph Explorer – IMiGEr

331

Figure 1: Visual clutter caused by large number of nodes and edges.

Figure 2: Overall SPADe architecture.

3 INTERACTIVE MULTIMODAL

GRAPH EXPLORATION

The aim to described approach is to be able to ﬁnd out

desired information quickly. When we speak about

software engineering project data, in most of the cases

we will be working with data containing at least hun-

dreds or rather thousands of elements (nodes). While

exploring such number of elements in a graphical way

shown all on one canvas we will see lot of visual

clutter. We are facing the contradictory requirements

of not seeing useful information when showing the

whole graph and not seeing the context of elements

when showing enough details about particular nodes.

Additionally, in software projects domain we have

multiple types of both nodes and edges, which creates

additional complexity to the information visualized.

While for single type of node and edge user can easily

understand the line between two nodes as connection,

for multiple types of edges simply showing single line

does not provide detailed information. Similarly, the

nodes visualization should reﬂect its type in way that

is effective.

As an outcome of above-mentioned challenges,

we designed the visualization to use interactivity and

details on demand principle to make the work with the

graph effective. One of the key aspects is to use infor-

mation hiding principle to reduce the number of nodes

in the graph to enable the mental model creation.

The approach described in this paper is able to dis-

play various types of data, which is enabled by gen-

eralized input graph structure. It is able to handle

IVAPP 2019 - 10th International Conference on Information Visualization Theory and Applications

332

Figure 3: Complex node Application Explorer tool demonstration.

multiple types of both nodes and edges. Thus, it can

suite to any graph data which could be converted to

given generalized structure (e.g., software engineer-

ing projects, intelligence agencies analysis, historical

data, computer networks, social networks).

To be able to verify techniques presented in this

paper we have created their implementation. It is

called Interactive Multimodal Graph Exploration ab-

breviated as IMiGEr. It is able to show the software

engineering data represented as a graph in the web

page. The demonstration of the IMiGEr’s GUI is

shown in Figure 3. It shows the tool interface on a

node diagram of a software engineering project. Fig-

ure 3 shows one instance of generalized graph struc-

ture (multiple types of nodes and edges) on depen-

dency graph showing relations among software engi-

neering project artifacts, people involved and tickets

created during software development project. These

data were collected by the SPADe (Picha and Brada,

2016) project data mining tool

from various data

sources, for instance Git, JIRA, Bugzilla. These data

were collected from 5 months software development

projects done by 4 members teams.

https://github.com/ReliSA/SPADe

3.1 Visualization Overview

The visualization consists of interactive graph area,

side area and toolbar. In the interactive graph area,

nodes are shown including their type, for instance

”wiki page”. Details like artifact authorship are hid-

den to reduce the visual clutter and provided on de-

mand by selecting individual edge (cf. the grey pop-

up). In such case the connected nodes and the edge

itself are highlighted, rest of the nodes are suppressed

using transparency.

The side area is showing nodes excluded from the

graph. For such nodes, there are no edges shown to

reduce a visual clutter in the graph area in ﬁrst place,

but they could be shown on demand. Another option

is to use so called proxy symbols instead. These sym-

bols indicate connection between node in side area

and node in graph area, as shown near each node.

For further visual clutter reduction and graph sim-

pliﬁcation, the concept of groups is used. These

are created in the side area for representing multi-

ple nodes as one while providing aggregate informa-

tion on connections. Each group can also be shown

back in the graph area as one node. The toolbar al-

lows to perform search, trigger force-directed layout,

and other standard operations (such as panning and

zooming). For easier navigation in the diagram the

minimap is shown in the right panel.

Software Engineering Projects Analysis using Interactive Multimodal Graph Explorer – IMiGEr

333

3.2 Interactivity Features

The approach we have proposed reduces the visual

clutter by multiple visualization techniques. In this

section we will focus on their explanation.

As we use removing of the nodes from the main

graph area to the SeNoA area as visual clutter re-

duction technique, we need two modes of manipulat-

ing the nodes. Its switching is done with appropriate

icons in the toolbar. First mode is for moving nodes

(A) where the user can manually adjust the layout of

the diagram. Second mode (B) serves for removing

nodes from the diagram area to the SeNoA area sim-

ply by clicking on the desired nodes, which should be

removed.

Last two icons in the toolbar serve for the auto-

matic removal of a conﬁgured number of nodes from

the diagram to the SeNoA area. The tool is currently

conﬁgured to remove most connected node, but more

sophisticated algorithms are subject of a future work.

The icon (C) is used for removing such node and

adding it to the SeNoA area as individual item. The

next icon (D) creates one group for all of them. These

techniques helps user to quickly reduce the initial vi-

sual clutter of the graph by removing the most con-

nected nodes. As in most of the graphs are very likely

present nodes having much higher number of edges

compared to other ones.

From our observations, we know that after remov-

ing most connected nodes, users usually focus on the

items they are interested in. For such cases a simple

search could be used. In Figure 3, one can see the

search result for a string “archite”. Two nodes in the

diagram contain this word as indicated by the number

(F). Nodes matching the search are highlighted by an

orange color (E).

3.3 The Separated Nodes Area

SeNoA (sidebar) is used for removing nodes with

a large number of connections from the main dia-

gram into a so called separated nodes area (abbre-

viated as SeNoA) placed on the border of a window

(the right sidebar in Figure 3). When a user moves

components from the main diagram to this area, the

lines between these components and remaining com-

ponents are elided. This essentially marks a compo-

nent as a “familiar one”. Once component removed,

the user may continue getting familiar with the rest

of the system. Often, a particular node (such as per-

son) is connected to large number of other ones. Such

nodes are good candidates to be place to SeNoA (U).

Then, graph’s visual clutter is usually considerably

decreased. One of the cases may be a node showing

person. Such a node is probably connected to most of

nodes in the system and its displacement reduces the

graph complexity.

We measured several projects lasting 5 months,

which were done by 4 people team. The results of

clutter reduction are shown in Table 1. Columns

Nodes and Edges represent the number of nodes and

edges created in the graph by the SPADe data collec-

tion and analysis. In the column Nodes Moved we

capture the number of nodes removed from the graph

area to the SeNoA. Finally, the column Edges Hid-

den shows the percentage of the edges reduced from

the graph area by moving the described number of

nodes to SeNoA. It shows that by using the proposed

technique, a signiﬁcant visual clutter reduction may

be achieved.

Table 1: Several Systems with the Number of Components

and Connections.

Project Nodes Edges

Nodes

Moved

Edges

Hidden

Prj 1 304 1775 5 36%

Prj 2 1236 9393 4 27%

Prj 3 1347 11095 6 19%

SeNoA consists of a list of items. Each item con-

sists of interfaces (indicated with the mark (T) in Fig-

ure 3), nodes (U) and one corresponding symbol (S).

Interfaces are clustered into sets (T) according to the

opposite node type. Numbers beside the clustered

connections denote the number of elements clustered

in the respective symbol.

It is also possible to remove all nodes from SeNoA

area back to the diagram area by using the red cross

icon (N). Which will allow user to analyze clusters

in the graph area step-by-step rather then get con-

fused by the clutter. There is also a possibility to sort

the nodes shown in SeNoA (L). Currently, there is a

possibility to sort nodes by their names or number of

nodes contained in a group. Both mentioned sorting

possibilities could be ascending or descending. This

allows the user to ﬁnd the nodes faster in the SeNoA

as the list could easily become long.

The purpose of symbols (S) is to create a clear

and easily recognizable key, which uniquely identi-

ﬁes items within SeNoA. These symbols can be used

in the diagram area to represent connections between

a given node and the corresponding item placed in

SeNoA. They are shown as small rectangles neigh-

boring to the displayed nodes (K) and containing the

symbol, which corresponds to the connected item (S).

One can also see a group consisting of three nodes

(M). These symbols have graphics, which may hu-

mans easily remember, to help build a mental model.

IVAPP 2019 - 10th International Conference on Information Visualization Theory and Applications

334

Figure 4: Adding nodes from diagram to SeNoA groups.

The symbols allow user to be able to understand that

there is a connection between particular nodes, but not

clutter the graph by using a line.

If one clicks to the connection icons (T) of a node

in SeNoA , its connections and connected nodes be-

come highlighted by blue or red color. For the outgo-

ing connections, a blue color is used for highlighting,

red one for the opposite ones. It can be seen between

outgoing of the JH node in SeNoA (T) and node (H)

in the diagram area. Each individual node shown in

SeNoA has its own button (R) to remove it back to

its original position in the diagram area. That is use-

ful when user already understands the diagram in the

main graph area and wants to for instance identify an-

other cluster.

There could be additional information associated

with a node. We apply details on demand principle

and hide such information by default. For showing

details, the tool offers grey tool-tip (P) when explor-

ing elements.

Identifying groups of nodes represent one of the

ways of graph simpliﬁcation. Thus, their easy iden-

tiﬁcation is important for process of getting familiar

with the diagram. There is a possibility to add desired

nodes from diagram area to existing items in SeNoA

by a context menu, as shown in Figure 4. The con-

text menu will appear after a right mouse click on any

nodes in the diagram area. It contains symbols and

colors of all items included in SeNoA. After adding a

node to an item, the group is created. Users are usu-

ally creating multiple groups mainly for two reasons.

First because the nodes belong to the same seman-

tic entity such as Sprint, Release or Team. Second,

they want to just eliminate the visual clutter of nodes

added, as they are not relevant for them. A group

could be shown both in SeNoA and in graph area. So

while having multiple nodes in one group user can

still see its relations with other nodes directly in the

graph area.

4 CURRENT STATUS OF THE

TECHNIQUE

IMPLEMENTATION

We have implemented described techniques in the

web application to enable anyone to analyze data from

any domain. The tool is running on a public URL

and available for general use. The source code is

available from a GitHub repository

IMiGEr uses servlets from the enterprise Java

technology as the back-end technology. Servlets are

used mainly because of the Java implementation of

the SPaDe tool. The front-end is based on HTML5

and JavaScript. Canvas and SVG elements from

HTML5 are used to represent the nodes of the dia-

gram. We also tested several graphical frameworks,

but we concluded that their customization would be

difﬁcult because most of our present and future fea-

tures require custom implementation. That is the rea-

son why we implemented the GUI from scratch.

5 WORK IN PROGRESS

The graphs showing static relations among nodes

shown are in most of the cases growing quite rapidly

along with size and duration of a project. We are

currently working on various concepts of manual and

automated ﬁltering of relevant data. We try to apply

principles of showing details on demand and provide

different perspectives to the same data such as time

perspective. We believe such principles will help to

reduce the number of elements in the graph area to

the relevant ones only and thus dramatically reduce

visual clutter and speed-up understanding of the data

shown.

Main part of current work is dedicated to addi-

tional connected view representing the Time-line of

all collected project events. So that the user can easily

analyze the data from the time perspective and ﬁlter,

the static relations in the graph area below only to rel-

evant time window selected. Our previous experience

shows that the Time-line based visualization, com-

bined with the relations among displayed events helps

users in orientation in complex data (Lipka, 2016) .

Our implementation of the Time-line allows dis-

playing events with required duration or only points

in time. The granularity of the time may range from

minutes to years, so it can serve both for a high level

overview or for the detailed analysis of the events in

some particular part of the project. The visualization

http://relisa-dev.kiv.zcu.cz:8085/imiger/

https://github.com/ReliSA/IMiGEr

Software Engineering Projects Analysis using Interactive Multimodal Graph Explorer – IMiGEr

335

Figure 5: Connected views between static relations and Time-line visualization proposal.

also supports various event renderers, allowing rec-

ognizing different types of events. When required,

relations between events can be displayed within the

Time-line, for example to indicate the authorship of

the artifacts or inclusion into a conﬁguration. The

Time-line also allows dividing the space into so called

”swim-lanes”, in order to separate nodes types.

In the proposed visualization, Time-line is placed

above the relations graph. It contains four swim-

lanes as SPADe meta-model is classifying the soft-

ware projects elements into four categories. These are

Artifacts, People, Conﬁgurations and Tickets. Each

such element can be positioned in time when appro-

priate rules are used. For example the developer can

be displayed in the People swim-lane, for the duration

of his or her involvement in the project, the conﬁgu-

ration can be displayed as a point in time when it was

speciﬁed.

Both Time-line and static relations graph show

the requested details on demand. For instance, if

user clicks on the node Wiki in the graph area (Y)

(in Figure 5) its static relations in the graph area as

well as the corresponding records (W) (in Figure 5)

in the Time-line area will be highlighted. The same

principle applies in opposite direction, once selecting

any record on Time-line items are highlighted in the

graph area. Additionally, as was mentioned before,

the Time-line is able to show the relations of selected

record directly in the Time-line itself.

To be able to better understand static relations in

larger project one could beneﬁt from viewing some

limited time frame of the underlying data. Viewing

whole project could easily contain thousands of nodes

and edges in the graph. For instance, project manager

is very often interested in current sprint or iteration.

For such cases, the Time-line should be able to easily

use start and end date ﬁlter for nodes and edges in the

static relations graph. Such principle is illustrated in

5 by yellow time positions (marked by (Q)).

We will also continue work on automated clus-

tering techniques, which will group suitable sets of

nodes into the groups. For instance, all nodes con-

nected only to particular node could be represented as

one group (e.g., wiki pages created by one author).

Another planned visual clutter reduction tech-

nique will show them people directly in the SeNoA

after the diagram is loaded and thus reduce the num-

ber of edges in the diagram area. As people are usu-

ally representing the most connected nodes (in soft-

ware engineering projects).

We have chosen force-directed algorithms to be

used in the application after a short survey. These

features should let the user get the insight faster by

showing most connected nodes in the center of the

diagram. We also plan to evaluate above-mentioned

ideas by a case or user study. Layout reﬂecting the

IVAPP 2019 - 10th International Conference on Information Visualization Theory and Applications

336

node types into position calculations will be also in-

vestigated. We are now considering circular tree lay-

outs separating sorting the nodes on the circle based

on the nodes type.

In addition, providing software process anti-

pattern visualization is currently being researched.

Anti-pattern can be detected by SPADe and the visu-

alization can be then used to provide the exploration

of the related data. The visualization challenge in this

topic is ﬁnding the proper way of visual representa-

tion of found patterns, as they are not only referring

to the static structure or time perspective.

6 CONCLUSION

In this paper, we addressed the problem of better un-

derstanding of complex structures, events and rela-

tions in software engineering projects. There are usu-

ally thousands of elements in the such projects thus

their visualization easily becomes hardly comprehen-

sible.

We propose several closely tied visualization tech-

niques which help dealing with complexity. These are

based on the details on demand principle, information

hiding and interactivity, and are manifested as a side-

bar used for node and edge reductions, cluster iden-

tiﬁcation and hiding, edge elimination and interactive

highlighting.

We have implemented a toolchain which is able

to collect, transform and visualize data from software

engineering project repositories. The user is dealing

mainly with the interactive visualization part. It is

useful in the project analysis process when the user is

interactively getting familiar with a relations among

artifacts and people. It helps with creating the mental

model by easing the process of clusters creation.

While the data mining process is delegated to the

SPADe tool in the toolchain, IMiGEr presented here

concerns with a user interaction when analyzing a

project and getting familiar with its structure. Pre-

liminary evaluation shows that the presented ideas are

helpful in a large graph visualization, where one suf-

fers from visual clutter caused by the large number

of connection lines. The future work is to bring ad-

ditional point of view in the form of time perspective

and its interaction with existing approach. It should

bring the possibility of advanced ﬁltering and includ-

ing its automated ways.

REFERENCES

Chen, K. and Liu, L. (2003). A visual framework invites

human into the clustering process. In Scientiﬁc and

Statistical Database Management, 2003. 15th Inter-

national Conference on, pages 97 – 106.

Chiricota, Y., Jourdan, F., and Melanc¸on, G. (2003). Soft-

ware components capture using graph clustering. In

Proceedings of the 11th IEEE International Workshop

on Program Comprehension, IWPC ’03, pages 217–.

IEEE Computer Society.

Ellis, G. and Dix, A. (2007). A taxonomy of clutter

reduction for information visualisation. Visualiza-

tion and Computer Graphics, IEEE Transactions on,

13(6):1216 –1223.

Feng, Q. (1997). Algorithms for drawing clustered graphs.

Ghani, S., Elmqvist, N., and Ebert, D. S. (2012).

MultiNode-Explorer: A Visual Analytics Framework

for Generating Web-based Multimodal Graph Visual-

izations. In Matkovic, K. and Santucci, G., editors,

EuroVA 2012: International Workshop on Visual An-

alytics. The Eurographics Association.

Hachul, S. and Jnger, M. (2007). Large-graph lay-

out algorithms at work: An experimental study.

http://jgaa.info/ vol. 11, no. 2, pp. 345369.

Holten, D. (2006). Hierarchical edge bundles: Visualiza-

tion of adjacency relations in hierarchical data. IEEE

Transactions on Visualization and Computer Graph-

ics, 12(5):741–748.

Holten, D. and Van Wijk, J. J. (2009). Force-directed edge

bundling for graph visualization. Computer Graphics

Forum, 28(3):983–990.

Holy, L., Maly, I., Cmolik, L., Jezek, K., and Brada, P.

(2015). An interactive uml-like visualization for large

software diagrams. Research Journal of Applied Sci-

ences, Engineering and Technology, 11:355–371.

Lipka, R. (2016). Scalable timeline visualization. In 2016

9th International Conference on Human System Inter-

actions (HSI), pages 150–156.

Mancoridis, S., Mitchell, B. S., and Rorres, C. (1998). Us-

ing automatic clustering to produce high-level system

organizations of source code. In In Proc. 6th Intl.

Workshop on Program Comprehension, pages 45–53.

Picha, P. and Brada, P. (2016). Alm tool data usage in soft-

ware process metamodeling. In 2016 42th Euromicro

Conference on Software Engineering and Advanced

Applications (SEAA), pages 1–8.

Picha, P., Brada, P., Ramsauer, R., and Mauerer, W. (2017).

Towards architects activity detection through a com-

mon model for project pattern analysis. In 2017

IEEE International Conference on Software Architec-

ture Workshops (ICSAW), pages 175–178.

Raﬁei, D. (2005). Effectively visualizing large networks

through sampling. In Visualization, 2005. VIS 05.

IEEE, pages 375 – 382.

Srinivasan, A. and Stasko, J. T. (2018). Orko: Facilitat-

ing multimodal interaction for visual exploration and

analysis of networks. IEEE Transactions on Visual-

ization and Computer Graphics, 24:511–521.

Software Engineering Projects Analysis using Interactive Multimodal Graph Explorer – IMiGEr

337