<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.0 20120330//EN" "JATS-archivearticle1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink">
  <front>
    <journal-meta />
    <article-meta>
      <title-group>
        <article-title>Process Attribute Visualization in 3D and Virtual Reality</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <string-name>Manuel Gall</string-name>
          <email>manuel.gall@univie.ac.at</email>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <contrib contrib-type="author">
          <string-name>Stefanie Rinderle-Ma</string-name>
          <email>stefanie.rinderle-ma@univie.ac.at</email>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <aff id="aff0">
          <label>0</label>
          <institution>University of Vienna, Faculty of Computer Science</institution>
          ,
          <addr-line>Vienna</addr-line>
          ,
          <country country="AT">Austria</country>
        </aff>
      </contrib-group>
      <abstract>
        <p>Being able to visually explore process attributes and their values supports process analysts in process understanding and optimization. The complexity of the analysis can range from a few to a variety of attributes, e.g., machining times and sensor parameter in the manufacturing domain. This paper introduces an innovative prototype to visualize process attributes within a 3D representation. The 3D representation can be displayed and explored on a monitor or a virtual reality system.</p>
      </abstract>
      <kwd-group>
        <kwd>Process Attribute Visualization</kwd>
        <kwd>Process Model</kwd>
        <kwd>3D Representation</kwd>
        <kwd>Virtual Reality</kwd>
      </kwd-group>
    </article-meta>
  </front>
  <body>
    <sec id="sec-1">
      <title>Introduction</title>
      <p>
        More and more data from various sources such as databases, services, and
realtime information is present in nowadays business processes. Incorporating as
much information as available into a process seems a good thing at rst.
However, companies need to understand their processes to achieve their business
goals. In order to support the analysis in such cases we use process models for
visualization. Such visualizations often comprise a plethora of attributes, e.g.,
machining parameters in production processes. These attributes are crucial in
understanding processes models. Various approaches e.g. views and abstractions
[
        <xref ref-type="bibr" rid="ref1 ref5">5, 1</xref>
        ], where introduced to deal with processes containing multiple attributes.
Usually these approaches abstract from the attribute data. These abstractions
lead to a loss of information and certain relations between attributes cannot be
identi ed anymore.
      </p>
      <p>To reduce the amount of information being lost as part of the visualization
process we developed a prototype called 3DViz. This prototype allows to
incorporate multiple attributes within one representation by using the additional
z-axis.
Analyzing and understanding a business process can be a complex and
overwhelming task. Some visualization approaches only show the control ow
elements and hide process attributes such as time within menus. However,
attributes can yield crucial information for process analysts. The most commonly
visualized attribute is the Role attribute, however there are certainly many more
relevant attributes within business processes e.g. data elements and resources.
These attributes can have di erent manifestations, e.g., cost, time and location.
These manifestations are not prede ned; they can be di erent in each process
model. One model uses only cost while another one requires a new attribute
and the next model has three or more attributes. This even gets worse as those
attributes might possess di erent data types like string or numerical values. Our
prototype allows for visualizing multiple attributes within a single
representation.
2.1</p>
      <sec id="sec-1-1">
        <title>3D Process Visualization</title>
        <p>
          We introduce a novel 3D Process Visualization prototype called 3DViz. 3DViz
and an overview on the design choices are explained in the following section,
details can be found in the technical report [
          <xref ref-type="bibr" rid="ref2">2</xref>
          ]. The 3DViz concepts are
prototypically implemented1. The prototype supports the visualization of process
models with multiple attributes.
        </p>
        <p>3DViz allows for visualizing process models within a 3 dimensional space,
hence allowing for utilizing an additional axis for process model visualization.
The 3D models enable positioning of activities on the x-axis, y-axis, and z-axis
and scaling of activities on these three axes. For activity positioning, 3DViz is
capable of attribute visualization on y-axis and z-axis. X-axis is restricted to a
processes control ow.</p>
        <p>
          Currently, we support up to 5 di erent attributes within a process model.
We could utilize more visual styles e.g., brightness, texture, and orientation
to incorporate more attributes within one visualization. However, the principle
of graphic economy [
          <xref ref-type="bibr" rid="ref3">3</xref>
          ] describes that up to 6 categories per variable can be
discriminated. Our design choice was to utilize the styles positioning, scaling,
colour, and shape. This seems to be the best trade-o between graphic economy,
perceptual discriminability, and visual expressiveness.
        </p>
        <p>In order to provide complexity management mechanisms we allow the user
to customize the amount of attributes and the visual styles used. Numerical
attribute values can be mapped to every axis for positioning and scaling. String
values are restricted to positioning within swimlanes on the y-axis and z-axis.</p>
        <p>Figure 1 depicts a sample process with 3DViz. When analyzing a process
there are two important axes, the positioning axis and the scaling axis. In this
example positioning axis in Fig. 1 1jdepicts Roles as swimlanes on the z-axis
and Cost on the y-axis. Positioning Cost on the y-axis leads to a visualization
where activities with high costs are displayed above activities with lower costs
2j. The scaling axis is visualized with colors 3j. This axis shows that Time
is scaled on the x-axis and Data Usage is scaled on the y-axis. 4jdepicts an
activity scaled on the y-axis. This activity possesses the highest amount of Data
Usage. Additionally a room (e.g., a seminar room or an o ce) is created around
the process model to ease navigation and orientation while exploring the model.</p>
        <sec id="sec-1-1-1">
          <title>1 https://bit.ly/2wTT5za</title>
          <p>In 3DViz, control ow elements (cf. Fig. 1) are designed similarly to their
2D counterparts in BPMN. The motivation behind this decision is to provide a
familiar setup to reduce training e orts.
2.2</p>
        </sec>
      </sec>
      <sec id="sec-1-2">
        <title>VR Visualization</title>
        <p>
          Literature suggests that viewing a 3D model on a monitor might not be an
appropriate way to transport the visual delity introduced by a 3D model [
          <xref ref-type="bibr" rid="ref4">4</xref>
          ].
Therefore we opted for extending 3DViz beyond monitors. We allow to view and
interact with 3D models through an HTC Vive VR device.
        </p>
        <p>From a visual point of view the model is represented in the same way as the
3DViz model. For this reason we want to refer to Figure 1 for a process model
visualization example. The main di erence is the interaction with the model.
While 3DViz requires mouse and keyboard input for interactions, e.g., camera
movement, the VR approach allows camera movement when the VR device is
moved, i.e., moving the head forward while wearing the device will translate to
movement within the virtual world.</p>
        <p>Within 3DViz a room is created around the process model. For the VR
approach the visualization of a room o ers additional bene ts. A room might
help to avoid dizziness and disorientation when exploring the process model.
In its current state the applicability of VR/AR is poor compared to Desktop
systems. However, when considering the development in recent years we believe
VR/AR will advance further and be better integrated in the normal day routine
of people.
2.3</p>
      </sec>
      <sec id="sec-1-3">
        <title>Visualization Con guration</title>
        <p>
          In the previous sections the visualization capabilities of the prototype are brie y
discussed and depicted in Figure 1. Figure 2 depicts the con guration menu. Nine
di erent styles for attribute visualization are available. For each style the user
can select an attribute and a mapping e.g. on the z-axis the role is visualized with
discrete mapping. For the attribute mapping we allow for 3 di erent variants,
direct mapping, relative mapping and discrete mapping [
          <xref ref-type="bibr" rid="ref2">2</xref>
          ].
At an public event called "Lange Nacht der Forschung" the 3D attribute
visualization was demonstrated. Based on a physical Lego showcase we visualized the
according process instances with respect to their attributes.
        </p>
        <p>
          Additionally we conducted a survey addressing applicability and usability of
3D visualizations. From the 42 participants 39 (92.8%) answered that 3D
visualizations are useful for process model visualization. Further we asked which
areas are particularly interesting when 3D visualizations are used. In total 33
(80.5%) participants answered that combining multiple attributes is an
interesting eld. Allowing the user to change the viewing angel an rotate around the
process model was also very well received by 31 (75.6%) participants. However,
to our surprise only 11 (26.8%) thought that 3D representations are particularly
useful to omit overlapping edges. More insights on survey design, demographic
data, results and discussion can be found in our technical report [
          <xref ref-type="bibr" rid="ref2">2</xref>
          ].
        </p>
        <p>
          An evaluation targeting the performance of 3DViz was conducted [
          <xref ref-type="bibr" rid="ref2">2</xref>
          ]. The
data ranged from simple graphs 2 Activities and 1 Control-Flow-Element element
till large graphs with 1024 Activities and 512 Control-Flow-Elements. We saw
a linear growth in time needed to draw graphs. This is a very good result as
large graphs e.g. 1024 Activities and 512 Control-Flow-Elements only take 2.2
seconds for drawing. In all our examples we used 5 attributes to be visualized.
For the creation of these models we used a random process model creation tool
2. This tool can be used to create large process models for demo visualizations.
4
        </p>
      </sec>
    </sec>
    <sec id="sec-2">
      <title>Conclusion and Outlook</title>
      <p>3DViz o ers an innovative approach to visualize process models and their
attributes. Attributes can be visualized by scaling or positioning of nodes. Scaling
and positioning on the x-axis, y-axis and z-axis allows to incorporated
multiple attributes within one visualization. Integrating multiple attributes leads to
faster recognition of associations between attributes.</p>
      <p>We are currently working on expanding 3DViz capabilities towards mining
and monitoring. Further we want to explore the visualization of multiple
processes in combination with constraint visualization.</p>
      <p>The prototype, a tutorial, a short video and further information is available
on our website3.</p>
      <p>Acknowledgment This work has been funded by the Vienna Science and
Technology Fund (WWTF) through project ICT15-072.</p>
      <sec id="sec-2-1">
        <title>2 http://gruppe.wst.univie.ac.at/ gallm6/ProzessGenerator/</title>
        <p>3 https://bit.ly/2wTT5za</p>
      </sec>
    </sec>
  </body>
  <back>
    <ref-list>
      <ref id="ref1">
        <mixed-citation>
          1.
          <string-name>
            <surname>Bobrik</surname>
            ,
            <given-names>R.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Reichert</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Bauer</surname>
            ,
            <given-names>T.</given-names>
          </string-name>
          :
          <article-title>View-based process visualization</article-title>
          .
          <source>In: Business Process Management</source>
          . pp.
          <volume>88</volume>
          {
          <issue>95</issue>
          (
          <year>2007</year>
          )
        </mixed-citation>
      </ref>
      <ref id="ref2">
        <mixed-citation>
          2.
          <string-name>
            <surname>Gall</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Rinderle-Ma</surname>
          </string-name>
          , S.:
          <article-title>Visualizing multiple process attributes in one 3d process representation</article-title>
          .
          <source>Tech. Rep</source>
          . arXiv:
          <year>1903</year>
          .
          <volume>00283</volume>
          ,
          <string-name>
            <surname>arXiv</surname>
          </string-name>
          (
          <year>2019</year>
          )
        </mixed-citation>
      </ref>
      <ref id="ref3">
        <mixed-citation>
          3. Moody, D.L.:
          <article-title>The physics of notations: Toward a scienti c basis for constructing visual notations in software engineering</article-title>
          .
          <source>IEEE Transactions on Software Engineering</source>
          <volume>35</volume>
          (
          <year>2009</year>
          )
        </mixed-citation>
      </ref>
      <ref id="ref4">
        <mixed-citation>
          4. Schonhage, B.,
          <string-name>
            <surname>van Ballegooij</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          , Elliens, A.:
          <article-title>3d gadgets for business process visualization a case study</article-title>
          .
          <source>In: Virtual Reality Modeling Language</source>
          . pp.
          <volume>131</volume>
          {
          <fpage>138</fpage>
          . VRML '
          <volume>00</volume>
          (
          <year>2000</year>
          )
        </mixed-citation>
      </ref>
      <ref id="ref5">
        <mixed-citation>
          5.
          <string-name>
            <surname>Smirnov</surname>
            ,
            <given-names>S.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Reijers</surname>
            ,
            <given-names>H.A.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Weske</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Nugteren</surname>
            ,
            <given-names>T.</given-names>
          </string-name>
          :
          <article-title>Business process model abstraction: a de nition, catalog, and survey</article-title>
          .
          <source>Distributed and Parallel Databases</source>
          <volume>30</volume>
          (
          <issue>1</issue>
          ),
          <volume>63</volume>
          {
          <fpage>99</fpage>
          (
          <year>2012</year>
          )
        </mixed-citation>
      </ref>
    </ref-list>
  </back>
</article>