Representing Software Project Vision by Means of Video:
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
| Title |
| |
| Subtitle |
| |
| Author | 0000-0001-5336-6899 (ORCID) | |
| Contributors | 0000-0001-5336-6899 (ORCID) | |
| License | CC Attribution 3.0 Germany: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor. | |
| Identifiers | 10.5446/54717 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language | ||
| Producer | 0000-0001-5336-6899 (ORCID) | |
| Production Year | 2021 | |
| Production Place | Celle |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
| |
| Keywords |
00:00
Streaming mediaSoftwareMachine visionBendingEndliche ModelltheorieInformationStreaming mediaProjective planeSoftwareModel theoryMachine visionArithmetic meanXMLUML
00:15
Normed vector spaceSoftwareStreaming mediaComputer configurationSystem programmingPhysical systemEndliche ModelltheorieSoftware engineeringHybrid computerMetric systemBroadcasting (networking)Hausdorff dimensionEuler anglesContent (media)CodecMachine visionModel theoryRepresentation (politics)Stability theoryFocus (optics)Helmholtz decompositionHierarchyProcess modelingArc (geometry)Complete metric spaceEquals signUsabilityLogistic distributionBinary fileLinear regressionCompact spaceMachine visionProcess (computing)Streaming mediaHelmholtz decompositionModel theoryMachine visionSoftwareProcess (computing)Dependent and independent variablesCharacteristic polynomialProjective planeGoodness of fitPhysical systemEndliche ModelltheorieComputer configurationProduct (business)Stability theoryContent (media)Focus (optics)Point (geometry)Complete metric spaceView (database)Linear regressionYouTubeThree-dimensional spaceInformationDimensional analysisMappingCombinational logicRepresentation (politics)ResultantFlow separationFile viewerLevel (video gaming)Plotter1 (number)Presentation of a groupFile formatMachine visionTerm (mathematics)Observational studyArchaeological field surveySoftware developerTwitterImplementationMereologyIdentifiabilityValidity (statistics)Generic programmingSubsetNumeral (linguistics)Order (biology)AdditionMultiplication signCodeElement (mathematics)Visualization (computer graphics)Context awarenessCondensationCycle (graph theory)Adaptive behaviorCASE <Informatik>IterationSet (mathematics)DivisorBasis <Mathematik>TelecommunicationExtension (kinesiology)Latent heatVelocityGroup actionProcedural programmingClient (computing)Maxima and minimaPhase transitionTheorySound effectCoordinate systemPersonal digital assistantRight angleTablet computerTotal S.A.HierarchyRevision controlNumberPrice indexMedianEstimatorShooting methodComputer animation
Transcript: English(auto-generated)
00:00
Hello and welcome to my talk with the title representing software project vision by means of video a quality model for vision videos. My name is Oliva Kavas and I'm working at the Leibniz Information Center for science and technology. So first of all let's start about the topic requirements communication for shared understanding. We all know we have two sides
00:23
the client and the suppliers and each of the persons involved has its own mental model about the future system we want to develop. In general the process of coordination and communication serves to develop or negotiate a shared understanding about the future system.
00:41
So at the end this also is in accordance with the goal of the requirements engineering we want to establish a shared vision about the future system in the hands of all involved people. When we want to support requirements communication for shared understanding
01:01
there are several different aspects according to the theory of shared understanding on how we can support the achievement of shared understanding and one of these aspects is use of documentation that supports proactive communication for shared understanding. So when we look at the
01:25
maxima representation we see at the lowest level we have paper this includes textual and pictorial artifacts which usually have a low richness and low effectiveness but are frequently used in RE.
01:40
On the other side we have the videos and as you can see they have a high richness and high effectiveness for communication. However when we consider literature we can see that they're infrequently used in RE. So now let me briefly introduce you about the topic of vision video.
02:03
Vision video is a kind of video in RE. We can define it as a video that represents a vision or parts of it for achieving a shared understanding among all parties involved by disclosing discussing and aligning the mental models of the future system. There are some frequently
02:22
well-known vision videos like the Apple Knowledge Navigator from 1987 which shows a tablet as a voice assistant which we all know at this time by the name Siri, Alexa or also Google. Further vision videos can be found in the EU science and innovation channel as this is hosted
02:43
since 2017 and this youtube channel already includes 381 vision videos of research and innovation projects. So we see a new trend that vision videos appear more often. However we want
03:01
to investigate the problem or the topic of using vision videos more closely and here in 2017 we conducted a survey on the topic videos as a documentation option in RE and one of the findings of the survey was that software professionals lack knowledge and skills to produce and use good
03:20
videos. So this leads us to the question what constitutes a good video and we considered literature and found out that the term video quality is a rather ill-defined concept due to numerous technical and subjective characteristics. So in 2018 we proposed the
03:42
idea of learn from the discipline video production on how to produce good videos by conducting a kind of literature study on numeric video production guidelines in order to develop some kind of quality model for vision videos which is inspired by
04:04
our well-known software quality model. So in particular what we've done and what I'm presenting in the following the process that we have the idea to consider the vision video as a part of the video and it's as a representation format and the vision as its content and so we did
04:22
two literature reviews first one on generic video production guidelines with the idea to identify quality characteristics of videos and also the related steps in the video production and use process but the second part on the vision literature we wanted to identify characteristics
04:40
of visions so that we can use those findings together in combination to provide a quality model for vision videos and we considered two representations the first one is a hierarchical decomposition in order to understand what is vision video quality in terms of its characteristics
05:00
and maybe some sub-characteristics. As a second part we wanted a representation along the video production use process so that we can see when can which characteristic be influenced in order to give some kind of guidance. At the end we validated our proposed quality model for visual videos in an experiment in academia the details will follow. So first of all due to
05:26
the time I can only briefly show you the idea of two literature reviews we found in the six generic video production guidelines and used a manual coding process consisting of two coding cycles with several iterations in order to find the video characteristics and on the other part
05:44
the process steps. If you want to get some more detailed information about this process I would refer to the discussion at the end. We did a similar manual coding process for the topic of software project vision in order to identify the vision characteristics. So what are the results
06:04
at the end and the results are two separate quality models one for vision and one for videos. We arranged the identified characteristics and sub-characteristics based on the three dimensions of a quality model. Consider the representation of an artifact of product
06:21
its actual content and its impact on an emotional dimension. So how can we get a combined quality model and this is simply a merge first of all of the different dimensions to propose a quality model for vision videos. And here in the first dimension we have the
06:44
sensorial characteristics combining video stimuli and also the aspect of focus a compact representation of a vision. The other ones are video characteristics that focus mainly on the presentation format. For the content dimension we have at the lowest levels two vision
07:06
characteristics for completeness and clarity and for further video characteristics I do not want to prevent all of them in detail but for example the plot considers the constructed representation of the content or the prior knowledge asked to consider what's the
07:25
prior knowledge a viewer needs to understand the content of the video. And at the third dimension we can see that there are the emotional characteristics so for example the pleasure of the regarding the enjoyment of watching a video but also for vision
07:43
support or stability that people accept the vision they see or that this vision is consistent over time and do not change. So this one was the hierarchical decomposition of the quality model but we also wanted another representation along the video production and use process
08:05
and for this we have the idea to count how often a specific coded element was also related to a specific step of the video production and use product that consists of for faces pre-production shooting post-production and viewing and based on these numbers we developed
08:22
a so-called indicator of impact based on the median to say the first rough estimation how strong and specific characteristic can be affected in a specific video production step. So at the end we have here on the right side the again hierarchical decomposition and now we
08:44
can map these characteristics onto the video production use process seeing that for example image quality has can can strongly be affected during the part of the shooting and the post production but since the video is already produced no longer not easily at the face of
09:05
the viewing. For the vision characteristics we could not do similar mapping so these are only proposed phases where we should consider the respective characteristics
09:21
in the respective steps of the production of the video production use process. So first of all since the vision is the main content of the video all characteristics are important during the pre-production and especially during the post-production we could affect the focus of the compact representation the clarity of the content and also its completeness
09:44
when we added the video and in particular the support through the acceptance of the vision and its stability at the end when we create the video are important factors during the viewing phase. So now let me start with the validation of our proposed quality model in particular we
10:05
had a research question regarding how do the individual quality characteristics at the lowest level of the quality model relate to the overall quality of the vision video from a developer's point of view we choose this point of view since we need a specific audience and developers
10:21
are one of the groups that can be a audience of the vision video. In order to set the end the velocity we defined two criteria and we said there should be relationships with all three dimensions of the quality model and we should also identify relationships with vision and video
10:41
characteristics otherwise our quality model would not cover a vision video. For our procedure we had an introduction with 139 participants and all of them were active developers at the time we showed them the vision videos. In particular we had the viewing phase with all of them in one
11:04
room and after the viewing of one video they performed an assessment of the 15 quality characteristics and whether they perceived the vision video as good or bad and repeated this process in total with eight vision videos so in the end we had a data set of the 952 assessments
11:24
of 119 subjects. To investigate the relationships between the individual characteristics and the overall quality we performed binary logistic regression and our dependent variable was the perceived overall quality coded as good or bad and for the independent variables we used
11:44
the assessments of the 15 video of the vision video quality characteristics. I do not want to show all of our findings in detail how and the entire process but the main findings is that we found six characteristics at the end with a relationship between the
12:06
individual characteristics and the overall perceived video quality and for example what we can see here is regarding the aspect of the prior knowledge in the left corner
12:23
and we can see that the more the people perceive that prior knowledge is necessary to understand the video the lower is the probability that the vision video was perceived as good. On the other hand in the top right corner you can see the characteristic pleasure
12:40
so the enjoyment of watching the video and you can directly see the more likely the people rated that they enjoyed the video the higher was the probability that the video was perceived as good. For the further discussions of the details I would still
13:01
if you're interested refer to the discussion afterwards. So what can we find or can we say as an answer to a research question we found that six out of 15 quality characteristics are related to the overall quality of the vision video perceived by developers and what we can say in a more general way the better the implementation of the respective quality
13:23
characteristic was perceived by the subjects the higher was the probability that the overall quality of the vision video was perceived as good. In addition we found relationships meet the criteria for validity and soundness of the quality model so we can say that the
13:43
quality model of vision videos contains at least a subset of relevant characteristics however we cannot conclude that the other nine characteristics do not or are not relevant for vision videos and their perceived quality. If you want to know the further details
14:02
why specific characteristics could not or did not have an impact on the overall quality based on our findings and our result from the experiment I would refer to again to the journal article where we discuss this aspect in detail. So in general the insights of our
14:23
experiment are not really that the individual relationships are important but there exists relationships between the characteristics and our proposed quality model and vision video quality that is perceived by the audience. So we can conclude based on our results that
14:42
the quality model is first of all according to our definition valid and sound and that's the basis for future refinements and extensions. So finally I want to conclude this talk and also highlight some of our planned future work. So what we've seen we've developed a
15:03
hierarchical decomposition of our quality model producing videos to understand what is vision video quality and we also met these individual characteristics along the steps of the video production and use process to understand when can which characteristic be influenced.
15:27
While this part provides first awareness about the topic of video quality and also gives some kind of first initial guidance we want to consider the question for future work how do I need to proceed as a software professional when I want to produce the video in order to accomplish
15:46
specific results and also explain why should I as a software professional follow this respective step. So our idea is to develop some kind of condensed guideline for vision videos
16:02
in order to support their production. The first step we've already done is that we used our manual coding and the extracted text passages and air processed these elements in a more structured
16:20
manner to provide a kind of recommendation that explains how should I proceed to receive a specific result so answering the question how and what and we also provide a rationale so answering the why should we proceed with this specific recommendation.
16:44
Due to the mapping and the coding we were also able to highlight the respective process set and the characteristics and this first guideline for generic videos was basis for our adaption to vision videos and for this purpose we refined the recommendations
17:02
by adapting them to the topic of vision videos and also highlighted in case of several steps the respective one where we perceive based on our experience which in which steps we can mainly benefit from the given recommendation and here we also highlight
17:20
based on which recommendation from the generic video production guideline we adapted this specific one for our vision videos. So first version of these two guidelines for the generic videos and also for the vision videos is published on archive and since this is also a topic
17:42
or was the topic of my PhD dissertation it is also further elaborated there. So the idea of these different questions is to provide some kind of awareness for software professionals to understand what is vision video quality and how can I influence these specific characteristics
18:03
so the idea is beside these awareness aspect we also want to provide them guidance in order to enable software professionals to produce these videos if they want to. Since the first article of this presentation is a part of my PhD dissertation as I previously mentioned and you're
18:25
interested into the topic I would like to highlight you my dissertation you can get which further elaborates on the topic of supporting requirements communication for shared understanding by applying vision videos and requirements engineering. Thank you for your interest in a
18:43
talk and have a nice day.