Please disable Adblockers and enable JavaScript for domain CEWebS.cs.univie.ac.at! We have NO ADS, but they may interfere with some of our course material.

Q2 Answers

• A more general question or problem I had was the description of the different datayset types. It is understandable what the differences are, but in the end, the data is always stored in a table. So the data set type would need to be decided more on the semantics of the data in my opinion. Is that correct? 

• The principle of fields is clear to me, but when it gets to examples, I got confused. What was also not that clear is the difference between fields and spatial fields. Is there a good example for a multidimensional field?  

• What is the difference between temporal data and fields? Is temporal data not a form of a field?  

• Is it correct to say, that a grid is needed to decide a sampling strategy for field data?  

• What exactly is a tensor?  

• In the book it was written, that sometimes the user makes the decisions about the vis. When does this happen? And what is the expert there for in such a scenario?  

• What is the difference between knotted string and two-dimensional display surface of a computer screen?  

• How much practice (years) does it need, to consider to manipulate one or more pictures with interaction?  

• Randomly choosing possibilities is a bad idea because the old of finding a very good solution are very low, but how do you find the good solution in the first place? Does it need more time, or more experience about visualization?  

• Why is the line between data and metadata not clear ? And why do we distinguish between them, if it is not so clear ?  

• Regarding the wide range and the specific side of various Visualization tools, how do we decide, which one will be the best for our use ? 

• How is the data provided in trees or networks? is the data represented by a adjacency or incidence matrix or is it delivered in a table, which is showing nodes and nodes which they are connected to? 

• Interpolation between measurements can lead to a loss of information, if I'm not wrong. Therefore there could be a visualization where we interpret the data not in the correct way. So why are we doing those interpolations? I mean, we want to generate sample points between two values to provide a closer look at the data. But is it really a more detailed view if we interpolate the measurements in between?  

• In part 3.4.4 it's saying, that identifying is just targeting a single item. But what if I'm identifying the context of a visualization. What I mean is, if I identify the context or the purpose, will it be summarized?  

• I am not so clear about the data types (section 2.3) and dataset types (section 2.4) or in what respect they are defined in the book. For example why distinguish between attributes and location. In what respect is location not an attribute or what advantage does it give to distinguish between them ? Further isn't spatial information given by coordinates, so basically it is a table. I am sure it is only my computer scientist mind that sees tables and boxes everywhere but I am not aware of the reasons why to distinguish data(set) types in the way as they are presented in the book.  

• Related with that is also my surprise that date is considered an quantitative measure (Figure 2.9) since i would have had assumed by the definition that it would be ordinal. Arithmetic operations, for example 09/12/2014 + 05/28/1999, are at least unusual, unless it is more specified or some kind of transformation is happening, for example 03/21/2015 + 3 days. The author seem to notice this opinion because later she says "A temporal key attribute is usually considered to have a quantitative type, although it's possible to consider it as ordinal data if the duration between events is not interesting." 

• I don't necessarily agree with, that produce is a part of the user goals. In my understanding I would "…produce output that is used immediately, as input to a next session" position it under how, and not why.  

• Why geometry dataset doesn’t need to have attribute?  

• At Table 2.1, Name is a categorical attribute that might appear to be a reasonable key at first. on this table there are two people has the same name, and why they said it is not a good choice? because even there are more than two people with the same name, but their ID is unique.  

• On page 58, they show how to reduce the data by filtering the data and put it back together. is there any technique to filtering or reducing the data?  

• In which context can "spatial fields" be used? Examples (maybe also visual examples)  

• Are there formulas to transform data in another form? Or must the data first be interpreted?  

• What is meant with domain-specific language?  

• zu Datasets: ist es nicht so, dass multidimensionale Tables für das menschliche Gehirn sehr schwer zu verstehen sind und sollten sie dann nicht lieber komplett weggelassen werden? (Stichwort: 2.5 dimensionale Wesen) 

• Designer or User: ist die Trennung nicht eine Grauzone wenn der User auch zum Produzenten/Designer wird?  

• Wie soll man Task Abstraction unternehmen wenn einem der Überblick und Hypothese über sehr komplexe Daten fehlt?  

• Is it possible to write a D3 chart, which saves user inputs as dynamic streams and changes the configurations itself?  

• Are there any other ways to represent a tree other than using link-node diagrams, because link-node has been described as most common used way and there hasn't been given any other examples. 

• Why are the quantitative categories: interval, ratio, nominal, ordinal not useful at visual encoding, and why have online interval and ratio been mentioned?  

• Ziemlich zu Beginn des 2. Kapitels auf Seite 23 werden die im Buch diskutierten Datentypen vorgestellt. Sie beinhalten Items, Attributes, Links, Positions und Grids. Hier würde mich interessieren woher diese Datentypen abgeleitet wurden, bzw. warum man sich gerade für diese entschieden hat und wie man diese Entscheidungen begründet. Wieso wird beispielsweise „position“ als eigener Punkt genannt und nicht als eine spezielle Art von „attribut“. Während auf andere Klassifizierungen und Einteilungen(z.B.: Dataset Types, Attribute Types) ziemlich genau eingegangen wird, wirkt dieser Abschnitt für mich etwas hastig übersprungen. 

• Zur Grafik 3.1 habe ich zwar keine Frage jedoch einige Anmerkungen. Die Unterteilungen finde ich wirklich spitze und nachvollziehbar. Dieses Framework dürfte so ziemlich alles abdecken und liefert einen tollen Überblick über die Verwendung von Visualisierungen. Die Grafik dazu, in der dies visualisiert werden soll, ist meiner Meinung nach jedoch sehr dürftig und unübersichtlich. Es brauchte für mich mehrere Anläufe um die Einteilungen zu erkennen und diese auch lesen zu können. Prinzipien wie das Beachten der Data-Ink-Ratio wurden scheinbar kaum umgesetzt, weswegen die Ansicht etwas überladen wirkt um nur ein Beispiel zu nennen. Dass dies in einem Buch über Visualisierung passiert, empfand ich als ziemlich witzig, weswegen ich es auch in diesem „Report“ erwähnen wollte .  

• Ist es gewollt, dass es für node, link, array und network mehrere Worte gibt? Denn vor allem in der Informatik ist man immer erpicht darauf, dass jedes Wort genau eine Bedeutung hat, damit es z.B in der Entwicklung nicht zu Missverständnissen kommen kann.  

• Time-Varying Data bezieht sich auf ein statische Daten und nicht welche, die sich verändern, d. h. man liest die Daten ein und die Grafik/das Diagramm bleibt statisch. Die Frage kam auf, da von Sensorwerten die Rede ist und sich diese dynamisch über die Zeit verändern, sodass sich die Grafik/das Diagramm immer updaten müsste. Kann man davon ausgehen, dass vis immer mit "statischen" Daten arbeitet?  

• Was genau genau sind Strahlers Nummern?  

• The strahler number is an interesting concept. Is there a common rule how many nodes show be shown in a tree? 

• Why is - in this book - everything declared by data, if the most common and understandable type is metadata? In my mind metadata has nothing to do with creating the chart. It just helps understanding it. Did I get something wrong?  

• I have problems by imagine a network in use. There always appears a computer network in my mind, but I can't imagine that this is the only type of usage. Is there any demonstrative example on networks?  

• I didn't understand what a tensor field works or rather looks like. It is a type of data that has a strong dependancy on other types of data and can't be just shown by his own value? I can't imagine a example which helps me understand that.  

• I do not fully understand the difference between lookup and locate. Is lookup only in a tree or a network?  

• It's hard for me to understand what a field really is. I have the conept of forms with fields in mind which does not really fit. Does it mean the information is shown via locality? 

• Dynamic Dataset: I was wondering how the user would interact with a visualization which uses a dynamic dataset. Which changes to the visualization are necessary to NOT confuse the user with the dynamic changes? How much harder is the development of the actual visualization? Are there any tools? Is it wise to update the data on the fly or use intervals?  

• Present: What is a good/common way to present a visualization to a group of persons/users which are not familiar with the data? Is it better to start with static vis and more explanation or dynamic vis and "let the data speak"?  

• Tools: Suppose you have the task of doing a visualization. You have your prototypes ready (on- paper) and have chosen a tool to do the implementation. After several hours of poor frustration you're finally at the point were you want to surrender. Is it wiser to change the prototypes or change the entire framework/tool and start from scratch with the implementation?  

• p. 32 (Categorial Data vs. Hierarchical Data): Often, a hierarchy is implied by the ordering of categories (f.e. in a bar chart that has no intended ordering, it could look like the categories that are shown first are more important than the ones that are shown later). How can one make sure that the ordering is understood to be arbitrary?  

• Ad task abstraction (a more general question): When (after how many iterations) is a visualization considered to be “done”? (Is that ever the case, or can a visualization rather be conceived as a process that never ends rather than a product?).  

• Ad user vs. designer (a more practical question): When there are conflicting options (f.e. shall we use blue or green as a color for indicating something in our vis) – how is the decision made and by whom? What if the designers and users have a completely different background: how does the mediation work? Is this communicative part considered to be part of the visualization process? If not, who is concerned with it?