Questions for 4-19

Few

Rating:  4

1) How do we balance ensuring that we present our information in multiple formats for higher level consumption with the issue he raises about our brains having a limited ability to hold multiple items in our minds at the same time? Look at Fig. 35.5. Here is a visualization that represents information in multiple formats, helping the reader to better understand the underlying data. But images such as this can sometimes be intimidating; a quick glance can make the display seem complicated, even if it actually simplifies the data upon closer inspection. When it comes to data visualization how much is too much and how much is not enough?

2) Do the principles relating to human perception that Few discusses hold true for different types of learners? Some people learn better by seeing or doing or hearing – would the principles of visualization discussed here apply equally to all these groups? Looking at his early example in figures 35.1 and 35.2 isn’t it possible that someone who works well with numbers and learns by doing would see figure 35.1 and compare the values of the numbers in their head almost automatically? Not that this would negate anything Few says. Such a person would still probably glean the same information more quickly from the graphical figure, but for such a person looking at figure 35.1 can we really say “this table fails” as Few does?

3) The node and link visualization can be an excellent tool for showing relationships between entities other than variables. The social media example used in the article is a good one. But I worry that we embrace it so readily because it looks nice. Is this really an effective method for demonstrating relationships? Looking at the figure can you tell me Amanda’s relationship to Nick? How about Scott’s relationship to Ken? Or Jason’s relationship to Christine? All of this is on the chart, but because the chart is cluttered with so many nodes it is not readily apparent (at least, it isn’t to me). To quote Few’s introduction, “’a picture is worth a thousand words’ - often more - but only when the story is best told graphically rather than verbally and the picture is well designed.” How could this visualization be better designed to overcome some of its weaknesses?

Icke

Rating:  3

1) Does it seem strange to anyone else that Icke spends 6.5 pages talking about the system aspects of VA, and crams user aspects and human-machine collaboration aspects into less than one page total while trying to tell us that user input and system computing power need to be balanced?

2) Icke says that the correct algorithms must be chosen for the dataset with which you are working but he doesn’t really explain how this is done. What factors are important to consider when deciding on analytics? What are the available options when it comes to analytics? There is a lot of information in the paper about the options available when it comes to visualization types, but it is quiet about the analytics themselves. Since various types of data and visualizations are discussed wouldn’t it make sense to expect a cursory discussion of available analytical tools?

3) What would Few say about this article? Would he agree with the visualizations Icke uses? Do they incorporate aspects of human perception? Looking at figure 35.9 from Few’s article, are Icke’s visualizations in-balance or out-of-balance when looking at the brain functions of seeing vs. thinking?

Lam et al.

Rating:  2

1) What is the difference between the process-oriented CTV scenario and the visualization-oriented UP scenario? They both seem to be evaluating how a program conveys information and how that information is received. It just seems to me that these types of evaluations would have a lot of overlap.

2) Why are the visualization scenarios so much more common than the process ones? I understand that some of it comes from the traditions of the HCI and CG fields, but the process scenarios are certainly not unknown since they’re 15% of the evaluations looked at in this study, and if they truly provide profitable information wouldn’t companies be interested in exploring those options? I wonder if the methodology of this paper had been changed to include papers from more than the 4 sources listed if that would have changed the percentages significantly.

3) Each year the number of papers that include evaluations increases. The authors mention that, according to the review by Barkhuus and Rode, the quality of these evaluations may have remained the same. What is the impact of static evaluations on the future development of visualization resources? Are the current evaluations sufficient or should we be investigating new methods that could spark developers to reach for a higher standard?

Questions for 4-12

Mislove, et al.

Rating: 3

1) Do people behave differently in a social network that is based off of sharing a particular type of media (photos in Flickr, videos on YouTube, etc.) as opposed to social networks specifically built for connecting socially(like Orkut)? If so, is this change in behavior significant? How would it manifest itself in this study?

2) The authors mention that their study will be of import to fields outside of just computer science. They mention sociology specifically. But who else benefits from this study? Since we are reading this article we can safely assume that iSchools would be interested, but what about business? Economics? Humanities? How would they find value in this study? What other fields are being overlooked?

3) The article’s methodology was a little different for Orkut than for the other sites because Orkut does not export an API for external developers. This meant that the information had to be gathered a little differently and the authors explain that this will have skewed the results slightly for Orkut. I get that partial BFS crawls have shown sampling bias in other instances, but couldn’t at least some of the differences between Orkut and the other sites stem from the differences between Orkut and the other sites? Orkut’s main purpose is social networking; the other sites are focused on blogs or photos or videos. Orkut has a user base that is largely Brazilian and Indian; the other sites are more globally balanced. Couldn’t these differences account for some of the differences in the results?

Bizer et al.

Rating: 2

1) Maybe I’ve just missed the point, but why is RDF format so much better than using traditional HTML documents? I get that RDF gives additional information about the relationship between two things, but can’t that relationship often be understood without being explicitly stated?

2) As demonstrated in the article, linked data has the potential to really improve several services/apps we use on-line and on mobile devices. How would linked data improve apps like: eReading technology? Games? Schedulers? Financial planners? The list goes on.

3) Is linked data really the next big thing? Or is it more of a buzzword that people get excited about today, only to forget about tomorrow like their now forgotten Palm Pilots and MySpace accounts? Sure there are advantages to linked data over standard HTML style browsing, but will people be willing to put forth the effort to get there?

I Horrocks

Rating: 4

1) How does WordNet fit into the semantic web? Horrocks talks about an ontology that would allow us to include the term SnowyOwl within the larger classification of Owl, and Owl within Raptor, and so on. Isn’t this closely related to WordNet? Is this a main motivation for WordNet being developed?

2) Horrocks tells us on page 6 that “query answering in OWL is not simply a matter of checking the data, but may require complex reasoning to be performed.” Should OWL and RDF searches ever really begin to take off wouldn’t this extra reasoning mean longer processing times on searches? Will people be willing to wait the extra time for the process to run if it means they get highly integrated answers from Linked data? Or will they stick to their old, tried-and-true methods that maybe aren’t as detailed in their response but can process much faster?

3) Future directions include continuing progress in ontology alignment – where ontologies whose domains overlap are reconciled and all are the better for it. Does this mean that the eventual end goal is one large ontology that covers everything? All subjects are going to overlap somehow into other subjects and everything can be connected in this way. How would it change the semantic web if we had an ontology covering almost every topic ready to go today?

Questions for 4-5

Golder, Huberman

Rating: 3

1) The largest category of tags from the study was those that identified what or who the URL in question was about. Rather than represent a need for extensive tagging could this represent a need for better title metadata? Even if this isn’t the case, how can we limit recording redundant information as we tag?

2) Golder talks about some of the strengths and weaknesses of tagging versus taxonomies. Has anyone ever tried to combine these systems? Could we eliminate having to search in as many areas of taxonomies if we could retrieve results from multiple areas of the taxonomy were tagged similarly? Would this simply result in the same problems we already experience?

3) Golder tells us that users are extremely varied in their tagging behavior. Looking at those who use many tags compared to those who use only a few are there trends in the types of tags used? Do those who only use a few tags use more personalized tags as opposed to descriptive tags? Or is there no correlation at all?

Marshall, Cathy

Rating: 4

1) Just to play devil’s advocate for a moment – why do we even care about the metadata for all these photos, most of which are intended for personal use? We recognize that we can’t possibly archive everything so who cares about the tags or description associated with the photo of Mr. John Tourist standing on bull testicles?

2) How would knowing the tendency for different kinds of metadata (place, artifact, story) to appear in different metadata types (title, tags, captions) help programmers working with image retrieval? Would it make their algorithms more efficient? What other related fields, like IR, would find this study relevant to their work?

3) Marshall mentions the problem associated with some user’s reluctance to refer to the bull’s testicles.  When relying on the masses to provide metadata how can we account for tendencies such as this where people edit their input based on personal beliefs, biases, opinions, etc.?

Kling

Rating: 2

1) Kling talks about the productivity paradox, explaining that the introduction of computers did not increase productivity as promised. What technologies today are new and promising, and making claims of revolutionizing our lives? Is there anything to separate these claims from those made in the 60s-80s about computing? Or will technology today leave us unsatisfied as well? Is technology destined to fall short of expectations? If not, how does it avoid that pitfall?

2) What was it about The Electronic Journal of Cognitive and Brain Sciences (EJCBS) that caused it to have poorer results than Electronic Transactions on Artificial Intelligence (ETAI)? Are the differences between these journals social or technological? In these journals, and in other settings as well, does social interaction influence technology or is it the other way around?

3) In section 6 “Why Social Informatics Matters” Kling begins by stating, “Social informatics research pertains to information technology use and social change in any sort of social setting, not just organizations.” How would Kling explain the benefits of social informatics to a religious leader? A university professor? A political candidate?