9.19 Respond To Kosara Schneiderman Fry Here

Transforming Data: Cultural Strategies in DataMining
Instructor: George Legrady
Posts: 160
Joined: Wed Sep 22, 2010 12:26 pm

9.19 Respond To Kosara Schneiderman Fry Here

Post by glegrady » Tue Sep 13, 2011 9:23 pm

To Transforming Data Students. Post your responses to this week's readings here. It may be best to first write them in a text editing program, and then to copy/paste here.
Last edited by glegrady on Fri Oct 07, 2011 8:01 pm, edited 1 time in total.
George Legrady

Posts: 8
Joined: Tue Sep 13, 2011 11:01 am

Response - David Lazarus

Post by lazad518 » Mon Sep 19, 2011 6:54 pm

Visualization Criticism - Kosara

The importance of criticizing and critiques in the art world is absolutely immense. Whether it was in high school studio art class or in a design feedback session in an advertising agency, critiques are always the most important and key moments in a creative process.
Furthermore, Kosara raises a very valuable point when he stresses the importance of students learning how to criticize in order to improve their aptitude of verbalizing their thoughts about art. I think that is an extremely important aspect of becoming a better artist and a respectable professional. There is an unnerving amount of people, students and professionals, that don't know how to clearly express their thoughts with regards to creative and artistic process, which usually voids their feedback as it doesn't make much sense. Thus wasting time during critiques or during creative process jams.

The Eyes Have It - Schneiderman

Schneiderman's 6 step process of delving into a data visualization was extremely interesting. I think I would have thought of only half of these elements, if I were to consider the important steps of a Data Viz. The importance of Relate and History seem more of a 'Usability' aspect of Data Viz creation, but they are vital as we make Data Visualizations to simplify information and to allow users to use these maps in easy and efficient ways.

Visualizing Data Ben Fry

I thought the outline of the creative process in this chapter was extremely useful. As Fry explains, with this process he allows the roles of graphic designers and statisticians to merge. He focuses therefore on the viewing of the data, and not so much on the tools used. He see beyond the process or the skills of the creator, and focuses on the end product, which is essentially the most important.
Within the 7 step process, one really struck me. The Refine part of the process was rather novel to me, but key. If you eliminate the Refine, then one is simply finding a new set of data that is being shown in a known way, it is a known representation of unknown data. However, by adding the extra step of Refining your visualization, you allow yourself to find new data, represent it, and then push it to the next level so that it is truly new and different, and can hopefully offer a new vision for the viewer.

User avatar
Posts: 11
Joined: Tue Sep 13, 2011 11:13 am
Location: brooklyn, ny

Response - Matthew Willse

Post by matthew » Mon Sep 19, 2011 7:12 pm

The Eyes Have It - Shneiderman

Schneider provides a good framework to consider while creating a visualization. While the terms are more dry and mechanical than the visualizations we hope to make, they offer a useful set of building blocks. For interactive applications, the list of user tasks give us clear targets to consider while we may otherwise be lost in data, math, or aesthetic decisions.

Most interesting to me was his discussion of history. In most web and application development, we think about a user origins and destinations, and aim to create clear paths between them. Schneiderman reminds us that exploring information is not a linear process -- no more so for the user than the visualization designer. He says:
It is rare that a single user action produces the desired outcome. Information exploration is inherently a process with many steps, so keeping the history of actions and allowing users to retrace their steps is important.
7 Stages to Data Visualization - Ben Fry

Ben Fry outlines the stages often used to create an visualization. He is careful to define these elements while recognizing that the exact process will vary between projects. He makes several observations and suggestions that could serve people well in many fields beyond design and data.

He notes that technological advances make it easier to disassociate from the meaning behind our tasks. In data visualization, the data might become severed from the questions that originally encouraged us to collect and dissect the data. Lose the narrative and information becomes noise.

In my consulting work to help public interest advocates, my central task was often to dissuade my clients from writing more and shouting louder. Instead, I urged them to make strategic statements to people with leverage. Fry eloquently argues that more “is not implicitly better, and often serves to confuse the situation.” He offers a reprise of the adage less is more -- “Perhaps making things simple is worth bragging about.”

Visualization Criticism - Kosara

As an undergraduate, I studied sculpture in a BFA program where I became very familiar with the role of critique in the creative process. Since that time, it has informed not only my process as a designer, but has shaped my approach to problem solving in general. The idea of bringing criticism to the discipline of data visualization as a means to advance the field is interesting. It can not only improve the work of individuals, but advance the profession by creating stronger tools for evaluating and understand the visualizations and their impact. The idea is one that makes so much sense it seems obvious and almost silly to discuss.

User avatar
Posts: 14
Joined: Tue Sep 13, 2011 11:10 am
Location: New York, NY

Re: 1. Respond To Kosara Schneiderman Fry Here

Post by deklerk » Tue Sep 20, 2011 4:52 am

There is an assumption that I would like to discuss in Ben Shneiderman proposition of a task based classification system in the paper, “The Eyes Have It: A Task by Data Type Taxonomy for Information Visualizations.” The proposition implies that a clear perception of information visualization comes with orienting information with human agency. In commercial applications, data visualization is part of “decision support systems,” and Shneiderman’s Visual Information Seeking Mantra is found on “dashboards” - the interface by which managers steer their businesses.

Overview first, zoom and filter, then details-on-demand. Let’s review what the mantra may look like in operation today. First: the “Overview first”. Seeing the aggregate data of an organization - sales, opportunities, inventory, etc - as from a distance, a summary perspective, presents the organization as an objective system. From the overview the user might see in which parts of the country sales are strong and where they are weak. Sales are weak in Ohio? Now we “zoom and filter”, getting closer to the problem. Here we see that one sales district is responsible for the lacklustre sales. We see the relationship right on the map. With a double click of the district, “details-on-demand” retrieves the name and contact information of the district manager. We give them a call, taking note of this action in order to remember it, to follow up on it, to evaluate it as an action that had measurable impact on improving sales.

Data visualization brings to management information systems not only a glass by which to see the organization through zooming, focusing, and filtering, but also a mirror in which to perceive our place in the organization. The clarity of tasks perceived through reflection, the security of agency optically stabilized.

By proposing to classify data types by tasks, Shneiderman foregrounds data to the background of action. Zooming, relating, extracting, etc - the seven tasks become tangible actions in the mediated environment of the visualization. The data types that Shneiderman cycles us through to illustrate the tasks are navigable spaces and theatres for action. Thus we might scroll through 1-dimensional data to find items with certain attributes, zoom and drill down into the details of mapped 2-dimensional data, positioning and orienting modeled objects in 3 dimensions.

The challenge, I think, for data visualization practitioners is to consider the affects of data visualization. Shneiderman makes his proposition in response to anxieties over information overload and the want for “smoother integration of technology with task”. He attempts to balance the chaotic experience of information with a sense of order clarified by the task. Has the perception shifted? Today we hear concerns of another risk: of filter bubbles - data views too narrow, insulating the user negatively from a variety of information. And what of the overwhelming expansion of management information systems - are the failures of the past few years to foresee the economic crisis unrelated to the visualization shift in organizational management? How many business leaders and managers sat confidently behind dashboards as their businesses crashed and the economy tanked? The eyes may have it, but to what degree are the eyes being had?

Posts: 9
Joined: Tue Sep 13, 2011 11:09 am

Alessandra's Response

Post by alessandrarosecampos » Tue Sep 20, 2011 1:41 pm

One interesting theme that I found across the texts that were read for class this week was the issue of viewership. What audience is the visualization intended for and how might this audience interpret the visualization? At the beginning of the first article, “Visualization Criticism”, the authors are speaking about peer criticism within the art school setting. They acknowledge that there is a “shared body of knowledge that’s assumed within the group, and comments by an uninitiated outsider, while sometimes providing an interesting new perspective, can also be irrelevant or even disruptive to a shared train of thought.” This observation is helpful within a university class dealing with data visualization, for example our Transforming Data class. As the article mentions, critiques can often devolve into expressions of personal opinion and preference or a defensive diatribe by the creator of the work. This is not very helpful within a classroom setting, as students are meant to base both the work and the critiques on the shared body of knowledge that is both presumed (from previous learning and experience) and learned within the class.

However, this approach may be problematic when applied to real world visualizations. Unlike art for art’s sake, data visualizations often serve a practical purpose. For example, the authors mention an experimental visualization piece at Brown University, which based a practical visualization of the university bus schedule on the abstract paintings of Piet Mondrian. While the visualization was aesthetically pleasing, and was able to be used practically by the creators, the experiment failed because outsiders (visitors to the cafeteria in which the visualization was displayed) were unable to identify the piece as a visualization and instead identified it as an art object. While the cafeteria goers may have been uninitiated into the nuances of visualization or artistic classroom culture, their reaction was far more helpful than it was disruptive to the artist’s creation. Whether or not the visualization was beautiful, or reminiscent of Mondrian, it did not hold up practically in a real world setting.

Ben Schneiderman additionally points to this issue in his exploration of a data type taxonomy of data visualizations. In speaking about an interactive map based on visual markers of U.S. cities, he states that a foreigner who is unfamiliar with U.S. topography may do better with an alphabetical list of city names. So, the artist creating such a visualization must take into account who is most likely to be utilizing his visualization. Is he a U.S. native who will find visual representations of familiar landmarks helpful in navigating the visualization, or is he a foreigner who may benefit from a textual representation of the location? Is it possible to embed textual information within the visualization to accommodate both parties (for example in the “details on demand” portion of Schneiderman’s ideal visualization)? Ben Fry places similar emphasis on the functionality of the visualization, taking care to steer the reader away from using stock visualizations. For him, it is important that each project be designed to fit the unique situation—how is it being used? Who is the intended audience? I find that this issue (who is this information intended for and how does this impact the information presented, as well as how it is presented?) is one that is applicable to all information within the university setting (in both “artistic” and “theoretical” classrooms). However, it is often ignored and information is taken at face-value. The practical aspect of data visualization makes this question about information more readily obvious and applicable, however, such questions are important across disciplines.

Posts: 6
Joined: Tue Sep 13, 2011 11:01 am

Christine Zenyi Lu

Post by luc393 » Thu Sep 22, 2011 5:36 pm

In the “Thinking About Data” section of Ben Fry’s Visualizing Data, he mentions AOL’s release of “randomized” search queries of its users. He does this to point out that we rarely think about the immense amount of information that can be gleaned from data. Immediately, I thought of a project by Sander Plug and Lernert Engelberts entitled “I Love Alaska: The heartbreaking search history of AOL user #711391.” http://www.minimovies.org/documentaires ... lovealaska In 13 episodes, each clocking under 5 minutes, the directors use the Internet search queries of user #711391 to tell a compelling narrative of a middle-aged woman in Alaska who may have health issues about to embark on an extra-marital affair. Through queries such as flight information, symptom checks, and Internet dating, we learn the intimate and incredibly personal details of one woman’s life. In the same vein, the New York Times was able to uncover one woman’s identity, Thelma Arnold (User No. 4417749) through an analysis of the leaked AOL search queries: http://www.nytimes.com/2006/08/09/techn ... wanted=all. The New York Times writes, “…the detailed records of searches conducted by Ms. Arnold and 657,000 other Americans, copies of which continue to circulate online, underscore how much people unintentionally reveal about themselves when they use search engines — and how risky it can be for companies like AOL, Google and Yahoo to compile such data.”

Ben Fry writes, “All data problems begin with a questions and end with a narrative construct that provides a clear answer.” What strikes me as amazing is the narrative structure inherent in the information that we unintentionally reveal through the collection of data. Ben Fry starts off his book with a quote: “The greatest value of a picture is when it forces us to notice what we never expected to see.” The challenge is to know what to look for. With Schneiderman’s “Visual Information Seeking Mantra”, we are given an equation to hone in on what matters: Overview first, zoom and filter, then details-on-demand. Kosara steps in and breaks it down into visual representations that span from the pragmatic to the sublime with varying forms of connection and purpose. The common thread that runs through these three articles is the need to ask the right questions and in doing so, reveal hidden truths to identify the patterns in what would otherwise be a random collection of a complex mess of meaningless data.

Posts: 13
Joined: Tue Sep 13, 2011 10:58 am

Pedagogical Approaches to Visual Info. - Andrew Bowe

Post by bowea324 » Fri Sep 23, 2011 8:08 am

Within Robert Kosara’s introduction to ‘Visualization Criticism’, Kosara points at the need for a tight bond between a community-based pedagogy in the process of constructing new forms of visualization. Here, Kosara argues that criticism is ‘a tool for pointing out and learning from mistakes. In a visualization context, critiquers can function as an impromptu test group assessing qualities of organization and usability’ (Kosara 2008). One might interpret Kosara’s understanding of critique not first and foremost as a reading of the work – as close-reading in Barthes or in most hermeneutical approaches to works of art—but rather as something that happens between the relation of an artist and an audience (classmates) that alters the outcome of a work. In moving away from a critique that relies on the content (final product), one becomes primarily pedagogical (socially) as an artist as they are open to social response that moves their work forward, rather than informing the viewer of their relation to the work (as an end product), or more historically to ask the reader to develop a reading of a concrete or stabile work. In this sense the interaction and production of art recognizes the potential of a maximum of interactivity from the beginning stage through the final product.

While this philosophy becomes important in the process of constructing a work that is either user friendly or enhanced democratically, one might question the capacity of the artist to produce visualizations that contradict the audience’s previous conceptions of works of art. If the role of the artist is to undo or question commonplace responses, it might be best to keep the production of the work under locked doors. At times the production of visualization(s) ought to be private in order to ensure that that certain unique vision are not lost for commonplace and/or democratic influence.

In Ben Schneiderman’s categorical approach in ‘The Eyes Have it: A Task by Data Type Taxonomy for Information Visualizations’ he takes the interactive nature of visualizations one step further by exploring ways in which the interactivity of visualizations must be a part of the most basic elements of design (by type: 1-D, 2-D, 3-D, Tree, etc.). Schneiderman writes, ‘…products will need to provide smooth integration with existing software and support the full task list: Overview, zoom, filter, details-on-demand, relate, history, and extract. These ideas are attractive because they present information rapidly and allow for rapid user-controlled exploration’ (Schneiderman 2009). In this sense, one might see Schneiderman’s text as more grounded in an exchange between a designer (worker) and a consumer/client (or potential consumer/potential client). That is because the process of construction cyclically consults the public’s demand for a feasible interaction. Though, one might also see Schneiderman’s analysis as giving the client more of a capacity to sort through material and to add to the structure of material without losing sight of the organization or ‘zoomed out’ perspective of what the product (or data type) might be.

Posts: 10
Joined: Tue Sep 13, 2011 11:02 am

Re: 1. Respond To Kosara Schneiderman Fry Here

Post by putzb642 » Fri Sep 23, 2011 3:04 pm

Visualization Criticism

Kosara does a good job pointing out the thought and the theory behind traditional critiques for artists from both the academic setting. It is important for us (by us I mean creators and theorists) to understand what are the ideas we are looking for and critiquing. Data visuals must be seen and critiqued in a very specific way as opposed to designs solely in an academic setting. It is important to understand the points of reference and make sure that it be being seen in the proper context. It is possible to have a successful visualization in a different setting than it is tackling but it must be clear that there are both design properties and informational properties instead of emphasizing one or the other.

Visualizing Data

Fry goes on to explain what is required of a data visualization and the context behind and why it is important. It is important to use a specific process to begin to make any sense of a large data set: acquire, parse, filter, mine, represent, refine, and interact. Once the monotony of following the processing of collecting and mining is complete you can than iterate through what you are trying to display in different methods. Each adding specific agency to what you are trying to convey. Pointing out the requirement for what makes a successful visualization stands out of course, and in some way is a no brainer. Orson Welles once stated “The enemy of art is the absence of limitations.” This stands true with all projects whether they are linear based or dynamic in nature. It is important to know that a meaningful experience is more often than not created by have a strict set of constraints which force the creator to think deeper into how anything can be successfully conveyed.

Task by Data Taxonomy

Schneiderman discusses in his paper the taxonomy of seven different data types. One, two, and three Dimensonal data, temporal and multi-dimensional data, and tree and network data. Moving forward from there it is crucial to use advanced filtering to specify which data would work best for the given need on that project. Much of the information that is collected would not be possible without the computer. It is both a tool for the creation and distortion, as well as a collector and sorter.

User avatar
Posts: 10
Joined: Tue Sep 13, 2011 11:00 am

Re: 1. Respond To Kosara Schneiderman Fry Here

Post by keats047 » Sun Sep 25, 2011 6:14 pm

Didactic Data - Stephen Keating

I think that one of the first questions everyone asks when examining a piece of data is "how is it useful?" But when I think about this question, it leads me to yet another question. What determines usefulness? It is a rather important question when attempting to understand didactic information. There's nothing inherently useful about any information we can be given, rather that information allows us to make decisions based on perceptions. These perceptions relate both to our own biases towards information generally as well as the information we're being given specifically. When I read about criticism and best practices for approaching data, it seems a bit awkward, and indeed there are instances of cognitive dissonance on how best to use the various pieces of data that any given person is bombarded with on a daily basis.

Below are some of my more personal thoughts on data and our usage of it in America more specifically, rather than the more general approaches of criticism and construction.

Data is something we, as Americans, have fetishized. There is a cultural construction, somewhat Randian in approach, that if we just have enough information, enough pieces of the puzzle, that we will reach a utopian freedom, all watched over by an ever-growing interconnected system. A beating heart that literally pulses information to us, perhaps at one point, through us--to be considered, consumed, and likely soon thereafter, thrown aside. Considering data as beautiful is likely something of little import, which means that information is less alive and malleable than it is a finished product. America, and Americans, expect the result rather than the potential, belieivng they themselves to be the potential, putting upon the data the burden of proof as to their mettle. Which makes sense particularly if you look at the standardized system of education, all of which is ruled by numbers and grades (merely a mask to conceal numbers).

In this way, we use data to discriminate negatively upon people, despite its potential to be used otherwise. While such degrading systems are somewhat ubiquitous, to attempt to reach the realm of a beautiful data visualization, it is my belief that the data being used must be used with caution, particularly as it has a great possibility to propagandize against the group or against the individual. While making a beautiful visualization is important, we must not confuse what that beauty intends to represent. For simple purposes such as maps, data serves a simplistic role, but when it serves a more complex role, such as tracking people or determining who is a "good" citizen and who is not, data can be divisive and dangerous.

Posts: 9
Joined: Tue Sep 13, 2011 11:10 am

Re: 1. Respond To Kosara Schneiderman Fry Here

Post by beth.c.carlson » Mon Sep 26, 2011 9:24 am

In The Eyes Have It, Ben Schneiderman discusses seven different data types: one dimensional, two dimensional, three dimensional, temporal, multi dimensional, tree, and network. It is important to understand the user, and what one wants to extract from the data set, in order to choose the best-fitted data type approach. I found Schneiderman's example of the different ways to visualize a map of the United States helpful in understanding this concept. Someone familiar with the geography of the country can find a city and click on it in order to access tourist information, but a foreigner who may know the name of a city, but not where it is located, would be better suited by a scrolling list of US cities. And a two dimensional visualization, like a map, is better for visualizing proximity and navigating paths, whereas a tree data type helps to visualize relationships, as in a family tree.

I found this chapter in Visualizing Data helpful. Ben Fry discusses the various questions and concepts that need to be tackled before creating a data visualization. Understanding the planning process is as important as the creation of the visualization itself. Is the data set fixed, or is it live, and constantly changing? What is the question you are attempting to answer with the visualization? Could the data set potentially reveal information other than what you are trying to express through the visualization? Is that potential information harmful or embarrassing to anyone? Fry explains how the path to creating a visualization is a multi-disciplined process, and planning is a large part of that process.

Kosara's article discusses the benefits of criticism in art and teaching, as well as information visualization. The example of the bus route visualization in the style of a Mondrian painting was helpful in illustrating that fact that art and information visualization are connected, but they are not the same. There are key differences between the two that cause each to be viewed in different ways. A bus map in the style of a art painting, and in the context of a cafeteria, does not read as a bus map. This point can be added to Ben Fry's planning process outlined in chapter one of Visualizing Data. Context and viewer perception are additional factors that must be considered before creating a visualization.