• Share
  • Email
  • Embed
  • Like
  • Save
  • Private Content
SLA Nov2009 Public
 

SLA Nov2009 Public

on

  • 1,247 views

Overview of Information Visualization. ...

Overview of Information Visualization.
Goal and “Soft” vs. “Hard” Definition
Human Visual Perception
Key InfoVis Design Principles
Visualizing Hierarchical Data: Treemaps
InfoCrystal | MetaCrystal | searchCrystal
Overlap between Search Engines | Wikipedia
Other Search Visualization tools
Data Visualization
Display Types | Baby Names | Gapminder | NY Times Visualizations
Visualizing Library Data
Google Motion Charts | ManyEyes | Tableau
Gigapixel Visualization
Gigapan | Photosynth | whereRU

Statistics

Views

Total Views
1,247
Views on SlideShare
1,247
Embed Views
0

Actions

Likes
2
Downloads
0
Comments
0

0 Embeds 0

No embeds

Accessibility

Categories

Upload Details

Uploaded via as Microsoft PowerPoint

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment
  • The titles of the 230 most popular Wikipedia pages are submitted as queries to the major search engines, where Google is used by more than 47.5 percent of people to search the Web, followed by Yahoo with 28.1 percent and MSN with 10.6 percent (comScore, 2007b). The queries were formulated so that special characters, such as colons or brackets, and words, such as "list of", were excluded from the queries, since people are not likely to include them in their search requests. The queries were submitted on 18 February 2007 and the positions of the Wikipedia pages that “originated” the queries were determined in the top 10 search results for each engine and query, respectively. Investigate where precisely a popular Wikipedia page is placed in the search results if you formulate a query that is identical to the title of the Wikipedia page. This will help to understand how specifically search engines drive traffic to Wikipedia.
  • Search result placement of websites can have profound effects on their success and profitability A consequence of placing Wikipedia pages in the top search results is that other websites are pushed further down in the result lists, and thus the probability that these sites are seen is decreased.
  • This triangular area, referred to as the “ Golden Triangle ” , was looked at by 100 percent of the participants in their study. The fact that Google, the dominant search engine, places popular Wikipedia results toward the very top of its result lists, which users pay most attention to and tend to click on, can help explain the popular and persistent content categories of the most visited Wikipedia pages. Besides paying users to click on a link, search result placement is the most powerful way at disposal of a search engine to promote specific web pages.

SLA Nov2009 Public SLA Nov2009 Public Presentation Transcript

  • Information Visualization Course Information Visualization Anselm Spoerri PhD MIT Rutgers University [email_address]
  • Talk Outline
    • Information Visualization
      • Goal and “Soft” vs. “Hard” Definition
      • Human Visual Perception | Key InfoVis Design Principles
      • Visualizing Hierarchical Data : Treemaps
    • InfoCrystal  MetaCrystal  searchCrystal
      • Boolean Queries | Overlap between Search Engines | Wikipedia
      • Other Search Visualization tools
    • Data Visualization
      • Display Types | Baby Names | Gapminder | NY Times Visualizations
    • Visualizing Library Data
      • Google Motion Charts | ManyEyes | Tableau Visualization
    • Gigapixel Visualization
      • Gigapan | Photosynth | whereRU
  • Goal of Information Visualization
    • Use human perceptual capabilities to gain insights into large data sets that are difficult to extract using standard query languages
    • Exploratory V isualization
      • Look for structure, patterns , trends, anomalies, relationships
      • Provide a qualitative overview of large, complex data sets
      • Assist in identifying region(s) of interest and appropriate parameters for more focussed quantitative analysis
    • Shneiderman's Mantra:
      • Overview first , Zoom and Filter , then Details-on-Demand
  • Visualization - Problem Statement: “ Soft” vs. “Hard”
    • Data | Map | Scientific Visualization
      • Use familiar visual displays suited for up to 3 data dimensions
      • Show abstractions, but based on physical space
      • Data has inherent 1-, 2- or 3-D geometry
      • MRI : density, with 3 spatial attributes, 3-D grid connectivity
    • Information Visualization
      • More than 3 data dimensions to be shown
      • Tabular, Networked, Hierarchical, Textual data …
      • Information has no obvious spatial mapping
    • Fundamental Problem
      • How to map non–spatial abstractions into effective visual form?
  • "Spatial" Data Displays IBM Data Explore r http://www.research.ibm.com/dx/
  • Abstract , N-Dimensional Data Displays
    • Chernoff Faces
  • Data Types and Visual Marks
    • Data Types
    • Numerical (can perform arithmetics)
    • Ordinal (obeys ordering relations)
    • Categorical (equal or not equal to other values)
    • Visual Marks
      • Points (position, color, size)
      • Lines (location, length, width, color)
      • Areas (uniform / smoothed shading )
      • Volumes (resolution, translucency)
  • Human Visual Perception – Pre-Attentive Processing
    • How many 3s ?
    0 8 0 2 8 0 8 5 0 8 0 8 3 0 8 0 2 8 0 9 8 5 0 - 8 0 2 8 0 8 5 6 7 8 4 7 2 9 8 8 7 2 t y 4 5 8 2 0 2 0 9 4 7 5 7 7 2 0 0 2 1 7 8 9 8 4 3 8 9 0 r 4 5 5 7 9 0 4 5 6 0 9 9 2 7 2 1 8 8 8 9 7 5 9 4 7 9 7 9 0 2 8 5 5 8 9 2 5 9 4 5 7 3 9 7 9 2 0 9
  • Human Visual Perception – Pre-Attentive (Color Pops Out)
    • How many 3s ?
    Pre-Attentive Demo by Christopher Healey 0 8 0 2 8 0 8 5 0 8 0 8 3 0 8 0 2 8 0 9 8 5 0 - 8 0 2 8 0 8 5 6 7 8 4 7 2 9 8 8 7 2 t y 4 5 8 2 0 2 0 9 4 7 5 7 7 2 0 0 2 1 7 8 9 8 4 3 8 9 0 r 4 5 5 7 9 0 4 5 6 0 9 9 2 7 2 1 8 8 8 9 7 5 9 4 7 9 7 9 0 2 8 5 5 8 9 2 5 9 4 5 7 3 9 7 9 2 0 9
  • Human Visual Perception – Motion Perception
    • Structure from Motion
      • Kinetic Depth demo
      • Anthropomorphic Form from Motion demo
  • Visual Perception Design  iPod Ads
    • Which Human Visual Capabilities are exploited in iPod ads ?
  • Visual Perception Design  iPod Ads
    • Which Human Visual Capabilities are exploited in iPod ads ?
  • Visual Perception Design  iPod Ads
    • Which Human Visual Capabilities are exploited in iPod ads ?
  • Visual Perception Design  iPod Ads
    • Which Human Visual Capabilities are exploited in iPod ads ?
     Structure from Silhouettes, Shading and Motion  Perception of Causality
  • Human Visual Perception – Key for Visualization Design
    • Visual System Detects CHANGES + PATTERNS
      • Luminance Channel More Important than Color
    • Stages of Visual Processing
      • 1 Rapid Parallel Processing
      • Slow Serial Goal-Directed Processing
    • Pre-Attentive Features
      • Position
      • Color
      • Simple Shape = orientation, size
      • Motion
      • Depth
    Proximity Similarity Continuity Symmetry Closure Figure + Ground Gestalt Law
    • Depth Cues
      • Occlusion
      • Relative Size
      • Motion Parallax
      • Binocular Disparity
      • Shape from Shading / Contour
  • Information Visualization – Key Design Principles
    • Abstraction
    • Direct Manipulation
    • Immediate Feedback
    • Linked Displays
    • Dynamic Queries
    • Overview  Zoom+Filter  Details-on-Demand
    • Focus + Context
    • Animate Transitions
    • Increase Information Density
  • Abstract  Hierarchical Information – Overview Treemap Traditional ConeTree SunTree Botanical Hyperbolic Tree
  • Hierarchical Data – Treemap  Space-Filling Design
  • Visualizing Hierarchical Data – Treemap
  • Treemap  Smart Money Map of the Market
  • Treemap  Newsmap
  • Treemaps – Examples
    • University of Maryland developed treemap
      • http://www.cs.umd.edu/hcil/treemap/
    • SmartMoney
      • http://www.smartmoney.com/marketmap/
    • Newsmap
    • http://newsmap.jp/
    • Honeycomb
    • http://www.hivegroup.com/
  • InfoCrystal  MetaCrystal  searchCrystal
    • InfoCrystal
      • Visualize Relationship between Search Results  Visualize Power Set
  • Goal – Compare Search Results
    • Show Overlap  Venn Diagrams
    How to Better Visualize ?
  • Goal – Compare Search Results Can be Generalized to N Sets Transform Explode
  • InfoCrystal
  • InfoCrystal
  • InfoCrystal A and (not (B or C)) A and C and (not B) A and B and C B and C and (not A) A and B and (not C) C and (not (A or B)) (not (A or B or C)) B and (not (A or C))
  • InfoCrystal  MetaCrystal  searchCrystal
    • InfoCrystal
      • Visualize Relationship between Search Results  Visualize Power Set
    • MetaCrystal
      • Visualize Overlap between Result Sets from Different Search Engines
      • Authority Effect : The more engines that find a result, the greater its probability of being relevant.
      • Ranking Effect : The higher up a result is placed and the more engines that find it, the greater its probability of being relevant.
  • MetaCrystal – Guide Users Toward Relevant Results
    • Visual Coding
    • Shape = Number of Engines
    • Colors = Which Engine Combination
    • Orientation = Which Engine Combination
    • Size = Number of Docs or Rankings by Different Engines
    • Visual Grouping
    • Location = Number of Engines Increases Toward Center
    • Containment = Docs found by Same Number of Engines Cluster in Same Ring
    • Proximity = Docs with Similar Rankings Cluster
  • MetaCrystal – Query Formulation
  • MetaCrystal – Cluster Bulls-Eye
    • Radial Mapping
    • Radius = Total Ranking
    • Angle Reflects Rankings
    • Size = Rankings by Engines
    Google Teoma AltaVista Lycos MSN
  • InfoCrystal  MetaCrystal  searchCrystal
    • InfoCrystal
      • Visualize Relationship between Search Results  Visualize Power Set
    • MetaCrystal
      • Visualize Overlap between Result Sets from Different Search Engines
      • Authority Effect: The more engines that find a result, the greater its probability of being relevant.
      • Ranking Effect: The higher up a result is placed and the more engines that find it, the greater its probability of being relevant.
    • searchCrystal
      • Compare, remix and share results from web, image, video, blog, tagging, news engines, Flickr images or RSS feeds.
      • Embed on a web or blog page
      • Demo of Most Visited Wikipedia pages (Sept 06 – Jan 07)
  • Visualization of Most Visited Wikipedia Pages
    • What is Popular on Wikipedia? Why?
      • 1 Visualize Popular Wikipedia Pages
        • Overlap between 100 Most Visited Pages on Wikipedia for September 2006 to January 2007 (using WikiCharts data)
        • Information Visualization helps to gain quick insights
        • Demo
      • 2 Categorize Popular Wikipedia Pages
      • 3 Examine Popular Search Queries
      • 4 Determine Search Result Position of Popular Wikipedia pages
      • 5 Implications
  • Category View
  • Information Visualization – Insight Gained
    • Category View
    • 40% of a month’s top 100 pages are visited in all five months
    • 25% are highly visited only in a single month
    • Top Pages in a Month
    • Sept 2006 – Death of Steve Irwin by Stingray
    • Oct 2006 – Halloween and North Korea
    • Sept 2006 – Sacha Baron Cohen (Borat) and Thanksgiving
    • Dec 2006 – Edvard Munch (Scream Logo) and Pearl Harbor
    • Jan 2007 – Pres. Gerald Ford buried and Deaths in 2007
    • Oct + Nov + Dec – Borat and Albert Einstein
    • Nov + Dec + Jan – James Bond (Casino Royale released Nov)
  • Cluster Bulls-Eye
  • RankSpiral
  • Information Visualization – Insight Gained
    • Cluster Bulls-Eye
    • Sexuality large % in all months
    • Spiral View
    • Entertainment becomes more frequent away from display center
    • Geography specific countries or places
    • Politics political figures
    • History World War I, World War II, Vietnam War Iraq War not incl.
    • Smaller Percentage of Popular Wikipedia pages related to Typical Encyclopedic Topics Than Expected
  • Search Result Position of Popular Wikipedia Pages 87% of Popular Wikipedia pages about Sex  Top 3 Google Result Positions Gain Insights into how specifically search engines fuel growing popularity of Wikipedia
  • Discussion
    • Top 100 Pages – Percentage of Total Daily Page Views
      • #1 = 13.5%
      • #2–#99 = 8%
    • List of Popular Search Queries = Edited, Sanitized
    • Wikipedia “Top 100” = Unfiltered, Real-time
      • view into what people are searching for on the Web.
  • Implications
    • Search Result Location, Location, Location
    • Fierce Competition to be placed in the search positions that have People’s Attention
    • Game of “Musical Chairs”
      • Three highly desirable chairs
      • Seven chairs of decreasing attractiveness
      • Millions websites wish to be placed on these ten chairs
    • Wikipedia in Top 3 Search Results  Increases Competition
  • Implications
    • EyeTools conducted Eye Tracking study
    • “ Golden Triangle of Search ” – users pay most attention to the triangle at the top of the search results page, which includes the top three results and top ads on the right.
  • Implications
    • Inclusion of Wikipedia pages in the top search results:
    • “ Win-Win” for Search Engines
      • Search engines recommend a Wikipedia page that comes from a trusted source
      • Wikipedia = “Safe Bet”
      • Increases Competition for Limited Resources search result positions that people notice
      • Increases Need to Purchase Ads to be noticed
  • Search Visualization - Clustering
  • Search Result Visualization  2.5D Maps Grokker Kartoo ThemeScape / Aureka
  • Search Result Visualization  3D Themescapes
  • Search Visualization – Tools
    • Quintura http://www.quintura.com/
    • AquaBrowser http://demo.aquabrowser.com/?q=visualization
    • Kartoo http://kartoo.com/flash04.php3
    • Viewzi http://www.viewzi.com/search/whitevoid-photocloud/
    • SpaceTime http://www.spacetime.com/
    • Aduna / Vound http://www.vound-software.com/
    • Closed Down
      • Grokker
      • SearchMe
  • Data Visualization – Common Display Types
    • Common Display Types
      • Bar Charts
      • Line Charts
      • Pie Charts
      • Bubble Charts
      • Stacked Charts
      • Scatterplots
  • Data Visualization – Baby Names Voyager http://www.babynamewizard.com/voyager
  • Data Visualization – Gapminder – Google Motion Charts http://www.gapminder.org/
  • New York Times Visualizations
    • New York Times has Visualization Team
      • http://www.nytimes.com/2009/01/19/business/media/19askthetimes.html
      • Leading adopter of visualization tools in the news media
      • At the forefront of enhancing new stories with interactive content
      • VizLab in Collaboration with ManyEyes http://vizlab.nytimes.com/
      • Community API http://open.blogs.nytimes.com/2008/10/30/announcing-the-new-york-times-community-api/
    • Combining Static Infographics and Interactive Visualization
      • Remade in America http://projects.nytimes.com/immigration/
      • Debt Trap http://nytimes.com/interactive/2008/07/20/business/20debt-trap.html
      • Election 2008 http://elections.nytimes.com/2008/results/president/map.html
  • Visualize Library Data
    • Helping RU Libraries Tell Its Story
      • Current MLIS InfoVis class working together on large-scale project
    • Visualization Tools Being Used
      • Google Motion Charts - Free
        • http://documents.google.com/support/bin/answer.py?hl=en&answer=91610
      • ManyEyes - Free ex1 ex2
        • http://manyeyes.alphaworks.ibm.com/manyeyes/
      • Tableau Visualization - Trial ex1 ex2 ex3
        • http://www.tableausoftware.com/
  • Visualizing Library Data – ManyEyes
  • Visualizing Library Data – Tableau Visualization
  • Gigapixel Visualization
    • Motivation
      • How to represent a large number of spatially related images
    • “ Infinite Zooming”
      • Seadragon DeepZoom
    • Gigapan
      • Stitching multiple images into image pyramid
    • Photosynth
      • Linking images based on pattern matching
    • whereRU: Virtual Experience of Rutgers
  • “ Infinite Zooming” – Seadragon’s DeepZoom
    • http://seadragon.com/
  • Gigapan – stitching multiple images into image pyramid
    • Gigapan  Google Earth  “Streetview” Anywhere
  • Photosynth – linking images based on pattern matching
    • http://whereru.rutgers.edu/photosynth.html
    • whereRU : Virtual Experience of Rutgers using Gigapans & Photosynths
  • Some Resources
    • Information Visualization Course
      • http://comminfo.rutgers.edu/~aspoerri/Teaching/InfoVisOnline/Home.htm
    • Blogs
      • Infoesthics
        • http://infosthetics.com/
      • Visual Complexity
        • http://www.visualcomplexity.com/vc/
      • Flowing Data
        • http://flowingdata.com/  
      • Smashing Magazine
        • http://www.smashingmagazine.com/2008/01/14/monday-inspiration-data-visualization-and-infographics/