CS 160 Lecture 24 Professor John Canny Spring

CS 160: Lecture 24 Professor John Canny Spring 2003 12/7/2020 1

Preamble 4 Quiz on last lecture. 12/7/2020 2

Information Architecture 4 The OAI model (lecture 20) was our starting point for information organization. 12/7/2020 3

Information Tasks 4 Specific Fact-finding: * Find the phone number of Bill Clinton 4 Extended Fact-finding: * What kinds of music is Sony publishing? 4 Open-ended browsing: * Is there new work on voice recognition in Japan? 4 Exploration of availability: * What genealogy information is at the National Archives? 12/7/2020 4

Database queries * Query languages like SQL are widely used, but are hard to learn and easy to make mistakes with. SELECT DOCUMENT# FROM JOURNAL-DB WHERE (DATE >= 1994 AND DATE <= 1997) AND (LANGUAGE = ENGLISH OR FRENCH) AND (PUBLISHER = ASIS OR HFES OR ACM) 12/7/2020 5

Visual Query Builders 12/7/2020 6

QBE: Query By Example 4 User chooses a record (Database) or document (search engine) and specifies “more like this”. 4 User can also pick a segment of text, even a paragraph, from a good document and use it as a search query (search engines only). 12/7/2020 7

Visualizing Search Results 12/7/2020 8

Multidimensional Scaling 12/7/2020 9

Multidimensional Scaling 4 Multi-Dimensional Scaling (MDS) is a general technique for displaying n-dimensional data in 2 D. 4 It preserves the notion of “nearness”, and therefore clusters of items in n-dimensions still look like clusters on a plot. 12/7/2020 10

Multidimensional Scaling 4 MDS applied to hand-classified discussion topics. 12/7/2020 11

Multidimensional Scaling 4 Clustering of the MDS datapoints (discussion topics) 12/7/2020 12

Discussion 4 Try to assign labels to (positive and negative) X and Y axes in the previous plot. 4 Note that X and –X may not be opposites in the usual sense – * this is an artifact of linear projection methods. 12/7/2020 13

Multidimensional Scaling 4 MDS can be applied to search engine results easily because they automatically have a high -dimensional representation (used internally by the search engine). 4 The MDS plot helps organize the data into meaningful clusters. You can search either near your desired result, or scan for an overview. 12/7/2020 14

Tasks for a visualization system 1. 2. 3. 4. 5. 6. 7. Overview: Get an overview of the collection Zoom: Zoom in on items of interest Filter: Remove uninteresting items Details on demand: Select items and get details Relate: View relationships between items History: Keep a history of actions for undo, replay, refinement Extract: Make subcollections 12/7/2020 15

Visualization principles 4 To support tasks 1 & 2, a general design pattern called “focus+context” is often used. 4 Idea is to have a focal area at high resolution, but keep all of the collection at low resolution. 4 Mimics the human retina. 12/7/2020 16

Distortion 4 Several visualization systems use distortion to allow a focus+context view. 4 “Fisheye lenses” are an example of strongly enlarging the focus while keeping a lot of context (sometimes the entire dataset). 4 Many of these were developed at Xerox PARC. 12/7/2020 17

Focus+Context: Document lens 12/7/2020 18

Focus+Context: Webbook lens 12/7/2020 19

Focus+Context: Table lens 12/7/2020 20

Focus+Context: Datelens is a PDA calendar program developed at U. Maryland (Bedersen et al. ) 12/7/2020 21

Navigation: Hyperbolic trees 12/7/2020 22

Navigation: Hyperbolic trees 12/7/2020 23

Navigation: Hyperbolic trees 12/7/2020 24

Navigation: Hyperbolic trees 12/7/2020 25

Navigation: Hyperbolic trees 12/7/2020 26

Navigation: Animation 12/7/2020 27

Network visualization 4 Often use mass-spring dynamic models. 4 Can be animated and interacted with. 12/7/2020 28

Discussion 4 Animation is used in viz schemes to smooth transitions, or increase “aliveness” of the display. 4 Discuss advantages/disadvantages of animation. 12/7/2020 29

Using 3 D 4 People perceive a 3 D world from 2 D views, so it seems like we could use 3 D structure to advantage. 4 Several systems (also Xerox PARC) have tried this. 4 Use 3 D spatial memory and organization to speed up navigation. 12/7/2020 30

Web. Book 12/7/2020 31

Web Forager 12/7/2020 32

Representing Hierarchies 12/7/2020 33

Summary 4 High-dimensional data can be visualized by 2 D via Multi. Dimensional Scaling. 4 Focus+Context is a design pattern for showing the area of interest and its relationship to the entire dataset. 4 3 D techniques can leverage the spatial capabilities of the human visual system. 12/7/2020 34