A Novel Visualization Model for Web Search Results Nguyen T, and Zhang J. 2006. IEEE Transactions on Visualization and Computer Graphics PAWS Meeting Presented.

A Novel Visualization Model A Novel Visualization Model for Web Search Resultsfor Web Search Results

Nguyen T, and Zhang J. 2006. IEEE Transactions on Nguyen T, and Zhang J. 2006. IEEE Transactions on Visualization and Computer GraphicsVisualization and Computer Graphics

PAWS MeetingPAWS MeetingPresented by Jae-wook AhnPresented by Jae-wook Ahn

February 9, 2007February 9, 2007

MotivationsMotivations

Search result presentationSearch result presentation Linear ranked list vs. visualizationLinear ranked list vs. visualization

InteractivityInteractivity ExplorationExploration Discover and analyze information by usersDiscover and analyze information by users

Simple term based visualization modelsSimple term based visualization models Multiple meaningsMultiple meanings Order of keywordsOrder of keywords Missing context informationMissing context information

Motivations (cont’d)Motivations (cont’d)

What is missing with current approaches?What is missing with current approaches? Semantic views/semantic relationsSemantic views/semantic relations Degree of relevance in terms of Degree of relevance in terms of subjects of subjects of

interestinterest Visualization model adapting to users’ Visualization model adapting to users’ subjects subjects

of interestof interest and contextual information and contextual information

Proposed approachProposed approach

Metaphor – solar systemMetaphor – solar system Query – SunQuery – Sun Documents – planetsDocuments – planets Dimensions and attributesDimensions and attributes

Semantic strength (= relevance)Semantic strength (= relevance) Gravity = (distance)Gravity = (distance) Rotation speedRotation speed ColorColor

What is unique?What is unique? Use Use movement, movement, speedspeed, , and and distancedistance to visualize the to visualize the

degree of relevance among a query and Web search degree of relevance among a query and Web search results with respect to users’ results with respect to users’ subjects of interestsubjects of interest and and contextual informationcontextual information

ArchitectureArchitecture

WebSearchViz – Java based meta search (Google) visualizationWebSearchViz – Java based meta search (Google) visualization Vector space model – document-term matrixVector space model – document-term matrix Term weightingTerm weighting Similarity measureSimilarity measure

Subject of interestSubject of interest

Subject = List of Subject = List of keywordskeywords

Users can Users can manually manually add/remove/edit add/remove/edit subject of interestssubject of interests

Weights are Weights are adjustableadjustable

Visualization spaceVisualization space

Location of documentsLocation of documents Similarity to the querySimilarity to the query

More similar, closerMore similar, closer Similarity to the subjectsSimilarity to the subjects

More similar, closerMore similar, closer

Rotation speed of the Rotation speed of the documentsdocuments Similarity to the subjectsSimilarity to the subjects

More similar, more identical to More similar, more identical to the speed of the subjectsthe speed of the subjects

Computed by the angles (Computed by the angles (ii))

RotationRotation Automatic or manualAutomatic or manual

Query

Document

Subject of interest

WebSearchVizWebSearchViz

(1) Google’s search results

(2) Users’ subjects

(3) Visualization

(4) Rotation control

Additional featuresAdditional features

Manual rotationManual rotation Users can view the impact of a moving subjectUsers can view the impact of a moving subject

ColorsColors Users can mark a document and can keep track it through sessionsUsers can mark a document and can keep track it through sessions

Center switchingCenter switching Any page can be made the centerAny page can be made the center

FilteringFiltering Users select a threshold by creating a filter circleUsers select a threshold by creating a filter circle Filter out low similarity documentsFilter out low similarity documents

GroupingGrouping Visualize only a group of documentsVisualize only a group of documents

ExperimentsExperiments

Time efficiencyTime efficiency 0.7 sec (parsing & indexing)0.7 sec (parsing & indexing) 0.5 sec (document-term matrix, 10,000 page, 54,000 keywords)0.5 sec (document-term matrix, 10,000 page, 54,000 keywords) Initial Visualization rendering – 1.2 secInitial Visualization rendering – 1.2 sec

Usability studyUsability study 20 undergraduates (natural & social science)20 undergraduates (natural & social science) Questionnaire & accuracy evaluationQuestionnaire & accuracy evaluation 89% recall, 92% precision89% recall, 92% precision More successful with WebSearchVizMore successful with WebSearchViz Subjects liked more manual rotationSubjects liked more manual rotation Subjects liked grouping, filtering, focus page shiftingSubjects liked grouping, filtering, focus page shifting Short learning curve – enhancement (WebSearchViz) to an existing Short learning curve – enhancement (WebSearchViz) to an existing

search service (Google)search service (Google)

ConclusionsConclusions

Web search visualizationWeb search visualization Considers context – subjects of interestsConsiders context – subjects of interests Subject editorSubject editor New dimension – rotation speedNew dimension – rotation speed

Not just decorativeNot just decorative