Graph-based RelationalData Visualization

pdf at: www.icmc.usp.br/pessoas/junio

Graph-based RelationalGraph-based RelationalData VisualizationData Visualization

Daniel Mário Daniel Mário de Limade Lima

José Fernando José Fernando Rodrigues Jr.Rodrigues Jr.

Agma Juci Agma Juci Machado TrainaMachado Traina

<danielm@icmc.<[email protected]>usp.br>

<[email protected]><[email protected]> <[email protected]><[email protected]>

Instituto de Ciências Matemáticas e de ComputaçãoUniversidade de São Paulo

17th International ConferenceInformation Visualization

15, 16, 17 and 18 July 2013SOAS, University of London ● London ● UK

pdf at http://www.icmc.usp.br/~junio/PublishedPapers/Lima-et_al_IV-2013.pdf

http://www.icmc.usp.br/pessoas/junio

http://www.icmc.usp.br/~junio/PublishedPapers/Lima-et_al_IV-2013.pdf


OutlineOutline1. Introduction

2. Method

3. Experiments

4. Conclusions



1. Introduction1. Introduction



IntroductionIntroduction• Large datasets are common

• unstructured: text• semi-structured: XML, RDF, sensor data• structured: relational (DBMS), network (graph-like)

• Analysis Process• Data Representation / Transformation• Storage / Retrieval• Statistics• Visualization• Analysis

Iter

ate



IntroductionIntroduction• How to spot interesting facts in the relationships

of large relational databases?• How are the entities on the database related to

each other?• How are the entities distributed over the

relations of the database?• How do the several attributes of the database

influence the relationships of the entities?• How do we quickly and intuitively browse the

relational database, considering its complex structure?



Our approachOur approach• Use graph representation• Graph-partitioning techniques• Graph-processing• Interactive Visualization

Database Graph Partitioning Visualization Analysis



2. Method2. Method



Relationships as GraphsRelationships as Graphs

Author Publish Work

DB m×mDB m×m GraphGraph GraphTreeGraphTreePartitioningPartitioning VisualizationVisualization AnalysisAnalysis




Author Publish Work

Alice ABob B

Charles C…

A 1B 2C 3A 2

…

1 Optic Fiber2 Networks3 Cryptography

…

11

22

33

AA

BB

CC





Author Publish Work

Alice ABob B

Charles C…

A 1B 2C 3A 2

…


…

11

22

33

AA

BB

CC





Author Publish Work

Alice ABob B

Charles C…

A 1B 2C 3A 2

…


…

11

22

33

AA

BB

CC




Graph PartitioningGraph PartitioningDB m×mDB m×m GraphGraph GraphTreeGraphTreePartitioningPartitioning VisualizationVisualization AnalysisAnalysis









Hierarchical PartitioningHierarchical PartitioningDB m×mDB m×m GraphGraph GraphTreeGraphTreePartitioningPartitioning VisualizationVisualization AnalysisAnalysis



Hierarchical PartitioningHierarchical Partitioning

cut 0

subgraph 1 subgraph 2





cut 0






cut 0






cut 0

DB m×mDB m×m GraphGraph VisualizationVisualization AnalysisAnalysisGraphTreeGraphTreePartitioningPartitioning




cut 1 cut 0 cut 2





cut 1 cut 0 cut 2

subgraph 1-1

subgraph 1-2

subgraph 2-1

subgraph 2-2




SuperGraphSuperGraph

SuperNode 1-1

SuperNode 1-2

cut 1 cut 0 cut 2

subgraph 2-1

subgraph 2-2





SuperEdge 1

SuperNode 1-1

SuperNode 1-2

cut 0 cut 2

subgraph 2-1

subgraph 2-2





SuperEdge 2

SuperNode 2-1

SuperNode 2-2

cut 0





cut 0





SuperGraphSuperGraphSuperNode 2SuperNode 1

cut 0




SuperGraphSuperGraphSuperNode 2SuperNode 1

SuperEdge 0




SuperGraphSuperGraphDB m×mDB m×m GraphGraph GraphTreeGraphTreePartitioningPartitioning VisualizationVisualization AnalysisAnalysis



SuperGraphSuperGraph• Further details in the paper




Attribute-based PartitioningAttribute-based PartitioningPaper Author


Left relation: Paper = {idPaper, country, year, title}Rght relation: Author = {idAuthor, age, dept, authorName}









PaperPaper AuthorAuthor

PP AA







PP AA

local







USUS

USUS BRBR

PP AA

BRBR

local







USUS

USUS BRBR

PP AA

BRBR

localyear







USUS

USUS BRBR

PP AA

BRBR

’00-’06’00-’06

localyear

’06-’11’06-’11







USUS

USUS BRBR

PP AA

BRBR

’00-’06’00-’06

’06+’06+ **

‘95+‘95+ ’02+’02+

localyear

’06-’11’06-’11







USUS

USUS BRBR

PP AA

BRBR

’00-’06’00-’06

’06+’06+ **

‘95+‘95+ ’02+’02+

agelocalyear

’06-’11’06-’11







USUS

USUS BRBR <40<40 >40>40

PP AA

<40<40

>40>40

BRBR

’00-’06’00-’06

’06+’06+ **

‘95+‘95+ ’02+’02+

agelocalyear

’06-’11’06-’11







USUS

USUS BRBR <40<40 >40>40

PP AA

<40<40

>40>40

BRBR

’00-’06’00-’06

’06+’06+ **

‘95+‘95+ ’02+’02+

agelocalyear dept

’06-’11’06-’11







USUS

USUS BRBR <40<40 >40>40

PP AA

<40<40

>40>40

BRBR

’00-’06’00-’06

IMEIME **

’06+’06+ **

EESCEESC

ICMCICMC

‘95+‘95+ ’02+’02+

agelocalyear dept

’06-’11’06-’11

FFLCHFFLCH







USUS

USUS BRBR <40<40 >40>40

PP AA

<40<40

>40>40

BRBR

’00-’06’00-’06

IMEIME **

’06+’06+ **

‘95+‘95+ ’02+’02+

agelocalyear dept

’06-’11’06-’11

FFLCHFFLCH

Connectivity SuperEdges







USUS

USUS BRBR <40<40 >40>40

PP AA

<40<40

>40>40

BRBR

IMEIME **

’06+’06+ **

‘95+‘95+ ’02+’02+

agelocalyear dept

FFLCHFFLCH

Connectivity SuperEdges





R-Mine PrototypeR-Mine Prototype• Based on the GMine System• Test platform with minimalistic design

• SuperNode tree:• node-link, radial layout, partial focus

• SuperEdge graphs:• node-link, bipartite layout, edge filtering

• Leaf SuperNode graphs: typical node-link




R-Mine PrototypeR-Mine PrototypeDB m×mDB m×m GraphGraph GraphTreeGraphTreePartitioningPartitioning VisualizationVisualization AnalysisAnalysis



3. Experiments3. Experiments




Tycho USP databaseTycho USP database• Data from several USP systems

• Personnel, Supervisions, Publications, Events…



Tycho USP databaseTycho USP database• Using 5 entities and 5 relationships

• 350k events• 380k examinations• 691k publications• 50k people• 26k supervisions

• 1.5 million nodes total• 1.8 million edges (relationships)



Q1: active authorsQ1: active authors• Which group of People (by age) have the

largest number of recent publications?

SQL: SELECT a.age, count(*) num FROM PersonPublication x JOIN Publication p ON p.id = x.publication AND p.year >= 2008 JOIN Person a ON a.id = x.author GROUP BY a.age ORDER BY num DESC



Q1: active authorsQ1: active authors



Q1.b: active authorsQ1.b: active authors• Who are them?• SQL: SELECT a.name, p.title

FROM PersonPublication xJOIN Publication p ON p.id = x.publication AND p.year >= 2008JOIN Person a ON a.id = x.authorWHERE a.age IN (SELECT age FROM (SELECT a.age age, count(*) num FROM PersonPublication x JOIN Publication p ON p.id = x.publication AND p.year >= 2008 JOIN Person a ON a.id = x.author GROUP BY a.age ORDER BY num DESC) T)



Q1.b: active authorsQ1.b: active authors



Q2: favorite countriesQ2: favorite countries• Which country receives the largest number of recent

publications from this group of people?• SQL: SELECT a.name, p.title

FROM PersonPublication xJOIN Publication p ON p.id = x.publication AND p.year >= 2008JOIN Person a ON a.id = x.author AND a.age BETWEEN 56 AND 63WHERE p.country IN (SELECT country FROM (SELECT p.country country, count(*) num FROM PersonPublication x JOIN Publication p ON p.id = x.publication AND p.year >= 2008 JOIN Person a ON a.id = x.author AND a.age BETWEEN 56 AND 63 GROUP BY p.country ORDER BY num DESC) T)



Q2: favorite countriesQ2: favorite countries



Q2: favorite countriesQ2: favorite countries



Q3: active authors per Q3: active authors per countrycountry• Now in one specific country, which group of People is the

most active recently?• SQL: SELECT a.name, p.title

FROM PersonPublication xJOIN Publication p ON p.id = x.publication AND p.year >= 2008 AND p.country = ‘Estados Unidos’JOIN Person a ON a.id = x.authorWHERE a.age IN (SELECT age FROM (SELECT a.age age, count(*) num FROM PersonPublication x JOIN Publication p ON p.id = x.publication AND p.year >= 2008 AND p.country = ‘Estados Unidos’ JOIN Person a ON a.id = x.author GROUP BY a.age ORDER BY num DESC) T)



Q3: active authors per Q3: active authors per countrycountry



Q3: active authors per Q3: active authors per countrycountry



Performance: individual Performance: individual queriesqueries150 analytical questions: PostgreSQL × R-Mine



Performance: accumulated Performance: accumulated timetime150 analytical questions: PostgreSQL × R-Mine



Performance: loading timePerformance: loading time

SuperNode Load(s)Connectivity

to all siblings (seconds)

SQL (seconds)

(initial loading) 6.032 - -

Person 0.057 5.847 7.349

Event 0.271 5.276 26.716

Publication 0.160 4.484 27.677

Total 6.520 15.607 61.742



4. Conclusions4. Conclusions



Our approachOur approach• Can use the Relational information

• To guide the partitioning• To give an initial context to the analyst

• Faster than running SQL queries

• Make neighborhood exploration easy• Interactive Visualization environment



ConsiderationsConsiderations• Initial parameters

• Which entities, relationships and attributes?• In which order?• How to define partitions? Ranges?• How many partitions?

• Different interaction tasks• Ongoing usability evaluation



ThanksThanks


Graph-based RelationalData Visualization

Data & Analytics

network graph

workalice abob bcharles

workdb mmdb

relational dbms

sensor data

text semistructured

cincias matemticas e

interesting facts