automated conceptual abstraction of large diagrams by daniel levy and christina christodoulakis...

46
Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Upload: lesley-green

Post on 18-Jan-2016

219 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Automated Conceptual Abstraction of Large Diagrams

By Daniel Levy and Christina ChristodoulakisDecember 2012

(2 days before the end of the world)

Page 2: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Introduction Big picture Clustering Algorithm Experiment & Results Conclusion

Outline

Page 3: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Introduction Big picture Clustering Algorithm Experiments & Results Conclusion

Outline

Page 4: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

So what is this “clustering” you speak of? Why do we need to cluster? Reduce cognitive load

Introduction

Page 5: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)
Page 6: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)
Page 7: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)
Page 8: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Introduction

Big picture Clustering Algorithm Experiment + Results Conclusion

Outline

Page 9: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Big Picture

Page 10: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Vision

Page 11: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Diagram Abstraction

Page 12: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Its been done before..

Related Works

Page 13: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Consider a diagram stripped of semantics, or pre processed using methodologies in previous work

Cluster graph

Evaluate clusters proposed based on closeness of meaning in the node names

Our Approach

Page 14: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Our Approach

Page 15: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Introduction Big picture

Clustering Algorithm Experiment + Results Conclusion

Outline

Page 16: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Min-Cut

Page 17: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Naïve Min-Cut Algorithm

Page 18: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

C

A

N B

1

2

3C

A

N B2

3

E4

E4

*Must result in exactly 2 partitions

Combinations / Creating partitions

*Assume there exist additional nodes

Page 19: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

C

A

N B

1

2

3C

A

NB

1

E E4 4

Page 20: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

C

D

A

B

21

3C

D

A

B

2

Minimum sets

C

D

A

B

21

3 C

D

A

B

2

3

Page 21: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

D

AB

1

3

2D

AB3

2

D

AB

1

3

2D

AB

2

Cycles

Page 22: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

E

D

C

A

B

12

3

4

5

Listing the min-cuts

Page 23: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

E

D

C

A

B

12

3

4

5

Listing the min-cuts

Page 24: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

E

D

C

A

B

12

3

4

Listing the min-cuts

5

Page 25: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

E

D

C

A

B

12

3

4

5

Listing the min-cuts

Page 26: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

E

D

C

A

B

12

3

4

5

Listing the min-cuts

Page 27: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

E

D

C

A

B

12

3

4

5

E

D

C

A

B

12

3

Outside-in approach

Page 28: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

E

D

C

A

B

12

3

4

5

E

D

C

A

B

12

35

Outside-in approach

Page 29: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

E

D

C

A

B

12

3

4

5

E

D

C

A

B

12

3

4

Page 30: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

E

D

C

A

B

12

3

4

5

Page 31: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)
Page 32: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)
Page 33: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

We use RiTa WordNet getDistance() function We calculate pairwise distances between

nodes. Select for each node the smallest distance

between it and another node Sum all minimum distances Average over all nodes in candidate cluster

Cluster Distance Measure

Page 34: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Introduction Big picture Clustering Algorithm

Experiments + Results Conclusion

Outline

Page 35: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Experiment 1

Page 36: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Experiment #1

Page 37: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Experiment # 1User 1 abstraction

Page 38: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

ExperimentationUser 2 abstraction

Page 39: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Experiment # 1automated abstraction

Page 40: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Experiment 2

Page 41: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Experiment #2

Page 42: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Simplified version

Page 43: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)
Page 44: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Introduction Big picture Clustering Algorithm Experiments + Results

Conclusion

Outline

Page 45: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Surprised at how similar manual clustering and automated clustering were.

Suggested improvements: Automatic distance threshold Creating subgraphs Strictness of clustering (min # of clusters Advanced min-cut discovery

Conclusions

Page 46: Automated Conceptual Abstraction of Large Diagrams By Daniel Levy and Christina Christodoulakis December 2012 (2 days before the end of the world)

Questions?Merry Christmas!