clustering the tagged web
play

Clustering the Tagged Web Daniel Ramage, Paul Heymann, Christopher - PowerPoint PPT Presentation

Oct 02, 2022 •437 likes •1.4k views

Clustering the Tagged Web Daniel Ramage, Paul Heymann, Christopher D. Manning, Hector Garcia-Molina Stanford University WSDM 2009 Images from del.icio.us, lbaumann.com, www.hometrainingtools.com Web document text Web document text Words:

  1. Clustering the Tagged Web Daniel Ramage, Paul Heymann, Christopher D. Manning, Hector Garcia-Molina Stanford University WSDM 2009 Images from del.icio.us, lbaumann.com, www.hometrainingtools.com

  2. Web document text

  3. Web document text Words: information about catalog pricing changes in 2008 welcome looking hands-on science ideas try kitchen projects dissolve eggshell grow crystals

  4. Web document text Words: information about catalog pricing changes in 2008 welcome looking hands-on science ideas try kitchen projects dissolve eggshell grow crystals Anchor Text: home science tools hometrainingtools.com links click follow supplies training experiments other pages

  5. Web document text Words: information about catalog pricing changes in 2008 welcome looking hands-on science ideas try kitchen projects dissolve eggshell grow crystals Anchor Text: home science tools hometrainingtools.com links click follow supplies training experiments other pages Tags: science homeschool education shopping curriculum homeschooling experiments tools chemistry supplies

  6. Why tags? – del.icio.us

  7. Why tags? – del.icio.us ≈120,000 posts / day 12- 75 million (≈10 7 – 10 8 ) unique URLs (versus 10 9 – 10 11 total URLs) Disproportionately the web’s most useful URLs (and those URLs have many tags)

  8. Using tags to understand the web  The web is large and growing: anything that helps us understand high level structure is useful  Tags encode semantically meaningful labels  Tags cover much of the web’s best content  How can we use tags to provide high-level insight?

  9. Web page clustering task  Given a collection of web pages

  10. Web page clustering task  Given a collection of web pages A B  Assign each page to a A cluster, maximizing similarity within clusters A B C A C

  11. Web page clustering task  Given a collection of web pages A B  Assign each page to a A cluster, maximizing similarity within clusters A B  Applications: improved user interfaces, collection clustering, search result diversity, language-model C based retrieval A C

  12. Structure of this talk Features Words Tags Anchors Vector Space Model: K-means Generative Models Model: MM-LDA

  13. Models: K-means and MM-LDA Features Words Tags Anchors Vector Space Model: K-means Generative Models Model: MM-LDA

  14. Model 1: K-means clustering  K-means assumes the standard Vector Space Model: documents are Euclidean normalized real-valued vectors  Algorithm: iteratively Re-assign documents to closest cluster centroid Update cluster centroids from document assignments

  15. Model 2: Latent Dirichlet Allocation  LDA assumes each Words: Document 22 document’s words information about catalog generated by some pricing changes topic’s word distribution 2008 welcome looking hands-on science ideas try kitchen Topic 5 Topic 12 science catalog experiment shopping … learning buy ideas Internet practice checkout information cart

  16. Model 2: Latent Dirichlet Allocation  LDA assumes each Words: Document 22 document’s words information about catalog generated by some pricing changes topic’s word distribution 2008 welcome looking hands-on science ideas try  Paired with an inference kitchen mechanism (Gibbs sampling), learns per- Topic 5 Topic 12 document distributions science catalog over topics, per-topic experiment shopping … learning buy distributions over words ideas Internet practice checkout information cart

  17. Features: words, anchors, and tags Features Words Tags Anchors Vector Space Model: K-means Generative Models Model: MM-LDA

  18. Combining features Feature Combination Feature Space Size Words Words Anchors Anchors Tags Tags

  19. Combining features Feature Combination Feature Space Size Words Words Anchors Anchors Tags Tags Tags as Words Tags as Words Anchors as Words Tags & Anchors as Words

  20. Combining features Feature Combination Feature Space Size Words Words Anchors Anchors Tags Tags Tags as Words Tags as Words Tags as New Words Words Tags Words Anchors Words Tags Anchors

  21. Combining features Feature Combination Feature Space Size Words Words Anchors Anchors Tags Tags Simple feature space modifi- Tags as Words Tags as Words cations for existing models Tags as New Words Words Tags Words Anchors Words Tags Anchors

  22. Combining features Feature Combination Feature Space Size Words Words Anchors Anchors Tags Tags Tags as Words Tags as Words Tags as New Words Words Tags Words + Tags Words Tags Words Anchors Words Tags Anchors

  23. Combining features Feature Combination Feature Space Size Words Words Anchors Anchors Tags Tags Tags as Words Tags as Words Tags as New Words Words Tags Words + Tags K-means: normalize feature input Words Tags vectors independently Words Anchors LDA : multiple parallel sets of Words Tags Anchors observations via MM-LDA

  24. Experiments Features Words Tags Anchors 1. Combining Vector Space words and tags Model: in the VSM K-means Generative Models Model: MM-LDA

  25. Experiments Features Words Tags Anchors Vector Space Model: 2. Comparing K-means models, at Generative multiple levels of Models Model: specificity MM-LDA

  26. Experiments Features Words Tags Anchors Vector Space Model: K-means 3. Do words and tags complement or substitute Generative Models for anchor text? Model: MM-LDA

  27. Experimental Setup  Construct surrogate “gold standard” clustering using Open Directory Project  Reflects a (problematic) consensus clustering, with known number of clusters ODP Category # Documents Top Tags Computers 5361 web css tools software programming Health 434 parenting medicine healthcare medical Reference 1325 education reference time research dictionary

  28. Experimental Setup  Score predicted clusterings with ODP, but not trying to predict ODP  Useful for relative system performance ODP Category # Documents Top Tags Computers 5361 web css tools software programming Health 434 parenting medicine healthcare medical Reference 1325 education reference time research dictionary

  29. Evaluation: Cluster F1 A Reference B Intuition: balance A pairwise precision A (place only similar B documents together) with pairwise recall (keep all similar documents C Health together) A C

  30. Evaluation: Cluster F1 A Reference B Same Different Label Label A Same A B Cluster Different Cluster C Health A C

  31. Evaluation: Cluster F1 A Reference B Same Same Different Different Label Label Label Label A Same Same A 5 B Cluster Cluster Different Different Cluster Cluster C Health A C

  32. Evaluation: Cluster F1 A Reference B Same Same Different Different Label Label Label Label A Same Same A 5 5 3 B Cluster Cluster Different Different Cluster Cluster C Health A Cluster Precision: 5/8 C

  33. Evaluation: Cluster F1 A Reference B Same Same Different Different Label Label Label Label A Same Same A 5 5 3 3 B Cluster Cluster Different Different 8 Cluster Cluster C Health A Cluster Precision: 5/8 C Cluster Recall: 5/13

  34. Evaluation: Cluster F1 A Reference B Same Different Label Label A Same A 5 3 B Cluster Different 8 Cluster C Health A Cluster Precision: 5/8 C Cluster Recall: 5/13 Cluster F1: .476

  35. Experiments Features Words Tags Anchors 1. Combining Vector Space words and tags Model: in the VSM K-means Generative Models Model: MM-LDA

  36. Result: normalize words and tags independently in the Vector Space Model Features K-means Words .139 Words Tags .219 Tags Words+Tags .225 Words Tags Possible utility for other applications of the VSM

  37. Result: normalize words and tags independently in the Vector Space Model Features K-means Words .139 Words Tags .219 Tags Words+Tags .225 Words Tags Tags as Words (×1) Tags as Words .158 Tags as Words Tags as Words (×2) .176 Words Tags Tags as New Words .154 Possible utility for other applications of the VSM

  38. Experiments Features Words Tags Anchors Vector Space Model: 2. Comparing K-means models, at Generative multiple levels of Models Model: specificity MM-LDA

  39. Result: MM-LDA outperforms K-means on top-level ODP categories Features K-means (MM-)LDA Words .139 .260 Words Tags .219 .270 Tags Words+Tags .225 .307 Words Tags

  40. Tagging at multiple basic levels People use tags to help find the same page later, often at a “natural” level of specificity Programming/ Languages Society/ Social Sciences (1094 documents) (1590 documents) Java PHP Python C++ Issues, Religion & JavaScript Perl Lisp Spirituality, People, Ruby C Politics, History, Law, Philosophy

  41. Tagging at multiple basic levels People use tags to help find the same page later, often at a “natural” level of specificity Programming/ Languages Society/ Social Sciences (1094 documents) (1590 documents) Java PHP Python C++ Issues, Religion & JavaScript Perl Lisp Spirituality, People, Ruby C Politics, History, Law, java applies to 73% of Philosophy Programming/Java pages but software applies to only 21% of Top/Computer pages

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Lecture 23: Spectral clustering Hierarchical clustering What is a good clustering?

Lecture 23: Spectral clustering Hierarchical clustering What is a good clustering?

Lecture 23: Spectral clustering Hierarchical clustering What is a good clustering? Aykut Erdem May 2016 Hacettepe University Last time K-Means An iterative clustering algorithm - Initialize: Pick K random points as cluster

932 views • 64 slides
Clustering Hierarchical clustering, k-mean clustering Genome 559: Introduction to Statistical and

Clustering Hierarchical clustering, k-mean clustering Genome 559: Introduction to Statistical and

Clustering Hierarchical clustering, k-mean clustering Genome 559: Introduction to Statistical and Computational Genomics Elhanan Borenstein A quick review The clustering problem: partition genes into distinct sets with high homogeneity

772 views • 40 slides
Graph Clustering Graph Clustering What is clustering? What is clustering? Finding patterns

Graph Clustering Graph Clustering What is clustering? What is clustering? Finding patterns

Graph Clustering Graph Clustering What is clustering? What is clustering? Finding patterns in data, or grouping similar groups of data-points together into clusters . Clustering algorithms for numeric data: Lloyds K-means, EM

688 views • 19 slides
Clustering Hierarchical clustering and k-mean clustering Genome 373 Genomic Informatics

Clustering Hierarchical clustering and k-mean clustering Genome 373 Genomic Informatics

Clustering Hierarchical clustering and k-mean clustering Genome 373 Genomic Informatics Elhanan Borenstein A quick review The clustering problem: partition genes into distinct sets with high homogeneity and high separation Many

768 views • 43 slides
Clustering DSE 210 Clustering in R d Two common uses of clustering: Vector quantization Find

Clustering DSE 210 Clustering in R d Two common uses of clustering: Vector quantization Find

Clustering DSE 210 Clustering in R d Two common uses of clustering: Vector quantization Find a finite set of representatives that provides good coverage of a complex, possibly infinite, high-dimensional space. Finding meaningful

396 views • 36 slides
Clustering: Hierarchical Clustering and K- Means Clustering Machine

Clustering: Hierarchical Clustering and K- Means Clustering Machine

Clustering: Hierarchical Clustering and K- Means Clustering Machine Learning 10-601B Seyoung Kim Many of these slides are derived from William Cohen,

660 views • 46 slides
Clustering and Dimensionality Reduction Preview Clustering K -means clustering

Clustering and Dimensionality Reduction Preview Clustering K -means clustering

Clustering and Dimensionality Reduction Preview Clustering K -means clustering Mixture models Hierarchical clustering Dimensionality reduction Principal component analysis Multidimensional scaling Isomap

1.61k views • 17 slides
_ b _ q b 1 0 1 g b g _ g b _ 0 q b 1 1 b ~ ~ Search for Gluino b b

_ b _ q b 1 0 1 g b g _ g b _ 0 q b 1 1 b ~ ~ Search for Gluino b b

_ b _ q b 1 0 1 g b g _ g b _ 0 q b 1 1 b ~ ~ Search for Gluino b b Search for Gluino b b 1 1 7 7 20 20 Tagged Jets Tagged Jets -1 Tagged Jets Tagged Jets -1 CDF Run II Preliminary, 156pb CDF Run II

361 views • 13 slides
Clustering A Categorization of Major Clustering Methods Partitioning Methods

Clustering A Categorization of Major Clustering Methods Partitioning Methods

Clustering What is Clustering? Types of Data in Cluster Analysis Clustering A Categorization of Major Clustering Methods Partitioning Methods Hierarchical Methods 1 2 What is Clustering? What is Clustering? Typical

647 views • 18 slides
Subspace Clustering Ensemble Clustering Subspace Clustering, Ensemble Clustering, Alternative

Subspace Clustering Ensemble Clustering Subspace Clustering, Ensemble Clustering, Alternative

LUDWIG- MAXIMILIANS- INSTITUTE DATABASE UNIVERSITT FOR SYSTEMS MNCHEN MNCHEN INFORMATICS INFORMATICS GROUP GROUP Subspace Clustering Ensemble Clustering Subspace Clustering, Ensemble Clustering, Alternative Clustering, Multiview

513 views • 9 slides
Finding Clusters Types of Clustering Approaches: Linkage Based, e.g. Hierarchical Clustering

Finding Clusters Types of Clustering Approaches: Linkage Based, e.g. Hierarchical Clustering

Finding Clusters Types of Clustering Approaches: Linkage Based, e.g. Hierarchical Clustering Clustering by Partitioning, e.g. k-Means Density Based Clustering, e.g. DBScan Grid Based Clustering Compendium slides for Guide to Intelligent

1.32k views • 82 slides
Introduction to Machine Learning, Clustering and EM Barnab s P czos Contents Clustering

Introduction to Machine Learning, Clustering and EM Barnab s P czos Contents Clustering

Introduction to Machine Learning, Clustering and EM Barnab s P czos Contents Clustering K-means Mixture of Gaussians Expectation Maximization Variational Methods 2 Clustering 3 K- means clustering What is clustering?

1.32k views • 66 slides
Accessibility in the L A T EX kernel experiments in tagged PDF Chris Rowley Ulrike Fischer

Accessibility in the L A T EX kernel experiments in tagged PDF Chris Rowley Ulrike Fischer

Accessibility in the L A T EX kernel experiments in tagged PDF Chris Rowley Ulrike Fischer L A T EX3 Team August 2019, TUG Meeting Palo Alto, USA Accessibility in the L A T EX kernel experiments in tagged PDF Chris Rowley,

722 views • 15 slides
PAC-Bayesian Analysis of Co-clustering, Graph Clustering and Pairwise Clustering Yevgeny Seldin

PAC-Bayesian Analysis of Co-clustering, Graph Clustering and Pairwise Clustering Yevgeny Seldin

PAC-Bayesian Analysis of Co-clustering, Graph Clustering and Pairwise Clustering Yevgeny Seldin Motivation Example Clustering cannot be analyzed without specifying what it will be used for! be used for! Example Cluster then pack

293 views • 28 slides
Clustering kMeans, Expectation Maximization, Self-Organizing Maps Outline K-means

Clustering kMeans, Expectation Maximization, Self-Organizing Maps Outline K-means

Clustering kMeans, Expectation Maximization, Self-Organizing Maps Outline K-means clustering Hierarchical clustering Incremental clustering Probability-based clustering Self-Organising Maps Classification vs. Clustering

1k views • 67 slides
Chapter 7: Clustering (Unsupervised Data Organization) 7.1 Hierarchical Clustering 7.2 Flat

Chapter 7: Clustering (Unsupervised Data Organization) 7.1 Hierarchical Clustering 7.2 Flat

Chapter 7: Clustering (Unsupervised Data Organization) 7.1 Hierarchical Clustering 7.2 Flat Clustering 7.3 Embedding into Vector Space for Visualization 7.4 Applications Clustering: unsupervised grouping (partitioning) of objects into

763 views • 49 slides
Intercity and Regional Bus Network Study I-70 Transit Advisory Group (TAG) Meeting #1 February

Intercity and Regional Bus Network Study I-70 Transit Advisory Group (TAG) Meeting #1 February

Intercity and Regional Bus Network Study I-70 Transit Advisory Group (TAG) Meeting #1 February 6, 2013 What is the CDOT Intercity and Regional Network Study Building upon the 2008 Colorado Statewide Intercity and Regional Bus Study, this

260 views • 25 slides
(TAG) at River Trail MS Spring 20 2020 20 TAG students are placed in Advanced or TAG classes

(TAG) at River Trail MS Spring 20 2020 20 TAG students are placed in Advanced or TAG classes

Talented and Gifted (TAG) at River Trail MS Spring 20 2020 20 TAG students are placed in Advanced or TAG classes based on several criteria: Alternative Placements: Class (2 must be met) 5 th Grade Teacher Placements Recommendation

382 views • 8 slides
2016 Fall Parent Meeting For Parents of TAG Iden1fied Students and Those Who Are Interested in

2016 Fall Parent Meeting For Parents of TAG Iden1fied Students and Those Who Are Interested in

2016 Fall Parent Meeting For Parents of TAG Iden1fied Students and Those Who Are Interested in Finding Out About TAG TAG Services 1 Goals of Our Time Together Introduce School Leadership, Building TAG Facilitators and District Support

345 views • 12 slides
TAG Acquisition April 8, 2019 DISCLAIMER This publication may include forward-looking statements

TAG Acquisition April 8, 2019 DISCLAIMER This publication may include forward-looking statements

ENGIE Brasil Energia S.A. TAG Acquisition April 8, 2019 DISCLAIMER This publication may include forward-looking statements The information and opinions contained herein should not on events or results pursuant to Brazilian and international

479 views • 14 slides
Improved Annotation of the Blogosphere via Autotagging and Hierarchical Clustering Chris Brooks

Improved Annotation of the Blogosphere via Autotagging and Hierarchical Clustering Chris Brooks

15th International World Wide Web Conference Improved Annotation of the Blogosphere via Autotagging and Hierarchical Clustering Chris Brooks and Nancy Montanez Department of Computer Science University of San Francisco Department of Computer

212 views • 18 slides
Replay, Relay and Inverse-Sybil Attacks on Proximity Tracing Apps Krzysztof Pietrzak 2020

Replay, Relay and Inverse-Sybil Attacks on Proximity Tracing Apps Krzysztof Pietrzak 2020

Replay, Relay and Inverse-Sybil Attacks on Proximity Tracing Apps Krzysztof Pietrzak 2020 Eurocrypt Rump Session, May 13th, Living Room Proximity Tracing . . . 8GD45AF3 8GD45AF3 . . . 8GD45AF3 Proximity Tracing . . . 8GD45AF3 8GD45AF3

223 views • 19 slides
Parenting a bright child is a unique challenge. Passion for Learning, Avid Reader, Zany

Parenting a bright child is a unique challenge. Passion for Learning, Avid Reader, Zany

Parenting a bright child is a unique challenge. Passion for Learning, Avid Reader, Zany Sense of Humor, Curiosity, Endless Questions, Creativity . . . Share a parenting tip you have found successful in meeting your childs

227 views • 12 slides
QuizZoodle capture your audience! J. Allali QuizZoodle J. Allali (QuizZoodle) QuizZoodle 1 /

QuizZoodle capture your audience! J. Allali QuizZoodle J. Allali (QuizZoodle) QuizZoodle 1 /

QuizZoodle capture your audience! J. Allali QuizZoodle J. Allali (QuizZoodle) QuizZoodle 1 / 6 Example of a survey How are you today? A Excellent B Very Good C Good D Not so Good #QDLE#S#ABCD#20# J. Allali (QuizZoodle) QuizZoodle 2 / 6

379 views • 9 slides

More recommend