KDD 2005
0 likes982 views
The document discusses the application of kernels in link analysis, particularly focusing on PageRank, HITS, and co-citation coupling used in bibliographic contexts. It aims to interpret diffusion kernels within link analysis and demonstrates how neumann kernels encompass both co-citation coupling and HITS importance, offering a spectrum of relatedness measures. Additionally, regularized Laplacian and diffusion kernels provide a new measure of relatedness that addresses limitations of traditional approaches.
![Kernels defining inner products of nodes in a graph. Diffusion kernels [Chung, 1997 ； Kondor & Lafferty, 2002] Neumann kernels [Kandola et al., 2003] Regularized Laplacian kernels [Smola & Kondor, 2003] … Background: Kernels for graphs](https://image.slidesharecdn.com/kdd-2005-110927012736-phpapp01/85/KDD-2005-3-320.jpg)
KDD 2005
- 1. Application of Kernels to Link Analysis Takahiko Ito † Masashi Shimbo † Taku Kudo ‡ Yuji Matsumoto † † Nara Institute of Science and Technology ‡ Google
- 2. Background: Link analysis PageRank and HITS evaluate the importance of web pages. Co-citation coupling is used by CiteSeer to estimate relatedness between papers.
- 3. Kernels defining inner products of nodes in a graph. Diffusion kernels [Chung, 1997 ； Kondor & Lafferty, 2002] Neumann kernels [Kandola et al., 2003] Regularized Laplacian kernels [Smola & Kondor, 2003] … Background: Kernels for graphs
- 4. Objective To give an interpretation of diffusion kernels in terms of link analysis.
- 5. Results Neumann kernels subsume both co-citation coupling relatedness and HITS importance. They also define a spectrum of intermediate measures between the two. Co-citation coupling HITS Neumann kernels
- 6. Results Regularized Laplacian kernels / diffusion kernels define a new relatedness measure that overcomes some limitations of traditional relatedness (co-citation and bibliographic coupling)
Editor's Notes
- #3: First I will talk about the background of this work. Link analysis is an essential tool for exploring networked data. For example, PageRank and HITS evaluate the importance of web pages. Co-citation coupling used by CiteSeer to estimate relatedness between papers.
- #4: On the other hand, several kernels for graphs have been proposed recently. In particular, a family of "diffusion" kernels defines an inner product of nodes in a graph, This family includes Heat kernels, Neumann kernels and regularized Laplacian kernels. Here, we have one question. What do these “inner products” represents when viewed as link analysis measures.
- #5: In this work, We give an interpretation of some diffusion kernels in terms of link analysis. More specifically, we show the interpretation of Neumann kernels and regularized Laplaian kernels. First We show Neumann kernels provide a unified perspective of relatedness and importance. Then We show regularized Laplacian kernels define a new relatedness measure that overcomes some limitations of traditional relatedness. This topic is discussed later.
- #6: In this work, We give an interpretation of some diffusion kernels in terms of link analysis. More specifically, we show the interpretation of Neumann kernels and regularized Laplaian kernels. First We show Neumann kernels provide a unified perspective of relatedness and importance. Then We show regularized Laplacian kernels define a new relatedness measure that overcomes some limitations of traditional relatedness. This topic is discussed later.
- #7: In this work, We give an interpretation of some diffusion kernels in terms of link analysis. More specifically, we show the interpretation of Neumann kernels and regularized Laplaian kernels. First We show Neumann kernels provide a unified perspective of relatedness and importance. Then We show regularized Laplacian kernels define a new relatedness measure that overcomes some limitations of traditional relatedness. This topic is discussed later.