The 1st ACTIVE Summer School 2009 - Bled

Link Mining

author:Lise Getoor, University of Maryland
published: Oct. 20, 2009, recorded: September 2009, views: 69

You might be experiencing some problems with Your Video player.

Slides

Slides
0:00	Link Mining - first
1:37	Roadmap
2:54	Link Mining
5:42	Linked Data
7:22	Sample Domains
13:58	Link Mining Tasks
14:58	Object Classification
17:02	Object Type Prediction
18:10	Link Type Classification
19:10	Predicting Link Existence
20:23	Link Cardinality Estimation I
21:42	Link Cardinality Estimation II
23:14	Entity Resolution
24:06	Group Detection
24:54	Subgraph Discovery
25:36	Graph Alignment
27:02	Link Mining Tasks
28:18	Link Mining Challenges
29:38	Logical vs. Statistical Dependence
30:17	Model Search
33:38	Feature Construction
34:14	Aggregation
36:54	Selection
37:30	Individuals vs. Classes
38:18	Instance-based Dependencies
39:10	Class-based Dependencies
40:02	Individuals vs. Classes
41:14	Collective classification
41:34	Collective Resolution
42:18	Labeled & Unlabeled Data
44:58	Link Prior Probability
45:21	Closed World vs. Open World
46:34	Link Mining Summary
50:30	Some Link Mining Algorithms
50:58	Collective Classification
51:18	Traditional Classification
53:39	Relational Classification (1)
56:57	Relational Classification (2)
57:29	Relational Classification (3)
58:42	The Problem
59:38	Example: Linked Bibliographic Data
61:22	Feature Construction
61:50	Simple Aggregation
62:14	Feature Construction
62:26	Aggregate Features Used
65:38	Formulation
67:42	CC Inference Algorithms
67:55	Formulation
68:30	CC Inference Algorithms
68:54	Local Classifiers Used in ICA
70:38	ICA: Learning
71:10	ICA: Inference (1)
72:30	ICA: Inference (2)
73:22	Experimental Evaluation
74:30	Results on Real Data
76:34	Effect of Structure
83:47	Results on Real Data
84:04	Effect of Structure
84:37	Entity Resolution
84:47	The Problem
84:50	InfoVis Co-Author Network Fragment
86:34	The Entity Resolution Problem
87:58	Attribute-based Entity Resolution
90:10	Roadmap: Relational Entity Resolution
90:20	Relational Entity Resolution
91:22	Relational Identification
92:06	Relational Disambiguation
92:46	Relational Constraints
93:26	Collective Entity Resolution
94:14	Entity Resolution with Relations
94:34	Algorithms
94:54	Example: CiteSeer
95:14	Example: CiteSeer (1)
95:42	Relational Clustering (RC-ER)
97:46	Relational Clustering (RC-ER) (1)
98:22	Relational Clustering (RC-ER) (2)
98:42	Relational Clustering (RC-ER) (3)
99:10	Cut-based Formulation of RC-ER
100:18	Objective Function
100:50	Measures for Attribute Similarity
102:01	Relational Similarity: Example 1
102:03	Relational Similarity: Example 2
102:06	Comparing Cluster Neighborhoods
103:30	Relational Clustering Algorithm
104:02	Probabilistic Model (LDA-ER)
104:14	Probabilistic Generative Model for Collective Entity Resolution
104:15	Discovering Groups from Relations
104:43	Latent Dirichlet Allocation ER
104:54	Generating References from Entities
104:56	Approx. Inference Using Gibbs Sampling
105:18	Faster Inference: Split-Merge Sampling
105:25	Experimental Evaluation
105:27	Evaluation Datasets
106:01	Baselines
106:34	ER over Entire Dataset
107:11	ER over Entire Dataset (1)
108:55	Performance for Specific Names
108:57	Trends in Synthetic Data
108:59	Link Prediction
109:01	Link Prediction: The Problem
109:03	Links in Data Graph
109:25	Links in Information Graph
109:45	Predicting Relations
110:15	Roadmap: Conclusion
110:20	Putting Everything together….
110:46	Learning and Inference Hard
111:23	Caveat: Link Mining & Privacy
112:08	Link Re-Identification
112:34	Attribute disclosure in OSNs
112:58	Conclusion
114:09	- Questions

Report a problem or upload files

If you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.

Lecture popularity: You need to login to cast your vote.

Description

Statistical machine learning is in the midst of a "relational revolution". After many decades of focusing on independent and identically-distributed (iid) examples, researchers are now studying problems in which the examples are linked together into complex networks. These networks can be a simple as sequences and 2-D meshes (such as those arising in part-of-speech tagging and remote sensing) or as complex as the collaboration structures produced by knowledge workers performing a variety of tasks in different contexts within an enterprise. In this talk, I will give an overview of this newly emerging research, focusing specifically on link mining tasks and algorithms.