en-de
en-es
en-fr
en-pt
en-sl
en
en-zh
0.25
0.5
0.75
1.25
1.5
1.75
2
Building blocks for semantic search engines: Ranking and compact indexing in entity-relation graphs
Published on Feb 25, 20079699 Views
We see an evolutionary path to supporting semantic search over text facilitated by 1. extractors and annotators for ever-growing collections of entity and relation types and 2. search systems that exp
Related categories
Chapter list
Building Blocks for Semantic Search Engines: Ranking and Compact Indexing for Entity-Relation Graphs00:00
Ranking and Indexing for Semantic Search00:26
Working Notion of Semantic Search00:37
Type-Annotated Corpus and Query e.g.02:38
The Query Class We Address05:25
Contribution 1: What is “NEAR”?07:25
Type-Annotated Corpus and Query e.g. (a)08:49
Contribution 1: What is “NEAR”? (a)09:12
Contribution 2: Indexing Annotations09:26
Part 1: Scoring and Ranking Nodes in Graphs10:39
Two Flavors of Ranking Problems10:46
Learning to Score Token Spans13:10
Learning the Shape of the Decay Function16:03
Learning to Score Token Spans (a)16:37
Learning the Shape of the Decay Function (a)17:07
Ranking Feature Vectors19:05
Benign Loss Functions for Scoring21:31
Learning Decay Function - Results23:45
Searching Personal Information Networks pt 127:16
Searching Personal Information Networks pt 229:11
Ranking Nodes in ER Graphs30:49
Edge Conductance32:39
Constrained Design of Conductance34:23
Breaking the p=Cp Recurrence37:25
Setting up the Optimization38:28
The Effect of a Limited Horizon39:16
Appropriateness of Loss Approximation40:05
Learning Rate and Robustness41:46
Discovering Hidden Edge Weights43:22
Part 1 Summary45:29
Part 2: Indexing for Proximity Search46:16
Part 2: Workload-Driven Indexing46:18
Pre-Generalize (and Post-Filter)47:53
(Pre-Generalize and) Post-Filter49:20
Estimates Needed by Optimizer49:45
Index Space Estimate Given R50:50
Processing Time Bloat for One Query51:35
Query Time Bloat - Results52:37
Expected Bloat Over Many Queries52:58
Smoothing Low-Probability Atypes53:51
The R Selection Algorithm54:34
Smoothing Low-Probability Atypes (a)56:20
The R Selection Algorithm (a)56:26
Optimized Space-Time Tradeoff56:34
Optimized Index Sizes57:26
Part 2 Summary58:03
The Big Picture58:24
Part 2 Summary (a)59:39
Conclusion01:00:10