en
0.25
0.5
0.75
1.25
1.5
1.75
2
Organizing Harvested Knowledge
Published on Jul 13, 20123582 Views
Related categories
Chapter list
Organizing Harvested Knowledge00:00
Our basic premise01:34
So it is natural to ask…02:15
Kinds of knowledge harvested03:01
Let’s talk about relations03:47
Approaches toward event roles04:54
The FrameNet solution07:26
Approaches 208:50
Taxonomy of top‐level case relations10:58
Entities: Noun-noun relations11:20
Tratz’s noun-‐noun relations12:38
Comparison to other studies12:54
Tratz: Mturk weighted Kappa agreements13:34
Summary for semantic relations13:45
Basically15:21
is‐a for taxonomizing concepts15:30
Aristotle, the father of it all15:47
The main relation: Subsumption17:08
The problem with Aristotle18:44
Tweety and the Nixon Diamond20:29
CYC has 2 subsumption relations21:56
shish kebabs22:45
Concept ‘facets’22:59
shish kebabs - example25:07
Why is this interesting?25:47
taxonomizing using dap26:20
Kozareva and Hovy26:36
DAP for taxonomy learning28:41
Experiment29:41
Results30:08
Evaluation woes: Precision30:57
Evaluation woes: Recall31:49
Evaluation measures31:58
Evaluation #1: Basic terms and Instances32:05
Evaluation #2: ISA links32:36
Is A above B?32:49
ISA relationship tests33:30
Evaluation #3: Intermediate concepts33:49
Effect of In-degree concept ranking34:02
Still…results are a bit of a mess34:06
How to taxonomize?34:28
Parallel is‐a hierarchies34:53
Evaluating concepts (1)35:01
Evaluating concepts (2)35:16
Evaluating concepts (3)35:18
Human category judgments35:19
More taxonomies… not so great35:27
Conclusion35:45
Today35:48
Thank you36:29