Sign inSign up
Open Information Extraction from the Web thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Open Information Extraction from the Web

Published on Jul 13, 20126281 Views

Oren Etzioni

Traditionally, Information Extraction (IE) has focused on satisfying precise, narrow, pre-specified requests from small homogeneous corpora (e.g., extract the location and time of seminars from a

AKBC-WEKEX 2012 - Montreal

Related categories

Information ExtractionNatural Language Processing Semantic WebText Mining

Chapter list

Open Information Extraction from the Web00:00
KnowItAll Project (2003…)00:17
Outline00:45
I. Machine Reading (Etzioni, AAAI ‘06)01:45
I. Machine Reading (Etzioni, AAAI ‘06) - Ontology free04:43
Lessons from DB/KR Research04:57
Lessons from DB/KR Research - a fortiori06:18
Machine Reading at Web Scale06:40
II. Open vs. Traditional IE08:39
Semantic Tractability Hypothesis10:12
Sample Relation Phrases11:32
Number of Relations12:26
TextRunner (2007)13:23
Relation Extraction from Web13:57
Open IE (2012)14:19
Open IE (2012) - NBA15:27
Open IE (2012) - ZDA president15:45
Towards “Ontologized” Open IE17:00
System Architecture18:34
III. Critique of Open IE25:08
Perspectives on Open IE26:12
A. New Paradigm for Search26:44
Case Study over Yelp Reviews27:54
RevMiner: Extractive Interface to 400K Yelp Reviews (Huang, UIST ’12)28:54
Revminer.com29:48
B. Public Textual Resources (Leveraging Open IE) (1)30:09
B. Public Textual Resources (Leveraging Open IE) (2)30:47
C. Reasoning over Extractions32:52
Unsupervised, probabilistic model for identifying synonyms33:33
Scalable Textual Inference34:45
Inference Scalability for Holmes35:16
Extractions -> Domain/range35:29
xxxxx36:03
Generative Story36:17
Examples of Learned Domain/range36:22
Summary: Trajectory of Open IE36:40
IV. Future: Open Open IE37:31
Conclusions39:21