en
0.25
0.5
0.75
1.25
1.5
1.75
2
Inducing Cross-Lingual Semantic Representations of Words, Phrases, Sentences and Events
Published on Jan 11, 20136269 Views
Cross-lingual representations of linguistic units (e.g., words or phrases) can facilitate transfer of annotation from resource-rich to resource-poor languages and have many potential multilingual appl
Related categories
Chapter list
Inducing Cross-Lingual Semantic Representations of Words, Phrases and Events00:00
Why semantic representations?00:12
Why cross-lingual semantic representations?01:02
Outline01:39
Representing events and their participants01:56
Syntactic-Semantic Interface02:57
Our task (1)04:12
Our task (2)05:11
Induction of Frame-Semantic Information (1)05:41
Induction of Frame-Semantic Information (2)06:01
Induction of Frame-Semantic Information (3)06:12
Induction of Semantic Classes: Definition07:00
Induction of Semantic Roles: Definition (1)08:01
Induction of Semantic Roles: Definition (2)08:32
Argument Keys09:00
Outline 210:26
Modeling assumptions10:35
The (Simplified) Model (1)13:06
The (Simplified) Model (2)13:28
The (Simplified) Model (3)13:38
The (Simplified) Model (4)13:54
The (Simplified) Model (5)13:58
The (Simplified) Model (6)14:02
The (Simplified) Model (7)14:12
The (Simplified) Model (8)14:20
The (Simplified) Model (9)14:26
The (Simplified) Model (10)14:29
The (Simplified) Model (11)14:30
Encoding modeling assumptions14:41
Outline 316:02
Crosslingual Induction of Semantic Roles (1)16:20
Crosslingual Induction of Semantic Roles (2)17:12
Crosslingual Induction of Semantic Roles (3)17:34
Crosslingual Induction of Semantic Roles (4)17:49
Crosslingual Induction of Semantic Roles (5)18:09
Crosslingual Induction of Semantic Roles (7)18:13
Crosslingual Induction of Semantic Roles (8)18:14
Crosslingual Induction of Semantic Roles (9)18:19
Crosslingual Induction of Semantic Roles (10)18:28
Outline 418:49
Application-based Evaluation (1)18:53
Application-based Evaluation (2)20:06
Application-based Evaluation (3)20:14
Benchmark Dataset: PropBank (CoNLL 08)22:07
Crosslingual Semantic Role Induction23:04
Outline 523:51
Why not clustering as before?25:23
Why Crosslingual Representations?26:20
Summary of our Approach (1)26:27
Summary of our Approach (2)26:36
Summary of our Approach (3)26:37
Summary of our Approach (4)26:46
Background: Multitask Learning27:31
What do we take from MLT?28:26
How to encode relatedness?29:41
Qualitative Evaluation31:18
Crosslingual Document Classification31:56
On-going work: machine translation (1)33:00
On-going work: machine translation (2)34:19
On-going work: phrases34:24
Conclusions: cross-lingual representations34:47