0.25
0.5
0.75
1.25
1.5
1.75
2
Machine Understanding for Interactive Storytelling: The MUSE project
Published on Dec 11, 20133079 Views
Related categories
Chapter list
Machine Understanding for Interactive Storytelling: The MUSE project00:00
MUSE project02:55
Outline03:36
Motivation for MUSE04:33
MUSE on one slide05:39
Overall goals of MUSE - 107:21
Use cases of MUSE08:18
Overall goals of MUSE - 208:36
Part 1: Machine understanding09:32
Goals of natural language processing in MUSE09:37
What can NLP already do?13:34
... far from what is needed in MUSE13:41
Challenges of the machine understanding13:48
Lack of training data14:23
Leveraging unlabeled data15:52
Language model16:49
Mapping to KR - 117:19
Mapping to KR - 217:31
Mapping to KR - 318:50
Mapping to KR - 421:11
Parsing discourse structure - 123:44
Parsing discourse structure - 224:25
Lack of domain/world knowledge - 125:55
Lack of domain/world knowledge - 229:57
A MSA of four event sequence decription30:29
Lack of domain/world knowledge - 331:07
Conclusions31:55
Special thanks to Oleksandr Kolomiyets, Quynh Do and Oswaldo Ludwig32:45
Interactive Narratives from Patient Education Documents32:52
The Problem33:29
Objectives34:21
Challenges35:03
New technology: Interactive Narrative35:52
Text-Based Generation37:21
Text-Based Generation of games …38:46
Interactive Patient Education “Games”39:35
Scenario: Bariatric Surgery - 140:22
Scenario: Bariatric Surgery - 241:14
Scenario: Bariatric Surgery - 343:03
Early NLP testing43:55
Feature Structures Extraction44:34
PDDL - AI Planning Representation44:53
Animation Techniques45:32
Example - 146:45
Patient examinations46:47
Dynamic change 1st 3rd person mode47:42
Direct/indirect Reference to actions49:08
The Difference49:55
Example - 250:53
Role of discourse structures51:34
How can deontic knowledge be represented in the system?52:14
Visualising non-compliance53:28
Representational issues53:32
Conclusions54:29