0.25
0.5
0.75
1.25
1.5
1.75
2
EARL: Joint Entity and Relation Linking for Question Answering over Knowledge Graphs
Published on Nov 22, 20182784 Views
Many question answering systems over knowledge graphs rely on entity and relation linking components in order to connect the natural language input to the underlying knowledge graph. Traditionally, en
Related categories
Chapter list
Joint Entity and Relation Linking for Question Answering over Knowledge Graphs00:00
Outline - 100:12
Outline - 200:27
Question Answering over KG00:30
Question Answering over Knowledge Graph00:42
Question Understanding00:51
Key components in QA pipeline - 101:14
Key components in QA pipeline - 201:28
Anatomy of a Entity Linking System01:34
Anatomy - 101:34
Anatomy - 201:44
Anatomy - 301:50
But what about questions02:01
Relation Linking02:31
State of the art for Entity and Relation linking in QA 03:09
Preliminaries04:17
Postulates04:43
General Architecture - 105:30
General Architecture - 205:37
Preprocessing - 105:44
Preprocessing - 206:54
Preprocessing - 307:07
General Architecture07:50
Approach 1: GTSP - 108:10
Approach 1: GTSP - 208:42
Approach 1: GTSP - 308:54
"Where was the founder of Tesla and SpaceX born?" - 109:41
"Where was the founder of Tesla and SpaceX born?" - 210:06
Approximate GTSP Solvers10:17
Drawbacks of GTSP solution10:44
General Architecture11:34
Connection Density - 111:45
Connection Density - 211:50
Connection Density - 312:12
Connection Density - 412:22
Connection Density - 512:26
Connection Density - 612:46
Connection Density - 712:58
Connection Density - 813:02
Connection Density - 913:16
Adaptive learning14:04
GTSP vs Connection Density14:49
Postulates ( a look back)15:52
Evaluation16:04
Experiment 1: GTSP vs Connection density16:40
Experiment 2: Reranking17:15
Experiment 3 : Entity Linking17:53
Experiment 4 : Relation Linking18:13
Discussion18:21
Conclusion18:39
Questions Suggestions Discussions18:45