Explore advanced techniques in scientific text mining and knowledge graph construction in this conference talk from KDD 2020. Delve into topics such as Science IE, conditional statements, sequence labeling for information extraction, and multi-output sequence labeling. Learn about evaluation methods and examine a detailed case study. Understand the motivation behind table extraction systems, explore system pipelines, and discover various table components and templates. Gain insights into problem definition, ensemble learning, and learning-based classifiers. Review the Tablepedia system and analyze results from recommender systems and entity resolution tasks.
Overview
Syllabus
Two Works
Science IE: Conditional Statements
Sequence Labeling for IE
Multi-Output Sequence Labeling
Sequence Labels
Multi-Input Multi-Output Sequence Label...
Evaluation
Case Study (cont'd)
Motivation (cont'd)
System Pipeline
Table Components
Table Templates (cont'd)
Problem Definition
Ensemble Learning
Assumption 1
Learning-based Classifier
Review: Tablepedia System
Results (RecSys)
Results: Asking ERD
Taught by
Association for Computing Machinery (ACM)
