Relation Extraction from Online PH News Source
summary
Pipeline extracts person‑to‑person relations from Inquirer articles by scraping ~900 pages, tokenizing with NLTK, tagging with wikifier.org API, encoding pairs with Universal Sentence Encoder, and classifying with scikit‑learn where a support vector machine is champion at F1 ≈ 60.73 after training on Wiki80 (≈56k samples, 80 classes, 90/10 split). Extraction of 246 candidate sentences takes about six hours, manual Wikipedia validation finds 99 valid relations yielding ~67.35% precision.
AI-written summary · about
chapters (12)
- 0:00Intro
- 1:46INTRODUCTION
- 9:29FRAMEWORK OF APPROACH
- 19:15Universal Text Encoder
- 20:59PROJECT OUTCOMES
- 21:31METHODOLOGY
- 27:55DATA COLLECTION
- 34:06MODELING DATA PREP
- 34:53MODEL TRAINING
- 39:38RESULTS
- 43:57ANALYSIS
- 48:21CONCLUSION & FUTURE STUDIES
related talks
01h 30mRelation extraction from texts
32mWhat is X up to? - NER and Relationship Extraction for Information Extraction
35mUnderstanding the News around the World with Web Scraping and NLP at Scale
23mExtracting Entities and Relations in Portuguese Documents
21mRelationships Matter: Mining relationships using Deep Learning
19mAddressing Treatment-Relevance in Biomedical Relation Extraction
26mDeep Learning Architectures for Semantic Relation Detection Tasks
39mSemi-Supervised Bootstrapping of Relationship Extractors