1. Event Focused URL Extraction from Tweets
By: Chris Bridges, Carter Tat, David Chun
CS 4624: Multimedia, Hypertext, and Information Access
Instructor: Edward A. Fox
Client: Liuqing Li
April 24, 2018
Virginia Tech, Blacksburg VA 24061
Slide Owner: Chris

2. Outline Project Goal Overall Design Testing /Evaluation Demo
References
Acknowledgements

3. Project Goal
Link existing Twitter collections and Event Focused Crawler (EFC)
Classify and rank relevance of URLs in Tweets to collection using deep learning and natural language processing techniques
Provide client with program that ties it all together

4. Overall Design

5. Testing/Evaluation 80% Training and 20% Testing Classifiers
Decision Tree
Random Decision Forest
Support Vector Classifier (SVC)
Gaussian NB
Cross-Validated using 10 subsamples

6. Results Classifier Decision Tree Random Forest Support Vector (SVC)
GaussianNB
Test Accuracy
Cross Validation Accuracy
0.94 (+/- 0.06)
0.95 (+/- 0.06)
0.75 (+/- 0.29)

7. Optimal Parameters

8. Demo
Slide Owner: David

9. Demo
“Future Florida Gators Softball Prodigy Is the Youngest NCAA Commit of All Time”

10. Demo
“Kentucky school shooting: 2 students killed, 18 injured”

11. References
“Sklearn.svm.SVC.” Sklearn.svm.SVC - Scikit-Learn Documentation, Web. Accessed 23. Apr 2018.
Moreira, Gabriel. “Discovering User's Topics of Interest in Recommender Systems.” LinkedIn SlideShare, 7 July 2016, Web. Accessed 23. Apr 2018
TextMiner. “Dive Into NLTK, Part IV: Stemming and Lemmatization.” Text Mining Online, 18 July 2014, Web. Accessed 23 Apr. 2018
"Events Archive (GETAR)." Events Archive. Web. Accessed 23 Apr
“Software Stanford Named Entity Recognizer (NER)." The Stanford Natural Language Processing Group. Web. Accessed 23 Apr
"Natural Language Toolkit." Natural Language Toolkit - NLTK Documentation. Web. Accessed 23 Apr
"Gensim: Topic Modelling for Humans." Radim Řehůřek: Machine Learning Consulting. Web Accessed 23 Apr

12. Acknowledgements Project Client: Liuqing, Li Instructor: Edward A. Fox
Global Event and Trend Archive Research (GETAR) is supported by NSF (IIS and )

13. Questions?