Distrbuted Agent system for analyzing Twitter content in case of finding car crashes and fire outbreaks
-
This project was made at Warsaw Univerity of Technology for the purpose of Agent Systems course.
-
Ths system finds car crashes and fire outbreaks related messages from Twitts
-
Classification whether particulat Tweet contains dangerous content (car crash / fire outbreak)
-
Implemeted Agent architecture for distribute system with JADE framework:
- Main agent container
- Defined custom agnt behaviour
-
Used Twitter4J in order to get http requessts from Twiter official API:
- needed obtaining API key
- some agents are crawlers and gather Tweets
-
Used openNLP for natual language processing part:
- some agents feed the model with pre-proessed Tweet content in order to train it
Contributors:
- breiker
- mjkrajsman
Folders:
- agent - whole agent system implemented with JADE
- openNLP - learning and testing NLP document categorizer model
- Twitter4J - handling Twitter API
Resources:
-
JADE: main page
-
openNLP: Sentiment analysis using OpenNLP document categorizer
-
openNLP document categorizer: Using Apache OpenNLP Document Categorizer
-
Twiter4J: A Java library for TwitterAPI