Iktishaf+: A Big Data Tool with Automatic Labeling for Road Traffic Social Sensing and Event Detection Using...

Area of Science:

• Computer Science
• Transportation Engineering
• Data Science

Background:

• Social media platforms serve as increasingly valuable data sources for real-world event detection.
• Traditional transportation monitoring methods face limitations in scope and cost-effectiveness.
• Big data analytics and machine learning applications in transportation are still evolving.

Purpose of the Study:

• To develop and implement an automatic labelling method for detecting traffic-related events from social media data.
• To create a software tool, Iktishaf+, for real-time traffic event detection using Arabic Twitter data.
• To leverage big data and distributed machine learning for enhanced transportation safety analysis.

Main Methods:

• Utilized distributed machine learning over Apache Spark for processing large volumes of social media data.
• Developed an automatic labelling method and a location extractor for Arabic tweets.
• Employed machine learning classifiers including support vector machines, Naïve Bayes, and logistic regression.
• Collected and analyzed 33.5 million Arabic tweets from Saudi Arabia via the Twitter API.

Main Results:

• Successfully detected and validated real-world traffic events such as fires, heavy rains, and accidents in Saudi Arabia.
• Demonstrated the effectiveness of the Iktishaf+ tool in automatically identifying events without prior knowledge.
• Extracted and visualized spatio-temporal information of detected traffic events.

Conclusions:

• Social media analytics, particularly Twitter data, is a powerful and cost-effective tool for real-time traffic event detection.
• The developed Iktishaf+ tool and automatic labelling method show significant promise for improving transportation safety and management.
• This research highlights the potential of big data and distributed machine learning in addressing critical transportation challenges.