Automatic detection and mapping live road closures -- 2
Budget: $30 – $250 USD
Road closures are very common due to severe weather conditions, accidents, and road
constructions. When disaster and unexpected situations happens navigation systems are slow to
share updates with the public. Fortunately, social media and new outlets provide live updates on
road closures. Through acquiring credible data about road closures, in this project, I will
implement supervised and unsupervised modeling techniques to map live closures. For the data
acquisition, I will extract various Transport departments tweets and information from social
media and news articles. There are different tools available to extract tweets and data mining
techniques to clean extracted data. Using Natural Language Processing and data analysis, I will
try to learn more about what makes a road closure announcement unique compared to all other
text. I will implement a supervised model on the cleaned tweets to run on unseen tweets to
determine if they are announcing a road closure or not. As well, an unsupervised model on news
headlines and Dept of Transportation announcements determines if they are announcing a live
road closure. Once the announcements, both tweets and news articles, were distinguished as
announcing road closures, will detect location from text using Natural language processing library
spaCy. Once we find the location name, will convert locations to coordinates and map them on
the HTML map
constructions. When disaster and unexpected situations happens navigation systems are slow to
share updates with the public. Fortunately, social media and new outlets provide live updates on
road closures. Through acquiring credible data about road closures, in this project, I will
implement supervised and unsupervised modeling techniques to map live closures. For the data
acquisition, I will extract various Transport departments tweets and information from social
media and news articles. There are different tools available to extract tweets and data mining
techniques to clean extracted data. Using Natural Language Processing and data analysis, I will
try to learn more about what makes a road closure announcement unique compared to all other
text. I will implement a supervised model on the cleaned tweets to run on unseen tweets to
determine if they are announcing a road closure or not. As well, an unsupervised model on news
headlines and Dept of Transportation announcements determines if they are announcing a live
road closure. Once the announcements, both tweets and news articles, were distinguished as
announcing road closures, will detect location from text using Natural language processing library
spaCy. Once we find the location name, will convert locations to coordinates and map them on
the HTML map