ConStory: Automatic story investigator of public perception on the mega urban infrastructure project

A project focused on extracting the public opinions of North Houston Highway Improvement Project (IH-45) from Twitter

• Copyright (C) 2022 The University of Texas at Arlington
• Copyright (C) 2022 HBE: The Humanized Built Environment, (https://hubilab.uta.edu/)
• Copyright (C) 2022 Alireza Shamshiri, Kyeong Rok Ryu, Steven McCullough, and June Young Park

Citation of this project

• Alireza Shamshiri, Kyeong Rok Ryu, Steven McCullough, and June Young Park. 2022. ConStory: Automatic story investigator of public perception on the mega urban infrastructure project: poster abstract. In Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation (BuildSys '22). Association for Computing Machinery, New York, NY, USA, 293–294. https://doi.org/10.1145/3563357.3567751

Data collection & Preprocessing

• 4103 Scrapped Tweets from Twitter
• Text preprocessing has been done using Natural Language Toolkit (NLTK)

Topic Modeling

• Topic Modeling Performed Using Gibbs sampling algorithm for a Dirichlet Mixture Model (https://github.com/rwalk/gsdmm)

Temporal Analysis of Tweets

- Number of Tweets between 2008 and 2021, Bottom: Most used words in posted tweets for each phase

Topic Modeling Results

- Weights of extracted topics based on the appearing in the number of tweets posted in each year

• Topic Weights = Number of topics assigned to each tweet / Each year tweets contain that topic