Skip to Main Content
 

Global Search Box

 
 
 
 

Files

ETD Abstract Container

Abstract Header

Modeling Crime Using Activities and Sentiment Generated from Geotagged Tweets

Abstract Details

2021, PhD, University of Cincinnati, Arts and Sciences: Geography.
In this dissertation, I show how activities and sentiment derived from tweets are valuable factors in analyzing urban crime patterns. Big social media data like geotagged tweets alleviate the census data’s weakness of showing the mobility in crime analysis. This is because geotagged tweets can measure the ambient population’s activities when modeling crime patterns. As crime is related to people’s routine activities, geotagged tweets shall help model crime better than census population for some types of crime. The first research in this dissertation assesses both the direct and spillover effects of tweets-derived ambient population on different types of crime and the intraday differences in day and night during weekdays and weekends. Besides, the mixture of spatial and contextual information of geotagged tweets makes it possible to measure each neighborhoods’ people’s collective sentiments. The contextual information taken from the negative or positive words and emojis in tweets may suggest forms of social cohesion. Social cohesion influences crime, crime impacts residents’ fear of crime, and fear of crime impacts residents’ mobility, guardianship, and victimization. Thus, the local dynamic collective sentiments may influence crime patterns. Inspired by Wilson and Kelling’s Broken Windows Theory, I propose Broken Sentiment Conjecture in the second research in my dissertation to explain the relationship between the neighborhoods’ collective sentiment and crime. Further, people may have varying sentiments at different places they visit. They may be more relaxed at home while more excited at other places like bars or stadiums. With this in mind, I ask: does Broken Sentiment Conjecture vary by types of major activity nodes? Are sentiments in homes correlate with crime more than sentiments in other places? I explore these questions in the third research in my dissertation. I use tweets’ spatial information to delineate people’s major activity nodes. After these, I assess sentiment at which type of activity node contributes more to Broken Sentiment Conjecture. In this dissertation, I test the tweets-crime relationship using negative binomial regression models. I collect four types of crime (assault, burglary, robbery, and theft) and all accessible public geotagged tweets in Cincinnati, OH in 2013. The results suggest 1) Tweets-derived ambient population and its spillover effect influence the magnitude of crime (including assault, burglary, robbery, and theft). However, the correlations vary by types of crime at different time periods of the day and week. 2) Broken Sentiment Conjecture is legit for certain crime types: a neighborhood’s collective negative sentiment significantly correlates with assaults, burglaries, and robberies. 3) Broken Sentiment Conjecture varies by types of activity nodes: people’s negative sentiment at home significantly correlates with burglary. Their negative sentiment at other major activity nodes significantly correlates with assault, burglary, and robbery. This dissertation suggests geotagged tweets count and its spatial lag are legit measures of the ambient population for crime analysis. This dissertation also supports Broken Sentiment Conjecture and implies sentiments may relate to people’s activities at different types of activity nodes. These findings bring new insights into existing criminology theories and have practical implications.
Lin Liu, Ph.D. (Committee Chair)
Diego Cuadros, Ph.D. (Committee Member)
John Eck, Ph.D. (Committee Member)
Kevin Raleigh, Ph.D. (Committee Member)
Tomasz Stepinski, Ph.D. (Committee Member)
136 p.

Recommended Citations

Citations

  • Lan, M. (2021). Modeling Crime Using Activities and Sentiment Generated from Geotagged Tweets [Doctoral dissertation, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1623167858208017

    APA Style (7th edition)

  • Lan, Minxuan. Modeling Crime Using Activities and Sentiment Generated from Geotagged Tweets. 2021. University of Cincinnati, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1623167858208017.

    MLA Style (8th edition)

  • Lan, Minxuan. "Modeling Crime Using Activities and Sentiment Generated from Geotagged Tweets." Doctoral dissertation, University of Cincinnati, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1623167858208017

    Chicago Manual of Style (17th edition)