Health and Crime Prediction on Social Media
edit
Data science is an interdisciplinary field on programs that extract knowledge from data derived from various sources. Because many people have access to social media e.g. Twitter and technological devices e.g. cell phones, data analysts have decided to use this data to improve individuals’ experiences; for instance, mobile data has been used in helping predict crime in certain geographic areas [1] Also Twitter, a microblogging site where web users post and read messages, or “tweets”, has been a prevalent source for studies in data science. In fact, Twitter data has been used to predict presidential elections [2], improve influenza forecasting[3], predict dark triad traits [4], improve understandings of health problems e.g. obesity [5], and improve online services to better suit health information seeking and sharing needs online [6]
Twitter and Influenza Forecasting
editInfluenza is a serious viral infection that kills hundreds of thousands of people each year, holding that accurate disease forecasts are critical. Unfortunately, forecasting is limited by the time necessary to acquire new, accurate data. In utilizing ILI data available at the time of forecast with Twitter data and Google Flu Trend (GFT) data, it is possible to generate a forecast model that can reduce forecast error by 17-30%. It can be stated that Twitter data improves influenza forecasting, and is a better predictor of influenza prevalence than GFT data [3].
Twitter and Crime Prediction
edit
Crime is another serious issue that can even reduce the economic growth of a geographic region; thus, finding a model to predict the crime levels in a particular area proves to be very useful. Since mobile phones are increasing in use worldwide, this is a useful source of data that can be used to generate a model to predict the crime levels in areas where crime has not been observed. For instance, one study obtained a dataset that included mobile data, criminal cases data and London boroughs files data; then, generated a model that could predict crime levels in the London Metropolitan Area with nearly 70% accuracy. This model demonstrated that human behavioral data computed by mobile activity improves the accuracy in predicting whether a city would be a crime hotspot or not [1]
Improving Online Health Information Seeking and Sharing
edit
The worldwide web has also become a common place for people to learn more about health information. From seeking medical advice from healthcare providers to sharing health information on social media, it is apparent that the use of both media for health information has become more popular. Since people are continuing to use the Internet to seek and share health information, it would be useful to further understand the "nature" of health information sought and why people choose a particular platform to search for health information. From one study, it has been has shown that search engines are more explicitly used to seek health information online [6]. Also, the primary intent of use for searching health conditions on a search engine is identifying treatment options and gaining an understanding of diagnostic processes for health conditions whereas Twitter is not. The primary motivating factors for seeking information on a particular platform is convenience, plurality of results, large audience, and the diversity of information available. Furthermore, the severity/type of condition and level of stigma does affect health information seeking behaviors. In assessing how people’s evaluation of their searching behaviors, people slightly agreed that there is privacy risks associated with search engines but that there are greater risks on Twitter; and people perceive search engines as providing higher-quality content and are more helpful in finding social support than Twitter. All in all, in observing health activities on both social media and search engines, a clearer picture of what health information people are searching for and why they choose a particular platform to search can be established; and understanding this could ultimately impact the development of these online services in the near future to better suit web user’s health activities on the Internet [6].
- ^ a b Bogomolov, Andrey, et al. "Once upon a crime: towards crime prediction from demographics and mobile data." Proceedings of the 16th international conference on multimodal interaction. ACM, 2014..
- ^ Andranik, et al. "Predicting Elections with Twitter: What 140 Characters Reveal about Political Sentiment." ICWSM 10 (2010): 178-185.
- ^ a b Paul, Michael J., Mark Dredze, and David Broniatowski. "Twitter improves influenza forecasting." PLOS Currents Outbreaks (2014).
- ^ Chris, et al. "Predicting dark triad personality traits from twitter usage and a linguistic analysis of tweets." Machine Learning and Applications (ICMLA), 2012 11th International Conference on. Vol. 2. IEEE, 2012.
- ^ Debarchana, and Rajarshi Guha. "What are we ‘tweeting’about obesity? Mapping tweets with topic modeling and Geographic Information System." Cartography and geographic information science 40.2 (2013): 90-102.
- ^ a b c De Choudhury, Munmun, Meredith Ringel Morris, and Ryen W. White. "Seeking and sharing health information online: comparing search engines and social media." Proceedings of the 32nd annual ACM conference on Human factors in computing systems. ACM, 2014..