Estimating the Locations of Emergency Events from Twitter

Question: Describe about the Estimating the Locations of Emergency Events from Twitter? Answer: Introduction: Position of a person can be identified from the activities on internet without using GPS services (Wu Zhu, 2015). As the number of social networking site users is increasing, use of information contained in these sites can be effective to understand the position of a person (Ao et al., 2014). In this report two articles are analyzed to understand the approaches which are used for identifying the location of a person. Both the articles contain discussions on the identification process of a persons location using information contained in tweets. However both the authors used different approaches for predicting location of a user of twitter. Research questions: Analysis of the two articles indicates that both researches are conducted for predicting location of a person. In both articles researchers analyzed the languages and contents of a tweet to identify the position. Use of GPS system can sometime provide wrong information regarding the actual position of a person as this system identify the location using geo co ordinates (Li et al., 2015). Using the information regarding a persons activities can provide more accurate information on the location. In both articles research questions are set to understand the approaches of predicting location of a person on basis of content and time of posting of a tweet. Mahmud et al. (2015) focused on predicting the location of a person on basis of content of the tweet and tweeting behavior. However the authors also conducted the research to understand the effectiveness of statistical and heuristic classifiers for predicting the location of a person from tweets. Li et al., (2015) conducted the research to predict the location from time of posting the tweet and language. However, the researchers also tried to finding whether the tweet user leaves any information on location or not. Rationale for the research: Use of GPS system may not be effective for differentiating the places which are closely placed. Information about position of a person also cannot be obtained due to failure of GPS system. However in such cases tweets are effective to understand proper location of a user. Before using tweets for identifying location of a person it is important to understand whether the predictions are accurate or not. In case of tweets the locations can be understood using POI tags. Use of POI tags is still limited among users due to privacy issues. As the comment and language of tweets vary according to the location of person, contents can be used as an indicator of POI. Although different methods are available for predicting location of person from tweets, effectiveness of these methods varies according to the algorithms. The current research will be effective to analyze the suitability of techniques proposed by other researchers. Results: Analysis of the results from both researches indicates that the location of a person can be predicted from tweets. In both researches different methods are used for predicting location. Use of city based classification method provides highly accurate information regarding the position. It is almost impossible for the users to hide the information about their location if city based classification method is used. Often the users of tweets do not mention the name of their location in the tweets. In such cases word and hash tag based classifiers can be used. The results indicate that locations can be predicted using hash tag and word classifiers also. However, use of heuristic and statistical classifiers improves the accuracy of prediction significantly (Mahmud et al., 2015). As POI (Places of Interest) tags do not contain only geographical co ordinates regarding, use of POI tags also provides accurate information regarding the location. Mahmud et al. (2015) opined that the hierarchical algorithm is better than other commonly used approaches for predicting the location of a person. Li et al., (2015) stated that the tweet users provide some information regarding their places unconsciously. My analysis on the results of both articles indicates that combination of which combines heuristic and statistical approaches improved the efficiency of hierarchical algorithm. However effectiveness of ranking approach can be affected if sufficient number of tweets is not available. Analysis on the ranking approach also indicates that effectiveness of this method also reduces when the contents of tweets reduce. Conclusion: Analysis on the researches indicates that the information regarding location is difficult to hide for a tweet user. The accuracy of prediction depends on use of frameworks. References: Ao, J., Zhang, P., Cao, Y. (2014). Estimating the Locations of Emergency Events from Twitter Streams. Procedia Computer Science, 31, 731-739. doi:10.1016/j.procs.2014.05.321 Li, W., Serdyukov, P., Vries, A., Eickhoff, C., Larson, M. (2015). The Where in the Tweet. Mahmud, J., Nichols, J., Drews, C. (2015). Where Is This Tweet From? Inferring Home Locations of Twitter Users. Sixth International AAAI Conference On Weblogs And Social Media. Wu, L., Zhu, Y. (2015). Inferring Locations of Mobile Devices from Wi-Fi Data. Intelligent Information Management, 07(02), 59-69. doi:10.4236/iim.2015.72006