Research Report On The Boston Police Department

1311 Words6 Pages
7. Ongoing Research Work Introduction The second part of the research is focused on studying the emotions and sentiments expressed by people who retweeted the Boston Police Department (BPD) tweets. The goal of this phase was to observe the sentiment among the users who were retweeting the BPD tweets. We wanted to observe the impact of BPD tweets on the users before they retweeted the BPD tweet and after they retweeted. By doing this we wanted to capture the shift in sentiment among the users, which would give us the impact of BPD tweet. In order to do the above analysis, the tweets of those users who had retweeted the BPD tweet were collected. Topsy API was used to collect the historical tweets for the time duration (April 15th 2013 19:50…show more content…
The next step was to observe the sentiment among the users, before and after they had retweeted. We aggregated the ‘before’ set for all users corresponding to a particular tweet and generated a set of 145 files. Similarly we generated a set of 145 files for ‘After’ set as well. Then we used LIWC to generate the sentiment score for files corresponding to each BPD tweet. Thus for each of the 145 BPD Tweet, we generated sentiment score of ‘before’ and ‘after’, and so we have 290 rows of data as an output by LIWC. When we request Topsy for 200 retweeters of a particular BPD Tweet, we got a random sample of 200 retweeters. For some of these users, the retweets were outside our time interval of consideration. Therefore the number of users considered is not the same across all BPD tweets, it is always a number less than 200. Therefore we added 2 new columns for analysis, the number of users considered and the total number of tweets considered. Later we divided the BPD tweets into two segments 1. Incident related (The tweets by BPD which spoke about the incident and things related to it) 2. Informational (The tweets by BPD which were sent out to public about the road being blocked or about parking related etc.). The reason for dividing into 2 segments was to further observe the difference of impact on users by these tweets, with the help of sentiment scores generated
Open Document