摘要: |
This project analyzed the spread of influence in social media, in particular, the Twitter social media site, and identified the individuals who exert the most influence to those they interact with. There are published studies that use social media to assess public perception and sentiment regarding public transit. The project had two goals. First, to identify the online sources of opinions as expressed in social media networks and second, to identify a set of social network users who collectively have the largest spread of influence on the social network. The sequence of methods to accomplish these goals include, 1) extraction of transportation related social media messages; 2) extract spatial relations between social media users; 3) parameter identification of influence spread models; 4) Influence maximization for transportation topics; 5) GIS visualization; and 6) datasets. After development of the methods were completed, the analysis produced by our approach was evaluated for correctness. We randomly selected subsets of the test data for manual annotation. Specifically, investigators manually determined if these test data messages were relevant to one of the pre-determined transportation topics and if these messages had been so identified by the topic modeling step. The following are the deliverables for this project: a) Analysis of online social media communications to identify statements related to transportation; b) Discovery of originators of frequently transmitted messages on transportation, c) Set of most influential social media users for specific topics (the number of such users varied by the planner), and 4) Output of results in format that can be ingested by industry-standard GIS tools. |