Towards development of FOPL based tweet summarization technique in a post disaster scenario: From survey to solution

In post disaster situation, the existing network infrastructure might be partly or fully damaged. In that case, a very popular online social network like twitter can be an effective tool, where people can share their views and knowledge about what is actually happening in the affected areas. It is a very challenging task to analyze the situation during the golden hours of any large scale disaster due to the absence of any renowned news media. All the tweets posted related to disaster are not genuine. Hence, some filtration must be performed to discard rumor tweets. After eliminating rumors, it has been observed that the volume of genuine tweets obtained is also very large. Thus, it is non-trivial for relief or rescue teams to analyze that large number of tweets and to take any decision regarding the relief and rescue as manual processing of those large numbers of tweets take significant amount of time. It is necessary to devise a summarization technique for efficient processing and analysis of genuine information at any point of time. In this work, an FOPL based summarization technique has been adopted to summarize the genuine tweets. From results it has been analyzed that the proposed technique has achieved better ROUGH-1 variant score compared to some other existing popular baseline techniques. The generated summary achieves an average precision, recall and F-measure score of 0.79, 0.39 and 0.55 respectively.