Probing of geospatial stream data to report disorientation

Abstract

Probing of data streams in a distributed environment for observation is considered to be one of the prime activities of Big Data Handlers. The notion of big data is efficiently leveraged through popular social networking sites such as Facebook, Twitter, LinkedIn, etc. Twitter is a most popular micro-blogging website enriched with many research issues. The users are allowed to put their ideas and thoughts in the form of messages called “Tweets” in twitter. In this study, the purpose of gathering the location specific tweets is to understand and surface the insights which are related to human dynamics. We have employed the data stream mining approach to process geo-spatial time invariant tweets in a distributed real-time environment to gain more useful information. Topic models were explored for identifying a particular topic of interest or to extract prudent information from the stream data. Our concentration is on the evolution of different topics at different places, a location-topic matrix is formed for the set of topics observed as most predominant for the specific locations. Then a user graph is generated for the volatile topics that help in analyzing the users who have tweeted or has been re-tweeted on a specific topic the most. From the properties of the generated graph, the disorientation of the topics is reported in the given locations by the use of a sentimental analysis that deems the topic discussed as positive or negative. These analyzes have shown that there is a possibility to outwit the useless and most rampant negative issues spread mutely on a specific location which later creates unnecessary panic to the society.