Harvesting Ambient Geospatial Information from Social Media Feeds

         

 http://www.dailymail.co.uk/sciencetech/article-2014748/Twitter-Flickr-use-Manhattan-How-social-networking-breaks-district.html

      Stefandis, Anthony.  Crooks, Andrew. Radzikowski, Jacek. “Harvesting ambient geospatial data from social media feeds” (2011) 

https://lms.southwestern.edu/file.php/3722/Literature/Stefanidis-2011-Ambient_Geospatial.pdf

      This article presents an interesting and emerging type of geographical data dubbed “ambient geospatial information.” AGI, as it is referred to in this text, is a new type of information that,   similar to VGI, has enormous potential for mapping areas based on seemingly non-geographical data. This paper’s main purpose is to show the origins of, design a framework for harvesting, and present case studies concerning incorporation of AGI within geospatial analyses.

                

      AGI and VGI’s origins share similar prerequisites and contributing factors. The most prominent being the advent of Web 2.0. According to this paper, Web 2.0 can be defined by six overlapping concepts: (1) Individual/user generated content (2) power of the crowd (3) data on a massive scale (4) participation-enabling architecture (5) networking (6) transparency. Some major Web 2.0 sites are Facebook, Wikipedia, Youtube, Flickr, and MySpace. The application of these 6 concepts to GIS is evident in certain examples of VGI like Wikimapia and Google Maps. However there are other forms of less obvious geospatial data (text describing a location, geotagged picture in a photo album etc. ) and these types of ambient information sources can be harvested for their insight into a wide range of social, political, and cultural subjects. 

 

   The real challenge with AGI is designing a system that allows you to collect and analyze the data. Above shows the graphic representation of a generalized architecture for harvesting AGI provided by the authors. This process entails 3 steps: extracting data from social media providers, integrating, parsing, and storing these data, and then analyzing these data. The extraction of data can be achieved by using various queries to filter out certain data. An integral point in this extraction is the discrimination done by the Social Media Ingestor, or SMI, which organizes data from diverse sources into common categories like time of submission, location, and username. 

    When storing data, the information must be made into an entry within the database. This data entry is essentially a record of a single piece of corresponding social media information.  The paper makes this definition of entries primarily to emphasize the importance of cross-entry relationships between data from different sources. These cross-entry connections allow for analysis into the social networking behind certain types of social media information. Harnessing this information that is not implicitly geospatial in nature can allow for innovative, novel, and out-of-the-box analyses.

The case studies mentioned in this article include the geospatial analyses of hotspots according to AGI (Arab Spring mentioned) and the mapping of social networks through the data provided by social media feeds (Earthquakes of 2011 in Japan mentioned). By geotagging certain things like twitter streams, certain geographic hotspots can begin to emerge as landmark footprints. Social network analysis (SNA) is another possible way to use AGI in order to explore how different parts of a social system are linked.  Other interesting examples of AGI mentioned in the article include exploring disease outbreaks via data mining of search engine results and blogs; using twitter messages to forecast influenza rates; finding social network clustering and using those clusters for emergency purposes. All of these represent the unprecedented amount of real-time data that can be geolocated and analyzed. 

The concluding paragraphs of this paper discuss the future of AGI as well as the differences between VGI and AGI. AGI relies on passive, indirect information that can be transmuted into geospatial data, while VGI provides data that is direct and geospatial in nature.  In addition, VGI is voluntary by definition and does not necessarily infringe upon any privacy rights. When collecting AGI on the other hand, the information is sometimes taken from personal sources and can therefore be viewed as an invasion of privacy. Various examples of sites and apps that already use this passively sensed AGI are listed in order to make this point. In conclusion, the author’s insist that the real power of AGI is its ability to better understand groups-- not individuals. The recent growth of social media presents a unique opportunity for GIS to expand beyond the conventional sources of data and utilize AGI for greater mapping.