Lecture9 - Social Media and GIS
Introduction
- The lecture will focus on social media and GIS, addressing its continued relevance despite the changing social media landscape.
- A recent paper by Phil Barty on the impact of social media around fracking in Lancashire will be discussed.
- Definitions, characteristics, data types, geographic extraction (geotagging, geocoding, geoparsing), examples, challenges, and ethical considerations will be covered.
- Social media involves informal and formal communication between individuals and organizations.
- Platforms like Facebook, Twitter, and LinkedIn serve different purposes.
- Mobile devices facilitate constant connectivity and data sharing.
Cynical View
- Social media can create a filtered view of reality, reinforcing existing opinions.
Data and Usage
- Most studies focus on Twitter, Flickr, and Webio due to their simple formats.
- Global average social media usage is around 144 minutes a day, varying by age group.
- Billions of active users on platforms like Facebook (3 billion) and Twitter (556 million).
- Twitter is effective due to its concise message format and use of hashtags.
- Tweets were initially limited in size before expanding to 240 characters.
- Social media data is not volunteer geographic information (VGI).
- Researchers harvest ambient information for analysis.
- Tools and techniques extract geographic and textual components.
- The characteristics of social media data: volume, velocity, variety, and fine-grained nature
- Spatial information can be explicit (latitude, longitude, timestamp) or implicit (place names).
- Geotagging offers precise location data; geocoding assigns coordinates based on place names.
- The location of the user matters, differentiating between being present at an event versus commenting remotely.
Mapping Data
- Explicit data allows direct plotting using latitude and longitude.
- Twitter removed the geotagging facility in 2019 due to privacy concerns.
Geocoding
- Geocoding involves assigning coordinates to a location based on place names.
- Accuracy depends on the size of the place; smaller regions provide better accuracy.
- For local analysis, detailed address strings and postcodes are necessary.
- Forward geocoding converts addresses to coordinates; reverse geocoding finds addresses from coordinates.
Geoparsing
- Geoparsing extracts location information from text using machine learning.
- It identifies place names and contextual clues to infer locations.
- Challenges include misspellings, abbreviations, and colloquial terms.
- Algorithms analyze surrounding words to confirm spatial references.
- Tools like the Edinburgh Geoparcer can automatically identify and map place names in text.
Usage and Analysis
- Analysis involves studying the time and location of posts, content, and user interactions.
- Applications include tracking traffic accidents, natural disasters, and reactions to events.
- Studies have analyzed events like the Arab Spring, earthquakes in New Zealand, and disasters in Haiti.
- The analysis allows to track