What is Geospatial Analysis?

Geospatial analysis is the act of capturing, manipulating, and visualizing information geographically, typically using a geographic information system (GIS), which is a software tool for mapping data. These systems can illustrate patterns and trends from previously unrelated information by applying the data to specific geographic coordinates on a map. The information can come from GPS sensors, mobile devices, surveys, satellite imagery, or even news reports and social media. Time can add another dimension; by adding timestamps, users can begin to map trends over both space and time in animated views.

While the market is filled with many providers, some of the more popular GIS vendors include ESRI (ArcGIS), Google (Google Earth, Google Maps), and BatchGeo. However, if you plan on enhancing your media analysis with GIS, you should not need a separate vendor. Any top social listening tool will include integrated GIS and features to help you apply it to your use case. Brand Intel, for example, includes ArcGIS or Google integration options as part of our standard deployment.

Top Geospatial Use Cases

There is a large variety of GIS-related use cases in the world today: from pure scientific applications (geology, hydrology, etc.) to navigation systems to supply chain management to urban planning. When it comes to location-based business and consumer services, most think about the outbound: ad-targeting through top platforms like Google Ads, Facebook Ads, Amazon Ads and the like. Users may configure content, demographics, budget, timeline, target locations, and then sit back while proprietary engines handle the rest: automatically bidding on placement and serving ads as populations move about the world.

What is Social Listening?

The explosive growth of mobile web and social media platforms has provided another layer of location-based services that leading public sector entities and private businesses are incorporating into their strategy. In the public sector, the combination of social listening and geospatial analysis for situational awareness has been quite beneficial to disaster management: enabling first responders to identify and respond people in need, while crisis management teams can disseminate key information through social channels. Additional public sector use cases include disease surveillance, traffic management, crime management, and tourism management.

Social listening for locations can have a huge impact on businesses too. Every day, consumers are sharing content: publishing experiences (good and bad), desires, observations, reactions, and fears. Many times, these experiences can be mapped to a location in the physical world: a favorite coffee shop, hiking trail, vacation spot, tips & tricks, restaurant dish, or trendy hotel. To truly understand your audience, it is important to understand where they are and what is important to them, so you can meet them with a relevant engagement. Geospatial-enhanced social listening allows you to further filter the noise and identify your audience of interest in specific areas.

Perhaps you have found that key prospects love a particular activity, hike, or selfie location. By incorporating this interest into a communication or physical event, you can ensure you are making impressions with the audiences that matter most to your brand. You may additionally identify potential cross-sell or partnership opportunities with brick-and-mortar locations. Of course, geospatial social listening can also provide these key insights to supercharge your outbound ad campaigns with the who, what, and where information needed to convert prospects to customers.

How do Social Listening & Social Analytics Tools Identify Location?

While the benefits of location-based social listening may be clear, there are practical challenges to its broad implementation. Most social media analytics and social listening tools will offer some support for locations, but not many do it well. This is because social location extraction is hard: there are data limitations and privacy restrictions that limit the amount of readily available geodata. Most platforms require users to specifically opt-in to location-sharing (geotagging) their posts. In Twitter alone, there are over 500 million tweets published per day, and less than 1% are geotagged with location data. In this challenging environment, how are social media analytics providers accomplishing this? Well, there are simple approaches practiced by most and advanced solutions provided by only a few key vendors.

Social Listening - Basic Location Classification

When it comes to determining a post, article, or mention location, there are a few basic approaches that most social media analytics platforms follow:

• Directly Geotagged data / Geo-Coordinates

  In this approach, data that has been geotagged by the user is able to be directly captured and visualized on a map. Typically, this data comes as part of the metadata accompanying the primary post content. The benefit here is that the information typically includes very precise locations within feet of the actual post/image location. Tags may also be captured at a business, neighborhood, or city level. The downside is that most posts will not include this information, due to the user’s own privacy settings. For some social analytics platforms, this is as far as they venture into location-based social listening. However, it is really just the beginning.

• Account location

  Without a geotagged post to rely on, social analytics platforms then typically move up to the user-level analysis. They may look at the location provided in the bio details of a post author’s account (eg Los Angeles, California) and then map posts accordingly. This information is only as good as the account author wants it to be – some may only include a state or country, and some may even use fictional names (Mordor, Winterfell). As you can imagine, accuracy here can be suspect.

• Miscellaneous Low-Resolution Options

  Without geotagged coordinates or user-location information, some vendors seek to identify a broad general location based upon the little remaining information. This can include using time zones (e.g. UTC-2 = 1 out of 20+ countries), extracting country from the top-level domain of a URL (e.g. caliente.mx = Mexico), or by looking up the location based upon the IP address of the web hosting provider. Each of these methods results in a very broad region that may not be very accurate.

Social Listening - Advanced Location Classification

For many, the basic social media location extraction approaches are insufficient - lacking in terms of coverage, resolution, and accuracy. Luckily, there are additional options. Advances in natural language processing (NLP), AI, computer vision and machine learning have made it possible to achieve more complete and more accurate location results within social and traditional media. These approaches go beyond the metadata, diving into the text, images, and other content to resolve the location.

“Geoparsing”, also called toponym extraction, involves processing free text and determining which parts reference a geographic location. As a follow-on process, “geocoding” converts place name into a set of geographical coordinates that can be placed on a map. In the example below, the tweet did not include geotagged metadata, nor did the author include a bio location (which probably would not have been relevant anyway). However, an accurate geoparser would have recognized the reference to “the Hollywood Bowl” as a place name and then geocoded the coordinates (34.112236, -118.339432) through a lookup.

The methods of geoparsing can be complex and may be proprietary amongst providers. They typically include a mix of:

1. Text processing with NLP methods (tokenization, Named Entity Recognition, parts of speech tagging, language modeling)
2. Building and managing a database of named locations
3. Place name matching and geocoding against the location database
4. Disambiguation or location refining

Additional, advanced methods include deep-learning based location classification, as well as using AI-models to infer location based upon interactions with other accounts.

Brand Intel - Your Location-based Social Listening Experts

While the more advanced methods of determining location can get complicated, rest assured that as a social listening platform user, you can access these insights without being an NLP expert or data scientist. If you plan on using geographic insights in your future, make sure you select a platform that can deliver beyond the basic. Only a select few social media analytics providers keep up with the latest advancements in location identification. If you would like any further information or a free demonstration, don't hesitate to reach out to the team today.