What is Geotagging from Unstructured Text?
Geotagging from unstructured text is an AI-based technology that not only identifies location information (e.g., place names) in natural language text, such as social media posts, emails, reports, or news articles, but also assigns it geo-coordinates.
Why is Geotagging from Unstructured Text important for GIS?
A Geographic Information System (GIS) is a computer system designed to display and analyze spatial or geographic data. It connects data to maps for geospatial intelligence and analysis. GIS enables users to identify patterns, relationships, and trends, enhancing decision-making in sectors like urban planning, emergency management, environmental science, and intelligence analysis.
While traditional GIS relies on coordinates that are often contained in structured data or document metadata, there is, however, a largely untapped wealth of locational information in unstructured text. The roadblock is that place names are not explicitly identified as such in unstructured text and also lack explicit geo-coordinates. In fact, traditionally the only way to incorporate unstructured data into GIS has been intensive manual curation and normalization of the unstructured data to make it usable by GIS tools.
Geotagging from unstructured text removes this roadblock by not only automatically finding place names in natural language text but also assigning geo-coordinates to them.
Where is Geotagging from Unstructured Text Used?
Geotagging from unstructured text is used across a wide range of sectors, such as:
- Law Enforcement & Public Safety
Police departments use geotagging to identify crime hotspots based on online discussions.
- Disaster Response & Humanitarian Aid
Organizations use geotagging to identify where crises are unfolding in real time. For example, during earthquakes or floods, responders analyze social media posts to locate affected areas and allocate resources quickly.
- Defense & Intelligence
Government agencies analyze unstructured text (e.g., internal reports, intercepted communications, online forums, social media, etc.) to detect threats or track activity in specific regions. This includes border security, counterterrorism, and geopolitical monitoring.
- Urban Planning & Smart Cities
City planners use geotagged text (like complaints, service requests, or social posts) to understand local issues—traffic, pollution, infrastructure needs, etc.—and improve city services.
- Journalism & Media Monitoring
Newsrooms analyze large volumes of text to track where events are happening and map stories geographically.
- Public Health & Epidemiology
Geotagged text helps track disease outbreaks or health trends. For instance, analyzing news articles and social media posts mentioning symptoms and outbreaks can help detect early spread patterns in locations mentioned in text.
- Insurance & Risk Analysis
Insurers monitor media to assess regional risks (e.g., political instability, natural disasters) that could impact markets or assets.
- Environmental Monitoring
Researchers analyze reports, social media, and field notes to detect environmental changes—like wildfire mentions or pollution incidents—in specific locations.
- Logistics & Supply Chains
Companies use geotagged customer complaints or news reports and other media to identify disruptions (e.g., delays at ports, regional shortages, political unrest) and adjust operations.
How Does Geotagging Work?
Geotagging takes place in two main steps:
1. Identify locational information such as place names in natural language text.
This is not a trivial task. First, place names are often ambiguous. For instance, Washington can be a city name, a state, a person’s last name, part of an organization name (e.g., Washington Post), etc.
Second, it is important to be able to identify place names that may be unknown, misspelled, or transliterated in a novel or unusual way. To overcome both challenges, geotagging must rely heavily on the linguistic context.
In addition to place names, there are other types of locational information that need to be identified. For instance, relative locations such as “50 miles northeast of Riyad” are very useful in particular for military and law enforcement applications, which deal frequently with unnamed locations.
2. Assign geo-coordinates.
This step requires a large list of place names and their geo-coordinates, known as a gazetteer. However, it is not just a matter of performing a list lookup. In fact, place names are often very ambiguous: the same place name can refer to multiple locations. For instance, there’s a “Springfield” in many U.S. states. Is it Springfield in Massachusetts, Virgina, Texas? Here again, the key is context.
Geotagging can resolve the ambiguity by evaluating the overall context supplied by the surrounding unstructured text. If there’s mention of “Worcester” or “Boston,” then it’s likely to be the “Springfield” in Massachusetts, not Illinois.
Beyond Place Names: Geotagging Relationships and Events
Geotagging is even more useful when it is combined with the ability to extract relationships and events from unstructured text.
Relationship extraction extracts key associations between entities in unstructured text. Consider a piece of unstructured data like “Hobson Corporation is located in Brussels.” Relationship extraction will identify or “extract” both “Hobson Corporation” and “Brussels.” It knows that the former is a company and the latter a location. Once the org-location relationship is identified and Brussels is assigned geo-coordinates, this information is then structured, which makes it easily convertible into a standard format that a GIS tool can accept and render in visual terms:
RELATIONSHIP: ORG-LOCATION
ORGANIZATION: Hobson Corporation
LOCATION: Brussels, Belgium (50.8477⁰ N, 4.3572⁰ E)
Based on this input from geotagging, a GIS tool can, for example, display not just locations but also other types of entities such as organizations, people, and artifacts on a map. All of this is the result of relationship extraction.
Geotagging can also be combined with event extraction to locate events on a map. Given a sentence like “John Donnelly visited New York City,” Event Extraction will extract the person “John Donnelly” and the place “New York City,” and also identify that there is a travel event for “John Donnelly” to “New York City”. This capability enables GIS to display events and their participants on a map.
As an added feature, event extraction can also identify the time of an event if the information is present in the unstructured text. This enables events to be aggregated and plotted on a timeline. For instance, the progression of a disease outbreak can be tracked over time visually on a map.
Other event types can be similarly combined with locational information and plotted on a map, such as meetings, attacks, movements of vehicles or troops, etc.
Summary
Geotagging allows conventional geospatial applications that traditionally required structured data to exploit unstructured data for geospatial intelligence. Geotagging enables GIS applications to display a far richer geospatial view that goes well beyond a “set of points” on a map.
