The Little Noticed Power of Geotagging and Natural Language Processing

Content analysis software Document categorization software Unstructured text


Geotagging documents

Data is a sleepy term that’s rising in worth every second. It’s estimated that, on a monthly basis, Facebook has 1.97 billion users around the world producing unquantifiable amounts of data every second.

Location is a large part of data sharing. We live on a big planet, but between text, pictures, and global positioning, pinpointing a person’s location within a very small margin is easy. Let’s look at the combined power of Natural Language Processing (NLP) and Geotagging.

Natural Language Processing

The market of text analysis (yes, it’s a market) is worth an estimated $3 billion and is predicted to jump to $6 billion by 2020. Being of such worth has software companies scrambling to build the most advanced software systems to analyze text information.

NLP is a branch of computer science–artificial intelligence–that looks at how computers understand and process human language. Your typed speech goes into a computer, is processed by the computer into useful information that breaks down meaning as accurately as possible. An important part of entity extraction, it breaks down raw text either input by a user or existing text applied to read information. From this data, NLP will attempt to extract exactly what you mean, harnessing data points that tell where you are, what you’re close to, what you want, who you are, and a range of possible meanings.

It’s incredibly complex, as humans have the ability to translate context, semantics, grammar, and meaning in natural language, whereas computers cannot without being so programmed. Therefore the programs have to disambiguate our language and attempt to best understand if we mean, say, Washington the state, the apple, or the president.

Smart Geotagging

Most people have a smartphone of some type. Hidden in a menu somewhere on your phone is some sort of built in location service based on GPS coordinates pinging your device’s hardware. As long as it’s running in the background and it’s in your pocket, there’s a live, accurate location of you/your device. A simple example is taking a picture and your device offering the option to locate your photo (e.g. the location the photo was taken) in a geotag.

While we won’t go in depth on geotagging software, there are two key parts: 1) named entity recognition tools, which use NLP to process text to identify possible places. 2) geo-resolution, which looks up possible location coordinates and runs them against named entity recognition to provide the closest possible match to a valid geographical location.

Together, these build advanced geotagging software. A complex process that most don’t realize is constantly running on their device or an option when they take a picture of their dinner at a restaurant. The data pulled from this can inform companies on the habits of individuals every second of the day, giving the ability to use that data to make informed decisions.

Leave a Reply