Yet Another Perl-Powered Company: Geolytica

Jan 16, 2025 by Ervin Ruci

Imagine you want to parse free-form address input and match it against a database representing the road network.

Example 1

• Input: "751 FAIR OKS AVENUE PASADNA CA"
• Output: "751 N Fair Oaks AVE, Pasadena, CA 91103-3069 / 34.158874,-118.151053" View example

Example 2

• Input: "5 Adne Edle Street, London"
• Output: "5 THREADNEEDLE STREET, LONDON, United Kingdom EC3V 3NG" View example

The database contains road names, shapes, numbers, zip/postal codes, city names, regions, neighborhood/district names, and more—billions of named location entities worldwide. Add another 100 million points of interest extracted from billions of webpages, and the problem becomes quite difficult.

Two decades of Perl coding, starting in 2005, and this problem is (mostly) solved at Geolytica.

Perl at Geolytica

At Geolytica, we harness Perl to manage and enhance vast geo-location datasets and build the application logic of the geocoding engines powering geocoder.ca and geocode.xyz.

Data Cleanup and Enhancement

We continuously update and enhance our location entities database, because ground truth changes - at the speed of life. One standout example is our work with OpenStreetMap’s POI data. A year ago, we utilized an in-house AI tool, (let’s call it “PerlGPT,”) to refine this dataset, correcting inconsistencies and enhancing data quality. The results were significant enough to share with the community at Free POI Data.

Perl’s Versatility and Stability

The beauty of Perl lies in its backward compatibility. Despite our codebase spanning over two decades, upgrading Perl across versions has never broken our code. In contrast, with other languages, we’ve observed issues like API changes causing extensive refactoring or even rewrites. Perl’s design allows for seamless integration of old and new code, which is vital for our specific needs.

Practical Implementation

At Geolytica, tasks like parsing location entities from text involve complex string manipulations. As shown in the examples above, these challenges would be difficult in any programming language, but Perl makes them easier than other options.

Final Thoughts

The best programming language for any job is the one that makes hard problems easy and impossible ones possible. For Geolytica, that language is Perl.

About the Author

Ervin Ruci has been immersed in Perl since 1998, initially building student information systems at Mount Allison University, then registry systems for CIRA from 2000 to 2005. In 2005, he became a location-independent entrepreneur, founding Geolytica to tackle the location intelligence problem.