Multi-Lingual Machine Learning API Deployments

March 13, 2018

Machine learning API ParallelDots has a story on launching their APIs in multiple languages. Allowing them to “serve a truly global customer base with following language options for our key APIs (Sentiment Analysis, Emotion Analysis, and Keyword generator)”. Something that I think more APIs providers are going to have to think about in coming years, as the need for API resources expands around the globe.

I’m tracking on the localization efforts of different API providers, and I’d say that having service availability in different regions, and multi-lingual support are the top two areas I’m seeing movement. I see two driving forces behind this, 1) the customers are demanding localization for their businesses, and 2) the governments in those countries are imposing regulations and other laws that dictate where resources can be stored and operate. I guess, something that can be seen as markets working things out, if you also believe in the value of regulations.

I also see a negative in all of this, specifically in this case, the imperialistic aspects of artificial intelligence and machine learning. Meaning, the models are trained here in western countries, but then being applied, injected, and imposed upon people in other countries. Selling the service as some sort of truth, watermark, and organic solution to sentiment, emotion, and intelligence without actually localizing the models. I’m not making any assumptions around ParralelDots motivations, just pointing out that this will be a problem, and should be ignored–if it makes you mad, then you are probably part of the problem.

As I conduct my API research, I will keep tracking on localization efforts like this. When I have the bandwidth I will dive in deeper and better understand how providers are defining these localization layers to their APIs in their API design, deployment, documentation, and other elements. I’ll be keeping an eye on which industries are moving fastest when it comes to API localization, and try to understand where it is deemed the most important by providers, consumers, and regulators. Lots to keep an eye on, and understand in 2018 when it comes to the expansion of the world of APIs.

Read more...

Previous Article
Google Releases a Protocol Buffer Implementation of the Fast Healthcare Interoperability Resources (FHIR) Standard
Google Releases a Protocol Buffer Implementation of the Fast Healthcare Interoperability Resources (FHIR) Standard

Google is renewing its interest in the healthcare space by releasing a protocol buffer implementation of t...

Next Article
API As A Product Principles From Zalando
API As A Product Principles From Zalando

As I’m working through the API design guides from API leaders, looking for useful practices that I can inc...