Tue Sep 15 2020

Talent Ticker’s Artificial Intelligence Now Speaks Over 100 languages!

As a SaaS platform built specifically for the recruitment industry, we have always understood that the future of recruitment is borderless, but how can a single team have eyes and ears in hundreds of different markets?

We found early in our development that AI was the only way we could deliver unique, easy to use and extremely powerful systems that met the goals of a recruiter.

The artificial intelligence platform uses natural language processing (NLP) to read and understand news from a multitude of sources including RSS feeds, social media posts, company and regulatory news feeds. This engine has the ability to process 100,000’s of news stories each day, sort them, understand them and only show the ones that are relevant to recruiters.

Even better these are all tagged, sortable and searchable.

The artificial intelligence model was initially developed and trained to read and understand English and therefore able to read and process information the world's information... as long as it’s in English.

As we know, recruitment is becoming more and more boarderless, surely we could do even more to enable this?

We asked the question to our AI team and recent developments have enabled the AI to learn over 100 other languages. We are now able to read and understand content in foreign languages and has become a truly multilingual system. Only 16% of the world's population speak English natively, this breakthrough enhancement now caters for greater global content coverage.

This enables a recruiter to have full market knowledge of their vertical, worldwide.

This remarkable achievement allows the inclusion of the worlds localised news feeds, regardless of the language. Over the coming weeks and months, there will be an influx of new content sources added to the platform, providing much deeper insight.

The language support is provided by an open-sourced neural network based technique for natural language processing (NLP). Using a pre-training method called Bidirectional Encoder Representations from Transformers, otherwise known as BERT. This provides a technique to pre-train a neural network the ability to understand languages by translating a random small set within two translated documents and using the remaining portion of the document to test and validate itself. Developed by Google the BERT method, in particular, allowed Google Search to better understand the placement of words and their importance in a search phrase.

Further enhancements to the Talent Ticker multilingual processing are planned, which will improve the multilingual event classification, making the publication of news faster.

Want to see how Talent Ticker would benefit your business? Let us take you through the software.