Crowdhelix Enhances its Recommender Engine by 50%

Crowdhelix has developed new AI technology to enhance its recommender engine by 50%, making it easier for users to collaborate with experts across research, innovation and business.

This massive enhancement is the result of a nine month project led by Crowdhelix’s Machine Learning Engineer, Alex Lilburn.

The pioneering system employs trailblazing transformer models, similar to those used in ChatGPT, to help users identify and collaborate with Crowdhelix’s network of over 11,000 expert innovators spread across 57 different countries.

‍

What Does This Mean for Crowdhelix Members?

Members of the Citizen Science Helix Community will continue to use the Crowdhelix platform as they did before. Changes have only been implemented in the background.

The new Recommender System makes it quicker and easier for members of the Citizen Science Helix Community to connect and collaborate with like minded experts from across the Crowdhelix global network.

Because the new recommender engine is aware of context, it recognises that people describe themselves and their skill sets in different ways and that opportunity posts published on the Crowdhelix platform vary hugely, even if they are published by experts working in the same field.

As such, Alex build the new recommender engine to see beyond superficial differences to bring researchers, innovators and business leaders together. Which means that members of the Citizen Science Helix Community will now see more opportunities for collaboration and innovation than ever before.

‍

How does the New Recommender Engine Work

The transformer models employed by the new recommender engine act as a neural network that understands the context in which platform users leverage keywords to search for collaborators.

As a result, Alex explains, the enhanced recommender engine is able to detect the most subtle connections between individual words and sentences, and the ways in which they interact with each other.

“We have significantly upgraded our recommender system, introducing some bleeding edge technologies that will transform how users can maximize the Crowdhelix platform. 

‍

While our previous recommender system might have identified the basic relevance between words with the same stem: the relationship between medicine and medical for example; our new engine can identify a broader set of meaningful words such as medicine, treatment, surgery and pharmacy. 

This means that our enhanced engine can appreciate context and how similar concepts relate to each other.” 

This encoding is more complex and context sensitive than any solution Crowdhelix has used before. In the same way as humans identify objects as closer or further away, the new recommender engine does the same for concepts in the very high dimensional space that it's encoding these pieces of text into. 

To begin using the new recommender engine, all you have to do is to log into your Crowdhelix account and look for opportunities.

‍