How Natural Language Processing (NLP) improves reference creation

Let’s face it, the process of creating references for medical affairs or promotional content is really slow. Finding the correct reference annotation can be like finding a needle in a haystack. Plus, teams are regularly hiring and training new people who need to ramp up quickly. When a reference annotation can’t be found quickly, this causes duplicate reference uploads to your library and results in bad data.

To help solve these problems we created Paige AI, an artificial intelligence assistant that recommends reference annotations based on past usage. We have been using Paige AI in beta with some customers since the summer and many of you have asked the same question – how does it work?

At its core, Paige AI uses a technology called “Natural Language Processing” or NLP for short. NLP is the ability for computers to read, decipher, understand and make sense of human language in a way that can then be valuable. 

Paige AI analyses the text you want to link to a reference, deciphers what has been written, and then searches for similar content that has already been used in past references.

Here are a few examples of some common scenarios:

Exact Text Match

Let’s use an exact text match as a baseline since it is common for content creators to reuse the exact wording verbatim from piece to piece. In this case, it has been used several times and Paige AI has a 100% confidence score. The more times this exact text is referenced, the higher the confidence level.

Partial Text Match

Let’s try this again but only highlight part of the text. This time Paige AI is less confident in the result with a 46% confidence score, but still suggests this annotation as the top match out of a database of hundreds of annotations. 

Rewriting Content

This example is the most interesting since it makes a recommendation despite the fact that the content has been rewritten. Since Paige AI understands the highlighted text to be communicating the same message as the original text, it comes back with an 83% confidence score. 

How NLP helps

Natural Language Processing helps you be more productive in several ways. First, you can create references much faster than before. Our analytics suggest that this process is ten times quicker when linking to a previously created reference. Second, onboarding new team members is faster since they will know if a reference annotation already exists or if they need to create one. Lastly, you will have less duplication in your reference database. Hard to find references causes your team to upload duplicate references and results in bad data. 

If you have any questions about Paige AI or would like a demonstration, book some time with us.

About Papercurve

Papercurve helps streamline compliance reviews in regulated industries like cannabis and life sciences. Manage approvers, comment, reference attached documents and archive past versions for audit – all in an easy to use interface.

Learn more