Improving text comprehensibility through Large Language Models

Use case

Making technical and complex texts more accessible

Opportunity

Accelerating innovation in the Public Administration and improving user experience through the use of Artificial Intelligence-based solutions.

Challenge

Measuring the desired readability level (for example, to make it easily understandable even for those who aren't very familiar with complex texts), assess the improvements made, and check that the original meaning of the text is accurately preserved, using dedicated indicators and language models allowing to:

• compare the modified text with the original and make revisions
• use standard metrics (e.g.: Gulpease index) and set target thresholds
• adjust lexical paraphrasing, reorder syntax, and reorganize content
• combine human expertise and automation for scalability and timeliness
• explain operations carried out (explainability).

Solution

Tools making technical and complex texts easier and more accessible in a semi-automatic way, thanks to the use of advanced language models (Large Language Models) and the subsequent review by a human user.