The Design and Implementation of APLOS: An Automated PoLicy DOcument Summarisation System

Many institutions write and publish policy documents to inform stakeholders or citizens about priority areas and legislations or regulations governing issues such as the environment, agriculture, food safety and business operations, to mention a few. These documents are often lengthy, have different formats, are full of jargon, may have many versions, and require domain knowledge to understand the policy issue. These characteristics make it challenging to develop a personalised tool for users to generate concise summaries based on their domain knowledge, refine the summaries without compromising the original meaning of the policy, and find related policies. To solve these problems, we leverage advances in foundation models (FMs) and Large Language Models (LLMs) to develop a tool called “APLOS” for summarising and gaining insights from policy documents. The architecture, implementation, and natural language processing challenges we faced in developing the tool and the future directions we are undertaking to address the challenges are presented and discussed.

Citation: