AI4PublicPolicy Policy Extraction and Recommendation toolkit part 1

Part 1 – Solution Design

The AI4PublicPolicy Policy Extraction and Recommendation module provides the tools required for the creation of an AI pipeline. The trained and tested AI models delivered through this pipeline allow the policymakers to design an appropriate policy.

The design of the module has been based on two primary goals. First, the goal is to analyze and select tools that provide the necessary functionality, satisfy the project’s requirements, and integrate well with each other. Second, the complementary goal is to design a user interface that binds all the functionalities together and guides the user so that, at every stage of the policy extraction, it is clear what the current status is and how to proceed further.

Hence the above, once the extraction process is finished, it should be easy for policymakers-users to navigate through all existing artifacts, including policies, models, and pipelines.

The diagram below shows the Policy Extraction process, which starts from the policy and the datasets retrieved through the ‘Policy and Data Management’ component, in order to deliver  AI models and dashboards as outcomes for the ‘Policy Evaluation & Optimization’ component.

This process involves two actors: the Policy Maker and the AI Expert. Both actors can perform the necessary steps to achieve the goal of creating the policy’s AI models. Throughout the process, the two actors also interact with the ‘Policy Interpretation’ component to gain an interpretation of the data and the AI models.