Skip to content

Latest commit

 

History

History
70 lines (36 loc) · 2.27 KB

debate_analysis.md

File metadata and controls

70 lines (36 loc) · 2.27 KB

Parliament debate analysis

Version 02

Your goal is to create AI agents using the platform crew.ai to extract information from the minutes of the European parliament debates.

please create a crew of agents

For each intervention :

  • extract speaker name, political group and role (NER)
  • extract the specific arguments, situation assessment and proposals for each speaker.
  • output the results in a JSON format

Input:

  • The input is a URL to the debate online page. not the full text

The code should

  • use LLMs for NER and topic / arguments extractions
  • instanciate an OpenAI LLM with model name and temperature

Instructions :

  • Verify that all speakers and interventions are accounted for
  • Keep the number of agents limited
  • Keep the code simple
  • Write the prompts in an external json file so that the prompt texts are not directly integrated in the code

version 01

https://claude.ai/chat/949665fb-b0a6-4211-a7ad-560c250afd63

The goal is to create AI agents using the platform crew ai to analyze, extract information and understand the minutes of the european parliament debates on multiple topics.

I want to create a set of crew.ai agents to understand the controversies and arguments from the different countries and political groups

Given a list of documents of the verbatim reports of the european parliament

I attached a pdf version of one of the debate for illustration of the raw content

The information extraction agent task is to

  • identify the topic of the debate
  • identify the speakers, their political affiliation, their country
  • summarize the main arguments of the different speakers

focusing on specific positions and do not take into account general statements on the danger of organized crime

The political analyst agent is an expert in EU politics Its task is to

  • find the oppositions in terms of assessment of the current situation or problem and suggested solutions ?

The data analyst agent will

  • organize the data into a JSON format with keys : Topic, Position, Details, list of Speakers

please walk me through creating crew.ai agents leveraging its tools (SerperDevTool, ScrapeWebsiteTool, WebsiteSearchTool and others if needed) to carry out the different tasks

https://www.europarl.europa.eu/doceo/document/CRE-10-2024-09-18-ITM-007_EN.html