Network analysis inside the Italian parliament
summary
Daily ETL pipes pull public SPARQL endpoints and scraped vote records into Postgres, sync to Neo4j and power interactive Linkurious views, using Django + Neo4j + NetworkX and a tiny oetl library. Voting is hot‑encoded voting similarity across five categories (yes/no/abstain/absent/secret), weighted, time‑decayed and multiplied into an agent similarity matrix to build an undirected weighted MPs graph. Degree, betweenness and closeness plus clustering reveal the current majority moves as a single entity, flag unexpected cross‑block connectors, and surface MPs who later switch parties.
AI-written summary · about
chapters (16)
- 0:00Introduction and welcome
- 1:52About Openpolis foundation
- 3:16Openpolis projects overview
- 5:20Italian parliament system explained
- 7:31Data collection and pipelines
- 9:18ETL libraries for data processing
- 12:30Social network analysis basics
- 15:49Sociological approach to analysis
- 16:53Voting weight metrics creation
- 18:43NetworkX graph construction
- 19:26Analyzing weighted networks
- 20:32Senate network during Electa government
- 21:35Current legislature network analysis
- 23:00Majority vs opposition dynamics
- 24:28Future improvements and challenges
- 25:11Q&A session
related talks
40mFriendly Borders: Graph algorithms reveal Eurovision voting patterns
39mUse of graph-oriented database in the detection of fraud in quotas for parliamentary activity
22mElection Night Reporting Using R & Quarto
01h 41mGraph Intelligence: Enhance Reasoning and Retrieval Using Graph Analytics
40mBridging voter data and social data using Apache Streams
24mOur daily graphs written in Ruby and Neo4j
43mGraphs: Datastructures to Query
42mFighting fraud with graph databases