Detection of Missing Data in Ballot Record Telegrams

THE COMPANY

Interdisciplinary team integrated by engineers, journalists, designers, programmers and data analysts to innovate the use of new technologies applying them to data and public documents for the journalistic research and coverage. The team is part of   an amportant Argentine media Company born in 1870 which objective is to provide communication solutions to their advertisers and quality and reliable products to their audiences.

THE CHALLENGE

The purpose of the Company was to visualize the ever-standing problem in all ballots and to reflect on the Argentina voting system which main instrument is the multiple ballot. This system has evidenced many failures, and our client required the analysis of effectivity in telegrams hand-filled in throughout different regions of the country upon closing of the ballot journey.

THE SOLUTION

By means of   Robotic Process Automation (RPA) and computer vision techniques, approx. 130,000 telegrams (2021 Argentina primary elections) were screened. Interpretation of different formats was firstly achieved in every district analysed (total of 5). 

Through the use of a computer vision algorithm, specially designed, potential mistakes (especially when drafting the documents) were investigated and detected clearly and precisely. Also dashboards were developed, they contained the information found in pre-set boxes, evidencing statistics on presence or absence of data (chairman of the voting station’s signature, ID #, number of envelopes in voting boxes, etc.)

BENEFITS

  • Analysis of different data sources.
  • Process automation for: telegram downloading, report generation and analysis; enabling thus the fast analysis of huge amounts of data.
  • Agile and easy visualization of statistics obtained.
  • SIGMA Awards: Prize to the World Best   Team of Journalists, rewarding this Project and other projects of the team.
Python