Big data, bad data -- Closing keynote at the Open World Forum 2013

2 558 vues

Publié le

In this talk, I partnered with Dr. Rand Hindi (Snips) and Romain Lacombe (French Prime Minister's Commission Etalab) addressing various facets surrounding Big Data. My point was to critically discuss common misconceptions in both Big Data meaning and analysis.

Publié dans : Technologie, Formation
0 commentaire
2 j’aime
Statistiques
Remarques
  • Soyez le premier à commenter

Aucun téléchargement
Vues
Nombre de vues
2 558
Sur SlideShare
0
Issues des intégrations
0
Intégrations
1 311
Actions
Partages
0
Téléchargements
16
Commentaires
0
J’aime
2
Intégrations 0
Aucune incorporation

Aucune remarque pour cette diapositive

Big data, bad data -- Closing keynote at the Open World Forum 2013

  1. 1. Big Data – Bad Data Rayna Stamboliyska, PhD Paris Descartes University RS Strategy
  2. 2. XXXL Data XXXL Issues
  3. 3. 300 years BC: "of making many books there is no end" 1935-1937, Social Security Act 1943-1945, 'Colossus' & Enigma 1997: first mention of 'big data' in Application-Controlled Demand Paging for Out-of-Core Visualization (NASA) “Is there anywhere on earth exempt from these swarms of new books?” ...
  4. 4. What is it?No, Big Data isn't new... but Maximizing computation power and algorithmic accuracy Drawing on a range of tools to analyse and compare large datasets Believing large datasets give greater objectivity and accuracy
  5. 5. No, Data doesn't speak of itself
  6. 6. Numbers never lie... o rly? * 51% of web traffic is non-human * there are 30 million fake Twitter accounts … poor bots on Thursday night :'(
  7. 7. Yes, Big Data discriminates between social groups Scientists are allowing their assumptions about race to shape their big-data genomics research If your past buying history indicates you are more likely to pay top euro for trousers, your starting price the next time you shop for clothes online might be considerably higher.
  8. 8. Big Data Fundamentalism? There's a word for that: apophenia! … Does big data change the erroneous 'correlation = causation' dynamic?
  9. 9. Math majors, rejoice! ''With big data techniques, you can get much closer to being able to predict causal relations.'' I beg to disagree Your question and your data are not random entities. Big Data needs several steps: ● of preparation ● in interpretation
  10. 10. And... If data are somehow subject to us, we are also subject to data

×