Session IV - Technical Data Strategy

Participants

Cindi Thompson, Chris Wensel, Reynold Xin, Greg Rokita

Members Excused

John Carnahan, Rahul Pathak, Ali Ghodsi, Mark Madsen

Administration

To encourage participation, only session summary will be transcribed.

Topic Discussion Summary

Topic introduction

• Business Data Strategy
• Relationship between Business Data Strategy and Technical Data Strategy

Technical Data Strategy Requirements

• Easy ingestion of any data by anyone
• Unified streaming and batch processing
• Easy transition from POC and Prod
• DWH as a property of the system (no loading or integration required)
• DWH for future data, not only historical Data
• Ease of building and deploying machine learning models
• Cost efficiency, minimal operational cost

The above may or may not be applicabe for a given comapony as there is no one unified Technical Data Strategy that fits every company

Value of data

• Differences between data and information
• Data Network Effect
• Infonomics

Deriving information from data

• Data on its own is not valuable
• Information ultimately needs to translate into knowledge
• RDF is very useful but very time consuming to curate
• Will machine learning be a scalable successor to RDF?
• Certainty of expert curated content needs to replaced with ML probability based model to make the curation scalable

November Session V Topic

Data Strategy: Architecture