Presentation about new data, methods and outputs to create knowledge for innovation policy. Presented at the OECD Blue Sky Conference, 20 September 2016.
Schema on read is obsolete. Welcome metaprogramming..pdf
New Data for Innovation Policy
1. New data for innovation policy
Hasan Bakhshi and Juan Mateos-García
OECD Blue Sky conference
20 September 2016
2. Roles of data in innovation policy
Set agenda
Design
Implement
Evaluate
Includes...
• Identifying and measuring new
industries
• Innovation in non-S&T intensive
sectors
• Assessing systems and networks
• Informing operational delivery
• Informing decisions by a variety of
actors in the innovation system
• Predicting as well as measuring
3. What do we mean by innovation
analytics?
Novel data sources and combinations of data,
methods to analyse them and formats for
dissemination that generate useful information for
innovation policy
• Not always ‘big’, and also including open and official
data
• Including existing methods that are becoming
easier/cheaper to apply
• Also disseminating data
4. Opportunity 1: Identify novel areas of
activity
• Web scraping of company
websites and directories
• Use of keywords/NLP for
classification and
measurement
• Includes non S&T sectors
Challenges
• Biases
• Reproducibility
• Consistency
SIC code distribution of UK video
games dataset obtained via web
scraping
References: Nathan and Rosso (2015), Gök et al (2014), Shapira et al (2010), Mateos-Garcia et al
(2014)
5. Opportunity 2: Analyse complex
systems
• SNA of relational data
(structure and change)
• Merge datasets and
analyse ecosystems along
more dimensions
• Use ML to model
complex dynamics
Challenges
• Communication
• Interpretability
Inter-cluster networking between UK
creative clusters based on Meetup co-
participation data
References: Hidalgo & Hausman (2014), Tambe (2015), Bakhshi et al (2013)
6. Opportunity 3: Open up innovation
data and analysis
• Interactive data
visualisation & dashboards
for dissemination and
exploration
• Open data for further
analysis
Challenges
• Privacy
• IP
• External capabilities
References: DataViva, DataUSA, Netsights.dk, Tech City UK cluster map, Mateos-Garcia and
Bakhshi (2016).
Economic complexity dashboard including
visualisation of potential growth sectors
based on ML analysis of past development
patterns
7. Conclusion
• Exciting opportunities, serious risks, potential need for
complementary investments
• Remove uncertainty through experiments
• Enhance collective learning through openness and
collaboration
• Embrace pragmatism in data sources & use of domain
knowledge. Toolkit thinking beats silver bullet mentality
• Unlikely that robots will steal innovation policymakers jobs in
the near future.