Tracking Technical Emergence:Can we predict the future?NILS NEWMAN

NEWMAN@SEARCHTECH.COM

SEARCH TECHNOLOGY

Why the Future? Identifying who, what, when and where

in scientific, technical and patent information is now a relatively well understood process.

Given current technical and computational advances, can we go beyond benchmarking and turn our attention to forecasting the future?

What could we actually forecast? There are several facets of future events that would be interesting.

Could we forecast:

The next “aha!” content and moment?

Invention?

Innovation?

Probably not. Not effectively and at least not directly.

So what can we forecast?

Technical Emergence

What is Technical EmergenceTechnical Emergence is a concept that has attributes of:

Novelty

Persistence

Community

Growth

For Example: The Quantum DotThe history of quantum dot

Aha! in 1981 Invention in 1986 Innovation -- 2019?

But Quantum Dot has trackable

Novelty Persistence Community Growth

The four dimensionsTo be emergent, a concept must have all four attributes.

All four attributes exist as traits in the scientific, technical, and patent literature.

All four traits can be measured using bibliometric and ‘tech mining’ techniques.

This combination means we might have a chance to do effective forecasting.

The forecastUnfortunately, each attribute presents unique forecasting challenges.

Novelty One cannot really predict the appearance of a concept that does not yet

exist; but one can analyze the past rate at which new concepts have emerged within a specified technical area.

One can track and forecast progressions of incremental change.

One can also use past activity to determine a probability for future radical change, but with a higher degree of uncertainty.

Persistence Persistence is easy to measure and allows for effective forecasts.

However, there is still a great deal of uncertainty as to how persistent a concept must be to be considered persistent.

The scale of the concept and the technical area both impact the behavior of persistence.

Persistence is also a great source of noise in forecasting process

Community Community can be difficult to measure. Complete and accurate information on all documents’ contributors is

not always available. Web of Science/Scopus work reasonably well. EI Compendex/INSPEC

are more problematic. Patents vary, depending on authority. Cleaning of organizational and personal names, and accurate matching

of people with organizations, is a challenge. Once cleaned, the social network analysis needed to measure and

forecast community is reasonably well understood. However, there is still debate on the level of analysis to apply when

looking to ascertain “a community.”

Growth Growth is tricky in that it comprises

multiple dimensions: Growth within the concept’s technology

space.

Growth into other technology spaces.

Growth within a community.

Once identified, there are a variety of ways to forecast future growth.

One technique involves curve fitting to logistics curves.

Sounds difficult…. So what can we actually do?

What would we do with it?

To simplify, let’s start with what to do with it One of the best places to apply

this kind of forecasting process is in Open Innovation (OI).

An effective forecast of emergence can yield new ideas within a technology space and, more importantly, find new partners for your organization.

So let’s focus on OI partners The emergence analysis would start with a

broad search of a general technology area of interest to your organization.

The search would cover patents, publications, and any other relevant information sources.

The search would include a back file of at least ten years.

The Emergence Calculations The emergence calculations would use

part of the data set as a benchmark file to determine normal behavior.

The most recent records would be analyzed for novelty, persistence, community, and growth.

This analysis would provide a set of likely emergent concepts.

Pivoting to People The current state of emergence research means that a list of terms will be

more than a bit noisy. The current processes have a weakness due to boundary issues:

• What is the most effective level of analysis?• What is the ideal community size?• What is the boundary between persistent and old?• What growth rate is optimal?

These issues introduce false positives into concept lists. One way around this issue is to focus on organizations more than on terms. Working hypothesis: Organizations with a higher concentration of emergent

concepts are better candidate partners than organizations with a lower concentration.

The output The output of an emergence analysis

would be: A list of organizations and researchers who

are operating in your area of interest.

These people and organizations show activity in cutting edges of your target space.

They could be useful resources for your Open Innovation efforts.

Back to Quantum Dots… Running this kind of analysis on

Quantum Dots yields roughly 30,000 phrases in around 4,000 patent abstracts, of which nearly 100 terms qualify as emergent concepts.

As expected, the organizations which strongly use emergent terminology are primarily University and Government labs and are mostly located in China.

The exercise produces a quick list of organizations on which to focus a human-centric OI vetting process.

Summary Analyzing Technical Emergence appears to be a

promising approach to forecast technical activity.

Currently the approach is more effective at targeting organizations and people. Specific topics can be extracted, but false positives are still a significant issue.

As research in this area increases, precision and recall of Emergence will improve.

Understanding of Technical Emergence could potentially have significant impacts on Open Innovation and research strategies in general.

Discussion

Thank You! Merci Beaucoup!