Seven roads to data-driven value creation

Predicting crime

Predicting deals

Predictive maintenance

The cold start problem

Risk missing the long tail, algorithmic discrimination, stereotyping

Neglect of novelty

Amazon’s product recommendation system

Google’s “Related searches…”

Retailer’s personalized recommendations

The cold start problem, managing serendipity and filter bubble effects.

Finding the value proposition which goes beyond the simple “you purchased this, you’ll like that”

Clarivate Analytics curating metadata(data, data curation) from scientific publishing

Nielsen and IRI curating and selling retail data

ImDB curating and selling movie data

NomadList providing practical info on global cities for nomad workers

Slow progress: curation needs human labor to insure high accuracy, it does not scale the way a computerized process would.

Must maintain continuity: missing a single year or month hurts the value of the overall dataset.

Scaling up / right incentives for the workforce: the workforce doing the digital labor of curation should be paid fairly, which is not the case yet.

Quality control

Selling methods and tools to enrich datasets

Selling aggregated indicators

Selling credit scores

Knowing which cocktail of data is valued by the market

Limit duplicability

Establish legitimacy

Search engines ranking results

Yelp, Tripadvisor, etc… which rank places

Any system that needs to filter out best quality entities among a crowd of candidates

Finding emergent, implicit attributes (imagine: if you rank things based on just one public feature: not interesting nor valuable)

Insuring consistency of the ranking (many rankings are less straightforward than they appear)

Avoid gaming of the system by the users (for instance, companies try to play Google’s ranking of search results at their advantage)

Tools for discovery / exploratory analysis by segmentation

Diagnostic tools (spam or not? buy, hold or sell? healthy or not?)

Evaluating the quality of the comparison

Dealing with boundary cases

Choosing between a pre-determined number of segments (like in the k-means) or letting the number of segments emerge

OpenAI with ChatGPT

Legal framework still being shaped (what about the copyright of the content ChatxGPT has been trained on?)

Still new, no "textbook" on how to use it at its fullest potential