When creating an NLG system, decision makers should bear in mind that the recipe for this simple dish includes only four primary ingredients:
- Structured data (you’ve got this in house)
- Subject matter expertise (you again)
- NLG writing talent (Qwerticulation)
- And the technology to pull it all together (Qwerticulation uses AX Semantics).
Data Forms the Foundation
Successful NLG projects are built upon structured data sets. This is why some of the earliest NLG efforts began with weather reports and sports statistics. Professionals in these fields reliably produce data sets that that contain neither null nor misplaced values.
Structured data can be found in nearly every field of business. Often, this data already serves as the foundation for content pieces that are produced linearly, one at a time, by humans writing about stocks, real estate, demographics, sales results, etc. NLG can be a huge help here, removing the requirements of prioritization and, ultimately, triage. But even if—and sometimes especially when—the data is not currently being analyzed at all, the resulting NLG output can amount to a windfall of monetizable content.
Ask yourself: If we had the luxury of hiring an unlimited number of subject matter matter experts who also happen to be talented writers, and if we turned this legion loose upon our structured data, what content could we instantly produce? How about on a regular (quarterly or monthly) basis? Whether monetizable by sale to third-parties, or by internal use among employees, the resulting output will constitute and enduring asset to the business.
Subject Matter Expertise is Essential
In any NLG project, structured data must be paired with subject matter expertise. You’ll likely find it among your colleagues, who’ll need to be on board to spend time on the NLG project. The output from the NLG system must reflect their knowledge, and the terminology and sensibility of your industry.
Within the context of NLG, the beguiling terms “machine learning,” “deep learning,” and “AI” can lead to the misconception that if we pour enough data about any given subject into the computer hopper, the computer will become a subject matter expert. Futurists may suggest that we will someday turn to machine learning and artificial intelligence not only to do most of our writing in various business fields, but also to produce creative work, such as plays and novels. For the sake of argument, let’s concede that the futurists may eventually turn out to be right . . . but not within the timeline of any current or foreseeable NLG project, just as the next car we buy will not fly over traffic jams. Therefore decision makers should furrow their brows at assurances that expertise can be gleaned instantly from data itself, without the extensive involvement of their companies’ subject matter experts. Indeed, an NLG project requires their continual engagement. Daily schedules should be planned accordingly.
The Project Begins and Ends with Writing
It may be tempting to approach NLG as a technological puzzle rather than as a writing assignment, which is exactly what it is. Written output is the final and, indeed, the only expression of the entire endeavor.
The success of an outsourced NLG project depends upon the pairing of in-house domain experts and third-party NLG writing talent. These project colleagues should meet weekly, if not more frequently, to review the prose output as it evolves. Over time, these meetings will transfer to the writers a measure of industry jargon, corporate sensibility, and subject matter expertise, and to the in-house domain experts a measure of NLG writing skills that will serve the firm well long after the engagement is complete.
Technology Conducts the Assembly
The possibility of NLG did not arise from the imaginations of traditional writers, who for millennia have been content to eke out one sentence at a time. Today’s rapidly evolving NLG capabilities are a tribute to the creativity and vision of computational linguists. For external publication purposes, however, the writers naturally set the standards for creativity and quality, with the goal of producing prose that should feel far removed from the computer science labs where NLG was first conceived. The finished prose should bear no whiff of the technology, just as a polished diamond does not call to mind the jeweler’s tools. For more on how this magic trick works, see NLG Demystified.
Copyright 2021 Qwerticulation LLC
About the author:
Greg Williams is the co-founder of Nila June Instant Property Descriptions, a natural language platform that produces property listing descriptions based on real estate agents’ responses to a simple survey. As head of product for Reis, Greg produced the CRE industry’s most widely deployed NLG system for the evaluation of commercial real estate markets, submarkets, and properties throughout the nation.
Qwerticulation is an AX Semantics Gold Managed Services Provider.
“Qwerticulation” is a portmanteau of “QWERTY” and “articulation.”
