Abstract
AbstractIt took a while, but natural language generation is now an established commercial software category. It’s commented upon frequently in both industry media and the mainstream press, and businesses are willing to pay hard cash to take advantage of the technology. We look at who’s active in the space, the nature of the technology that’s available today and where things might go in the future.
Highlights
Since the early days of computational linguistics, research in natural language generation (NLG) – traditionally characterised as the task of producing linguistic output from underlying nonlinguistic data – has often been considered as the ‘poor sister’ in relation to work in natural language understanding (NLU)
NLG was for a long time a niche interest
I haven’t carried out a detailed analysis of the papers published in worthy venues over the last few decades, but I think a reasonable guess is that, averaged over the years, only around 10–15% of the papers published in NLP conferences and journals are concerned with generating language, whereas 85–90% are concerned with language understanding
Summary
Since the early days of computational linguistics, research in natural language generation (NLG) – traditionally characterised as the task of producing linguistic output from underlying nonlinguistic data – has often been considered as the ‘poor sister’ in relation to work in natural language understanding (NLU). As in many other technology domains, it’s common for the vendors to offer bespoke application development via a professional services model: the customer believes they have a problem that can be solved with NLG, and the vendor will build and maintain the corresponding solution for them. Generally developed on the basis of learnings and experience acquired from activity in the first stream, it’s common for the vendors to offer self-service toolkits that third parties can use to develop their own applications, typically these days via a SaaS model. Making money under the first stream is a challenge for a new technology area, because rational price points are not well established, and companies risk either underbidding to get the business or scaring away the customer with sticker shock. Is a wee table that shows which NLG vendors offer, or have offered – sometimes the current status of the product is unclear – integration with each of the main BI tools
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
