NLP (Natural Language Processing)
I’ve been reading a lot lately about machine-learning algorithms and neural networks, particularly about their potential to turn the science of business forecasting and supply chain management on its head. I’m not here to make that case or to delve deeply into supply chain AI tools. Blogs, chat rooms and newsfeeds are already abuzz with the latest claims to AI by everyone from supply chain execs and consultants to market researchers and publishers. Instead, I’ll review Natural Language Processing (NLP) – specifically, the potential for increasingly smart human-computer interaction to make sense of unstructured data (news, social media, etc.) “signals” of upstream supply-chain activity.
Think about it. When you’re a key link in the middle of a multi-billion-dollar supply chain, mastery of supplier quality management is crucial to your ability to satisfy downstream demands. If your clients are sought-after consumer brands, you’re under constant pressure to deliver high-quality products and supplies, on time, and at the lowest cost possible.
This is easier said than done given potentially enormous disruptions from shortages, recalls, contaminations, outbreaks, human rights investigations, safety infractions, environmental hazards, and so much more. Yet getting goods moved through your place in the value chain is your reason for being. It’s fundamental to your reputation and financial solvency.
Advantages of NLP
Ask yourself, how great an advantage would the ability to scan, pinpoint and manage potential supplier disruptions before they actually occur? I contend that this advantage is potentially enormous, and increasingly viable due to continued advances in data mining and Natural Language Processing.
Let’s quickly define Natural Language Processing. NLP blends computer science, artificial intelligence, and computational linguistics in ways that improve the interaction of computers and humans, or rather the ability of computers to accurately interpret the natural language of our texts, tweets and news feeds. Indeed, technology is getting better. And this improvement will only accelerate as their self-learning algorithms gain exposure to increasingly massive amounts of digital media (unstructured data).
From this vantage, I imagine increasingly intelligent supply chains. Think of supplier-health scorecards that factor potential hazards, trends and other phenomenon gleaned from Natural Language Processing and analyses of news, social media and other unstructured data. That’s essentially adding a layer of insight and analysis to your existing historical time-series framework, which most organizations already use to forecast various aspects of their supply chain. Now, what if organizations can also capture data on the health of suppliers, or on potential health impacts to suppliers? BTW: When I say health, I mean as it pertains to suppliers’ ability to meet order fulfillment targets without disruption, delay, or loss of quality.
Consider the following. One of your key suppliers suddenly loses a huge contract that up until now drove half of their revenue. Are they more likely to face liquidity shortages? Probably yes. And you wouldn’t know about it until it’s too late.
Leveraging NLP for Supply Chain AI Tools
That’s unless you were leveraging Natural Language Processing and text analytics to monitor the internet for signals of instability. Various combinations of various words and phrases could have generated a potential disruption signal, prompting a set of alternative forecasts and tests meant to simulate any and all impacts to your supply chain. The same scenario could play out if your supplier gets acquired, hit with a law suit, or trapped under a government inquiry. The point is just that the more insight you have into your supply chain and its health, the better off you and your customers will be.
Stay tuned for our forthcoming series on supply chain AI tools and the latest innovations in AI for supply chain.