Using Sentiment Analysis to Detect Disruptive Events in Supply Chains

Contemporary supply chain operations operate on a global scale connecting multiple organizations. Beyond internal processes, supply chain performance is influenced by external events, which can lead to disruptive scenarios. Sourcing activities are particularly susceptible to disruptions. Classifying an event as disruptive or non-disruptive depending based on the perspective of a focal company can serve as a trigger system for sourcing decision-making. We propose a framework for classifying supply chain events based on risk type, geographical impact, occurrence frequency, and sentiment. Leveraging transfer learning, we train a sentiment analysis model to assess the relevance of news headlines related to supply chain events.