IBS Software Built Bilingual AI NER Model for Cargo Logistics on Bedrock

AWS Machine Learning Blog published a technical breakdown on how IBS Software built a bilingual named entity recognition (NER) system for cargo logistics using token-based knowledge distillation on the Amazon Bedrock platform.

What is NER and Why Does Logistics Need It

Named Entity Recognition (NER) is a natural language processing task where a model finds and classifies specific entities in text: company names, addresses, container numbers, dates, port names. In cargo logistics, documents (invoices, customs declarations, instructions) often arrive in different languages, and manual data extraction from them is a labor-intensive and error-prone process.

How Token-Based Knowledge Distillation Works

Knowledge distillation is a technique where a smaller, more compact model is trained to reproduce the behavior of a larger model while preserving most of its accuracy with significantly lower computational costs. The token-based approach applies distillation at the level of individual text tokens, which is particularly important for the NER task, where the accuracy of marking each word in a sentence is critical.

• Technology—bilingual NER model for cargo logistics
• Training method—token-based knowledge distillation
• Platform—Amazon Bedrock, leveraging AWS knowledge distillation capabilities
• Case author—IBS Software company

Using knowledge distillation instead of deploying a full-size large language model allows IBS Software to achieve acceptable entity recognition accuracy at lower inference costs—critical for logistics companies processing large volumes of documents daily.

What This Means

The IBS Software case demonstrates a practical pattern for industries with narrow-specialized and multilingual data: it is not necessary to keep a heavy universal model in production—a compact distilled model trained for a specific task like NER can solve it more efficiently and cheaply.