AI set to tackle Increasingly complex logistics challenges

Information is key and with the internet awash with data, and both trusted and reliable a method to rapidly sift through this information quagmire, and to discard poor quality data in favour of valuable trusted material can offer invaluable support to companies with long supply chains with distant partners.

Safety and security in supply chains often comes down to knowing those that you are co-operating with, and when it is not possible to drill down to small suppliers, AI systems can recognise and link regional events or activities that may have caused issues within supply chains.

One such issue is the transportation of barbeque charcoal, which can have an accelerant added, and is known to have caused fires on board ships. Identifying regional suppliers of charcoal, often small holdings, can alert companies to potential problems.

Recognising localised challenges is one positive attribute of collecting, analysing and storing AI data, but the more readily available data on larger companies can offer alternative insights into supply chain partners, including managerial changes, financial developments and mergers and acquisitions.

Information collected globally, analysed and cross-referenced by AI could give supply chain operators critical information early on distant partners, engendering trust in those partners.

Related:Leveraging Artificial Intelligence and Machine Learning to Drive Data Visibility and Smarter Decision Making

One such company offering a data mining service is Semantic Visions (SV), based in the Czech Republic’s capital, Prague, SV has developed an AI based system that seeks data on particular companies and regions that will offer insights for business partners.

Julius Rusnak, SV’s COO, told Seatrade Maritime News that the company has the capability to download and process 2 million data pieces per day - that is news articles, blogs open-source data and proprietary data.

“We have our own database of companies that we continuously enrich with data and interconnect these companies stored in our database with events and with the relationships between them. And we also make sure that we recognise if the source is writing about a particular company or a brand in general,” explained Rusnak.

SV’s system has the capability to search in multiple languages and is backed by a multilingual team, searching items in Spanish, Chinese, German, Italian, French, Japanese and English among other languages. This team “understands the intricacies and differences in cultures and happenings in different parts of the globe”, said Rusnak, adding that the company does not merely rely on machines to analyse data.

Related:New fire detection technologies developing at speed

Rusnak concedes that to identify a business a company must have an internet presence, and that some smaller suppliers in less developed countries may not be present on the internet.

“We are experts at recognising events,” said Rusnak, “We not only use the latest, machine learning, but we also use our vast ontology that we constructed before the ontake of technology.”

This database of more than 600 predefined events and their history, which have occurred around the world, go into every type of human work, including environmental, social and governance events, both positive and negative and the system will analyse these events for their relevance.

According to Rusnak an AI system is capable of describing opportunities and recognising “patterns and meanings and semantic structures in the text” which means that the AI system is “good at recognising what’s going on”.

Constant monitoring of the internet can also act as an early warning system of an event, perhaps, port congestion, an oil spill or just a delay in transportation.

“We figure out which regions and which companies are actually mentioned in the input data and we correctly map these suppliers to the incident,” said Rusnak.

The goal of an AI data source is to give customers filtered and structured information that effectively cuts out the ‘noise’ around an event.

In effect, said Rusnak, “We take a vast amount of information in its unstructured form and convert it into high value data in a very structured form.”

Rusnak explains that while AI can offer tools in certain instances, it will be down to the reader to interpret the data that is being presented.

One illustration of this was SV’s Russian reports, which saw, what Rusnak describes as “sentiment”, stable for some years then one year ahead of the invasion “sentiment towards Ukraine skyrocketed crazily,” he said.

“It doesn't mean that we were able to make that conclusion, towards what actually happened in the world,” he explained, “But analysts that had the report in their hands, they would have been able to draw their own conclusion. So, I say, we don't interpret this stuff per se, we provide the data.”