A new recommendation engine technology has been designed without the need for an Artificial Intelligence (AI) model. This technology instead utilises hypergraph theory (a powerful method for visualising complex relationships between products), enabling businesses to provide customers with improved product and service recommendations, while reducing privacy issues.
As customer services move increasingly onto digital platforms (such as online shopping and streaming platforms), there is a growing need for improved recommendation engines for customers. Currently, automated recommendations for related products are often viewed poorly by users, with low quality choices that don’t have clear links between them. This not only reduces customer satisfaction and brand trust, but in some cases, it can even pose a health and safety risk – such as substitution recommendations for grocery food items not taking allergens into consideration.
Privacy is another issue with the current recommendation engines on the market today. Customer recommendations are driven by user data and previous customer purchases or viewings, however there are growing concerns over the reach of what information is collected and how this data is used and shared. This is only increased by the use of AI models, which lack transparency and auditability. Using customer data is also computationally expensive, particularly for large AI models, and requires a high amount of power, increasing cost and environmental impact.
A new solution from Dstl overcomes these current issues with a highly sophisticated and powerful data insight and recommendation engine technology: Ordinal Relationship Briefing (ORB). ORB does not employ an AI model; instead, ORB uses pre-processed datasets and hypergraph theory to analyse a huge range of specific product details and attributes to identify more elevated relationships between them, without relying on user data. This has the potential to greatly improve product recommendations for customers – whether for retail products and food items in online shopping, or TV, movie, and music recommendations on streaming platforms.
ORB directly compares products to each other only, removing the need for the collection, storage, and use of customer data for recommendation purposes. As the technology does not use an AI model, computational power and costs are also drastically lowered.
This technology can also be used by brands to measure trends and connections in purchasing habits and product features down to minute details, with no individual customer data required.
Customers can benefit from an improved online shopping experience at supermarkets and broader online retail shopping, and a better streaming experience for movies, TV shows, music, and more. Recommendations and product comparisons can be made from specific product details without the need to use customer data or previous purchase histories.
As the recommendation is powered by specific details about the product, recommendations can take into account cost, stock levels and supply chain complexity to provide significant benefits to both the sellers and customers.
Brands can benefit from increased sophistication in their analysis abilities. For example, the ability to measure if products with specific ingredients or certain levels of vitamins are more popular in grocery items, or if a certain fabric, style of sleeve, colour or hem line is most popular in fashion retail, or if certain genres or sub-genres, lead actors, or song lengths are most popular for streaming platforms. This can all be measured and analysed without collecting individual customer data.
This technology provides transparency and auditability to the horizon scanning process – enabling an understanding of what has already been done in a technology space, identifying capability gaps, and visualising and exploring previously unseen connections between capabilities. ORB is compatible with a vast variety of data sources including patents and publications.
If you would like to discuss this technology or collaboration opportunities with our team, please get in touch below.
"*" indicates required fields
© Ploughshare Innovations Ltd 2025. Registered in England and Wales No. 04401901
