Recommendation systems are ubiquitous on digital platforms. However, many systems rely either on the availability of large amounts of data that allow for data-driven optimization (collaborative filtering), or they are rather simple and lack the possibility of intelligent recommendations. Especially in the fields of life sciences and facility management, there are a large number of potential applications of recommendation systems which cannot be supported by large amounts of data. In such a case, the possibility of intelligent recommendations is typically implemented by an expert system, with the disadvantage that the data-based learning capability of the system is mostly very limited.
To overcome this limitation, we are working together with Dayzzi to develop a novel hybrid recommendation system that combines an expert system with a learning neural module that enables online learning.