When working with large data collections to obtain, in a short time, the most relevant information they contain, it is possible to perform pattern extraction by means of Data Mining. Among the most used patterns are Association Rules, which measure the co-occurrence of items in large collections of transactions. After many trials, one type of transaction has been found that can be treated more efficiently than the one that has been used up to now. In this study, a new method was applied to this type of transactions, which allowed to obtain in the first tests much faster execution times and more information than the one obtained with the classic Association Rules Mining Algorithms. This will allow to improve the response times of a web recommendation system to provide answers to the users in real time. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021.