| Abstract [eng] |
This paper presents a new text recommendation algorithm that is a combination of ItemKNNCFCBF and MostPopular algorithms. The main objective of this research was to create an algorithm that would be capable of recommending texts written in Lithuanian, assuming that users’ details are obscure. Other existing algorithms were taken into consideration to accomplish the aim. The main criteria and conditions for the algorithm were formulated, and the key goals were raised to satisfy these conditions. Content-based methods were combined with collaborative filtering methods to create a new text recommendation algorithm. This way, a hybrid method was created, and main recommendation algorithm problems were solved. The evaluation part proved that the created algorithm generates better results compared to other algorithms that were analyzed in this research. The effectiveness of text recommendation algorithms was evaluated in terms of Precision, Recall, F score, Mean Average Precision, Normalized Discounted Cumulative Gain, Area Under the ROC Curve, Novelty, and Coverage. |
