Knowledge graph for recommendation system: Enhanced relation reliability and prediction probability (ERRaPP)

With the current explosion of information, the end-users find it challenging to filter this information. Recommendation systems present solutions to filter and prioritize the information to overcome the problem of information overloading. However, one of the main challenges associated with RS is accuracy. A knowledge graph (KG) is one solution to improve the recommendation system’s performance. The existing solutions do not consider the side information and semantic relationship from the knowledge graph, which results in the problem of accuracy of the recommendation and the processing time. Our proposed solution aims to increase the accuracy and decrease the processing time by exploring semantic relations between entities and considering the importance of relationships. The proposed system consists of a collaborative knowledge graph (GCN) with Enhanced Relation Reliability and Prediction Probability (ERRaPP) algorithm to enhance the recommendation accuracy and minimize the processing time. This algorithm includes the importance of relation specialized in an entity to get more reliable paths. It also has an attention mechanism with a sigmoid function to replace the inner product between entities embedding to improve the prediction. The results are obtained for different model stages (training, evaluation, test) for 4 other datasets (Book-Crossing, MovieLens-20 M, MovieLens-1 M and Last.FM). The results show that the proposed solution achieves better recommendation accuracy with less processing time for all three stages and 4 datasets. The proposed solution provides the recommendation accuracy of 0.705 against the current accuracy of 0.665 on average for the Book-Crossing dataset and a processing time of 7.884 seconds against the current processing time of 12 seconds on average for the testing stage. The proposed solution focuses on enhancing the overall accuracy and reducing the processing time of the knowledge graph-based recommendation system by using the ERRaPP algorithm. Finally, the solution with enhanced relation reliability and score prediction improves the recommendation accuracy by considering semantic relations between entities.