|
S. no | Publication | Technique used | Advantages | Disadvantages |
|
(1) | An application of fuzzy geographically clustering for solving the cold-start problem in recommender systems [39] | Data like user demographic information, their opinion, and social tags are used to determine the best neighbours for the new user | Analogous users are more accurately determined | Very less user’s demographic information is considered |
(2) | Semantics-aware recommender systems exploiting Linked Open Data and graph-based features [40] | LOD-based features, Random Forests, Naïve Bayes, and Logistic Regression | Accuracy of recommendation framework is improved | Similarity degree between users is not considered, and sparsity is ignored |
(3) | A recommender system based on collaborative filtering using ontology and dimensionality reduction techniques [41] | EM clustering for clustering and nonincremental SVD for dimensionality reduction, Ontology-based similarity | Solve two main drawbacks of recommender systems, sparsity and scalability, using dimensionality reduction and ontology techniques | User’s demographic or its browsing information has not been considered |
(4) | A new user similarity model to improve the accuracy of collaborative filtering [24] | Determining the analogous users using new similarity technique clustering algorithms, decision trees | Enhancement in the similarity degrees between users and no requirement for additional data | Needs consideration in choosing the optimal number of groups and the splitting criteria |
(5) | Addressing the new user cold-start problem in recommender systems using Ordered Weighted averaging operator [42] | (i) Optimistic exponential type of ordered weighted averaging (OWA) operator is applied (ii) Fusion of CF and demographic and fusion of CF and CBF classifiers have been used | Improvement in performance of the hybrid recommender system under “new user cold-start” problem | Missing values are not handled |
(6) | Proposed approach | (i) Ontology-based clustering for item (ii) LOD and social network graph features to build user’s profile | (i) Determination of similar user is more accurate (ii) Improvement in recommendation system’s performance for “pure new user cold-start” problem (iii) Missing values handled, and sparsity prediction is more accurate | (i) Scalability (ii) Filtering features based on domain |
|