Utility-Based Multi-Criteria Recommender Systems by Zheng
In this research activity, they introduced a utility-based multi-criteria recommendation algorithm. In this algorithm, they studied customer expectations by dissimilar learning to rank approaches. Their experimental outputs are depending on practical datasets. It demonstrates the usefulness of these approaches [3].
3.4.4.1 Experimental Dataset
In this research activity, they used two practical datasets where ratings are scaled between 1 and 5. The TripAdvisor data had used. In this dataset, it has more than 22,000 ratings provided by more than 1,500 clients with around 14,000 plus hotels. Every client rated at least 10 ratings. These ratings relate to multi-criteria ratings on seven criteria. Those criteria are cost-effective, convenience, quality of rooms, check-in, and cleanliness of the hotel and general standard of facility and specific business facilities. The Yahoo! Movies dataset was used here. There are more than 62,000 ratings given by more than 2000 clients on around 3,100 movies. Every client rates minimum 10 ratings. These ratings are related with multiple-criteria ratings on furrieries. Those critters are acting, direction, stories, and visual effects. They compared their utility-based models with some approaches. The approaches are MF, linear aggregation model (LAM), hybrid context model (HCM), and criteria chain model (CCM) [3].
They evaluated the efficiency of recommender form on the top 10 recommendations by using accuracy and NDCG to calculate the efficiency. To calculate the utility scores, they used three measures. By applying Pearson correlation, they get little improved results rather than applying cosine similarity. They found that Euclidean distance was the bad choice. They represented the best outcome by using Pearson correlation [3].
3.4.4.2 Experimental Result
As we can see in Figure 3.5, it represents the results the experiment. FMM becomes the best performing baseline method for the TripAdvisor data, but LAM and CCM beat MF by 1%. Here, HCM performs even lower than the MF approach. Through applying the utility-based method, the UBM by applying the listwise ranking can perform well the FMM method. If they use the pointwise and pairwise ranking optimizations, then the other UBM models will fail to beat FMM. From Yahoo! Movies dataset, all methods can perform the MF method that does not consider multi-criteria ratings. To be to detail, the UBM using listwise ranking can upgrade NDGC and precision by 6.3% and 5.4% in the TripAdvisor data, and 4.1% and 8% in FMM in comparison with Yahoo! Movies data [3].
Figure 3.5 Experimental result.
3.4.5 Multi-Criteria Clustering Approach by Wasid and Ali
In this research activity, they suggested a clustering method to use multiple-criteria rating into conventional recommendation system successfully. To generate more on the mark recommendations, they evaluate the intra-cluster client matches by applying Mahalanobis distance approach. Then, they collated their method with the conventional CF [2].
Now, we will take a look on their experimental evaluation and result for its efficiency.
3.4.5.1 Experimental Evaluation
To implement this proposed approach, they have used Yahoo! Movies dataset. This dataset consists of 62,156 rating provides by 6,078 users on 976 movies. To make it simple, they have extracted those clients that have rated to minimum 20 movies. This condition satisfies 484 users and 945 movies, and they have total 19,050 ratings. Then, every client’s rating is splitted arbitrarily into training and testing set. They took 70% of data for training purpose and remaining 30% of data for testing purpose. Then, they calculate the distance between clients successfully. They selected top 30 most equivalent users for the neighborhood set formation. To evaluate this proposed approach, they used the most popular Mean Absolute Error (MAE) performance matrix. MAE is very popular because of its simplicity and accuracy as we have seen before. It matches the goal of the experiment. The mean absolute error estimates the derivation of actual and predicted client ratings [2].
3.4.5.2 Result and Analysis
The dataset which they have used contain both single-criteria and multiple-criteria user provided ratings. Table 3.4 represents the difference between their way with the conventional approach for both non-clustering and clustering environment. Here, they have compared their experimental result with Pearson collaborative recommender (PCRS) approach. Table 3.4 conveys the result between PCRS with their proposed Mahalanobis distance recommendation scheme (MDRS). It is very transparent that their method MDRS performed much superior than traditional PCRS. These results are based on mean absolute error method. This proposed MDRS approach works better for both on clustering and clustering environments which is shown in Figure 3.6 [2].
They also have shown the graphical implementation of this table which is in both non-clustering and clustering environments. The Mahalanobis distance–based method gives better result than the Pearson collaborative recommender approach. If the mean absolute error values are lower, then it means it is a better result. By that result, we can also see that the non-clustering–based technique always have less performance than clustering approaches. So, the clustering approach is better [2].
Table 3.4 Comparison among clustering and non-clustering approach.
| Approach | MAE (non-clustering) | MAE (clustering) |
|---|---|---|
| PCRS | 2.4577 | 2.2734 |
| MDRS | 2.3094 | 2.1751 |
Figure 3.6 Result.
They have consolidated the multiple-criteria ratings into the conventional CF-based recommender system using K-means algorithm. Their method treats the third dimensional as multi-criteria, the clustering parameters as the clustering parameter of the clients, to handle the dimensionality. Their approach depends on thinking like each user has unique opinion and criteria. Therefore, to compare each user, the most important concern of this work is to find out clients’ segments with alike client. Mahalanobis distance method is used here to create most exact neighbors for every client within the cluster. In the result, we can clearly see that this technique is more effective and accurate than the traditional approach [2].
3.5 Advantages of Multi-Criteria Recommender System
Multi-criteria
