Fast Recommendations With the M-Distance
Memory-based recommender systems with m users and n items typically require O(mn) space to store the rating information. In item-based collaborative filtering (CF) algorithms, the feature vector of each item has length m,and it takes O(m) time to compute the similarity between two items using the Pe...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2016-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/7452337/ |
| _version_ | 1818876708236820480 |
|---|---|
| author | Mei Zheng Fan Min Heng-Ru Zhang Wen-Bin Chen |
| author_facet | Mei Zheng Fan Min Heng-Ru Zhang Wen-Bin Chen |
| author_sort | Mei Zheng |
| collection | DOAJ |
| description | Memory-based recommender systems with m users and n items typically require O(mn) space to store the rating information. In item-based collaborative filtering (CF) algorithms, the feature vector of each item has length m,and it takes O(m) time to compute the similarity between two items using the Pearson or cosine distances. In this paper, we propose an efficient CF algorithm based on a new measure called the M-distance, which is defined as the difference between the average ratings of two items. In the initialization stage, we compute the average ratings of items and store them in two vectors, which requires O(m) space. Scanning the rating dataset then takes O(mn) time. In the online prediction stage, a threshold δ is employed to identify similar items. To predictp ratings, our algorithm requires O(np) time compared with the O(mnp) time of the cosine-based kNN algorithm. Experiments are undertaken on four well-known datasets, and the proposed M-distance is compared with the cosine-based kNN, Pearson-based kNN, and Slope One methods. Our results show that the new algorithm is significantly faster than the conventional techniques, especially for large datasets, and that its prediction ability is no worse in terms of the mean absolute error and root mean square error. |
| first_indexed | 2024-12-19T13:46:41Z |
| format | Article |
| id | doaj.art-ce88d994cf304142a1127fff33cdd441 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-19T13:46:41Z |
| publishDate | 2016-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-ce88d994cf304142a1127fff33cdd4412022-12-21T20:18:50ZengIEEEIEEE Access2169-35362016-01-0141464146810.1109/ACCESS.2016.25491827452337Fast Recommendations With the M-DistanceMei Zheng0Fan Min1Heng-Ru Zhang2Wen-Bin Chen3School of Computer Science, Southwest Petroleum University, Chengdu, ChinaSchool of Computer Science, Southwest Petroleum University, Chengdu, ChinaSchool of Computer Science, Southwest Petroleum University, Chengdu, ChinaSchool of Computer Science, Southwest Petroleum University, Chengdu, ChinaMemory-based recommender systems with m users and n items typically require O(mn) space to store the rating information. In item-based collaborative filtering (CF) algorithms, the feature vector of each item has length m,and it takes O(m) time to compute the similarity between two items using the Pearson or cosine distances. In this paper, we propose an efficient CF algorithm based on a new measure called the M-distance, which is defined as the difference between the average ratings of two items. In the initialization stage, we compute the average ratings of items and store them in two vectors, which requires O(m) space. Scanning the rating dataset then takes O(mn) time. In the online prediction stage, a threshold δ is employed to identify similar items. To predictp ratings, our algorithm requires O(np) time compared with the O(mnp) time of the cosine-based kNN algorithm. Experiments are undertaken on four well-known datasets, and the proposed M-distance is compared with the cosine-based kNN, Pearson-based kNN, and Slope One methods. Our results show that the new algorithm is significantly faster than the conventional techniques, especially for large datasets, and that its prediction ability is no worse in terms of the mean absolute error and root mean square error.https://ieeexplore.ieee.org/document/7452337/Computational complexitydistance measureneighborhoodrecommender systems |
| spellingShingle | Mei Zheng Fan Min Heng-Ru Zhang Wen-Bin Chen Fast Recommendations With the M-Distance IEEE Access Computational complexity distance measure neighborhood recommender systems |
| title | Fast Recommendations With the M-Distance |
| title_full | Fast Recommendations With the M-Distance |
| title_fullStr | Fast Recommendations With the M-Distance |
| title_full_unstemmed | Fast Recommendations With the M-Distance |
| title_short | Fast Recommendations With the M-Distance |
| title_sort | fast recommendations with the m distance |
| topic | Computational complexity distance measure neighborhood recommender systems |
| url | https://ieeexplore.ieee.org/document/7452337/ |
| work_keys_str_mv | AT meizheng fastrecommendationswiththemdistance AT fanmin fastrecommendationswiththemdistance AT hengruzhang fastrecommendationswiththemdistance AT wenbinchen fastrecommendationswiththemdistance |