# Addressing the Cold-Start Problem in Recommender Systems Based on Frequent Patterns

^{*}

## Abstract

**:**

## 1. Introduction

- It proposes a novel hybrid approach, which combines clustering, discriminative itemset mining and the user’s/product’s context.
- It can handle both the sparsity and the cold-start recommendation problems.
- It can handle the out-of-matrix prediction (pure) cold-start recommendation problem, where users have no rating history. In most of similar approaches, several initial ratings are needed.
- It does not need any interview process to ask the user to fill in some extra information (e.g., extended profile data, answer questions and rate items), before using the system for the first time [5].
- It does not include many modules that are interrelated with multiple dependencies.

## 2. Literature Review

## 3. Proposed Methodology

Algorithm1: Recommendation Algorithm based on Discriminant Frequent Itemsets |

Step 1: InputThe set of users $U=\left\{{u}_{1},\dots ,{u}_{nofu},\right\}$, where $nofu$ is the number of users The set of products $P=\left\{{p}_{1},\dots ,{p}_{nofp}\right\}$, where $nofp$ is the number of products The $\left(nofu\times c\right)$ user matrix $UM=\left\{\left({u}_{i},{u}_{i1},\dots ,{u}_{ic}\right)|{u}_{i1},\dots ,{u}_{ic}\mathrm{are}\text{}\mathrm{the}\text{}\mathrm{characteristics}\text{}\mathrm{of}\text{}\mathrm{user}\text{}{u}_{i}\in U\right\}$ The $\left(nofu\times nofp\right)$ user–product matrix $UP=\left\{\left({u}_{i},{p}_{1},\dots ,{p}_{nofp}\right)|{u}_{i}\in U,{p}_{1},\dots ,{p}_{nofp}\in P\right\}$ Step 2: Extract $C=\left\{{c}_{1},\dots ,{c}_{n}\right\}$ clusters of usersStep 3:For each cluster ${c}_{i}\in C$ do Form the $\left|{c}_{i}\right|\times nofp$ sub user–product binary matrix $U{P}_{{c}_{i}}=\left\{\left({u}_{i},{p}_{1},\dots ,{p}_{nofp}\right)|\left({u}_{i},{p}_{1},\dots ,{p}_{nofp}\right)\in UP,{u}_{i}\in {c}_{i}\right\}$ Extract frequent itemsets $FI{S}_{{c}_{i}}=\left\{fi{s}_{{c}_{i}1},\dots ,fi{s}_{{c}_{i}m}\right\}$ of $U{P}_{{c}_{i}}$ End For Step 4:For each cluster ${c}_{i}\in C$ do Form the $\left|{c}_{i}\right|\times nofp$ sub user–product binary matrix $U{P}_{{c}_{i}}=\left\{\left({u}_{i},{p}_{1},\dots ,{p}_{nofp}\right)|\left({u}_{i},{p}_{1},\dots ,{p}_{nofp}\right)\in UP,{u}_{i}\in {c}_{i}\right\}$ Extract discriminant frequent itemsets $DFI{S}_{{c}_{i}}=\left\{dfi{s}_{{c}_{i}1},\dots ,dfi{s}_{{c}_{i}q}\right\}$ of $U{P}_{{c}_{i}}$, where $DFI{S}_{{c}_{i}}\subseteq FI{S}_{{c}_{i}}$ End For Step 5:For each ${u}_{i}\in U$ do If $\left({u}_{i},{p}_{1},\dots ,{p}_{nofp}\right)\in UP$, ${p}_{1}={p}_{2=}\dots ={p}_{nofp}=0$ * cold user Then For each $dfi{s}_{{c}_{i}k}\in DFI{S}_{{c}_{i}}\wedge {u}_{i}\in {c}_{i}$ do ${p}_{s}=1,s\in dfi{s}_{{c}_{i}k}$ End For Else * sparsity For each $dfi{s}_{{c}_{i}k}\in DFI{S}_{{c}_{i}}\wedge {u}_{i}\in {c}_{i}$ do If $\left|\left\{{p}_{s}|s\in dfi{s}_{{c}_{i}k}\wedge {p}_{s}=1\right\}\right|ct$ If $s\in dfi{s}_{{c}_{i}k}\wedge {p}_{s}=NULL$ Then ${p}_{s}=1$ End If End If End For End If End For |

#### 3.1. Input Matrices

#### 3.2. Clustering Procedure

#### 3.3. Extraction of Frequent Itemsets

#### 3.4. Extraction of Discriminant Itemsets

#### 3.5. Recommendation

#### 3.6. An Illustrative Example

## 4. Empirical Results

- To increase the density of the initial matrix by recommending more movies to users who have a small rating history (in-matrix prediction);
- To recommend movies to new users with no rating history at all, based on the rating history of users with similar characteristics (out-of-matrix prediction).

#### Comparison to Similar Methods

## 5. Discussion

## 6. Conclusions

#### 6.1. Limitations

#### 6.2. Future Work

## Author Contributions

## Funding

## Data Availability Statement

## Acknowledgments

## Conflicts of Interest

## References

- Feng, X.; Zhang, H.; Ren, Y.; Shang, P.; Zhu, Y.; Liang, Y.; Guan, R.; Xu, D. The deep learning–based recommender system “Pubmender” for choosing a biomedical publication venue: Development and validation study. J. Med. Internet Res.
**2019**, 21, e12957. [Google Scholar] [CrossRef] [PubMed] - Montaner, M.; López, B.; de la Rosa, J.L. A Taxonomy of Recommender Agents on the Internet. Artif. Intell. Rev.
**2003**, 19, 285–330. [Google Scholar] [CrossRef] - Sidana, S.; Trofimov, M.; Horodnytskyi, O.; Laclau, C.; Maximov, Y.; Amini, M.-R. User preference and embedding learning with implicit feedback for recommender systems. Data Min. Knowl. Discov.
**2021**, 35, 568–592. [Google Scholar] [CrossRef] - Lin, J.; Sugiyama, K.; Kan, M.-Y.; Chua, T.-S. Addressing cold-start in app recommendation: Latent user models constructed from twitter followers. In Proceedings of the 36th International ACM SIGIR Conference on Research and Development in Information Retrieval, Association for Computing Machinery, Dublin, Ireland, 28 July–1 August 2013; pp. 283–292. [Google Scholar] [CrossRef]
- Nadimi-Shahraki, M.-H.; Bahadorpour, M. Cold-start problem in collaborative recommender systems: Efficient methods based on ask-to-rate technique. J. Comput. Inf. Technol.
**2014**, 22, 105–113. [Google Scholar] [CrossRef] - Seyfi, M.; Nayak, R.; Xu, Y.; Geva, S. DISSparse: Efficient mining of discriminative itemsets. J. Inf. Knowl. Manag.
**2022**, 21, 2250009. [Google Scholar] [CrossRef] - Cheng, H.; Yan, X.; Han, J.; Yu, P.S. Direct Discriminative Pattern Mining for Effective Classification. In Proceedings of the 2008 IEEE 24th International Conference on Data Engineering, Cancun, Mexico, 7–12 April 2008. [Google Scholar] [CrossRef]
- de Carvalho, L.C.; Rodrigues, F.; Oliveira, P. A Hybrid Recommendation Algorithm to Address the Cold Start Problem. In Hybrid Intelligent Systems; Springer International Publishing: Cham, Switzerland, 2020. [Google Scholar]
- Yanxiang, L.; Deke, G.; Fei, C.; Honghui, C. User-based Clustering with Top-N Recommendation on Cold-Start Problem. In Proceedings of the 2013 Third International Conference on Intelligent System Design and Engineering Applications, Hong Kong, China, 16–18 January 2013. [Google Scholar] [CrossRef]
- Parambath, S.A.P.; Chawla, S. Simple and effective neural-free soft-cluster embeddings for item cold-start recommendations. Data Min. Knowl. Discov.
**2020**, 34, 1560–1588. [Google Scholar] [CrossRef] - Peng, F.; Lu, J.; Wang, Y.; Xu, R.Y.-D.; Ma, C.; Yang, J. N-dimensional Markov random field prior for cold-start recommendation. Neurocomputing
**2016**, 191, 187–199. [Google Scholar] [CrossRef] - Park, S.-T.; Chu, W. Pairwise preference regression for cold-start recommendation. In Proceedings of the Third ACM Conference on Recommender Systems, Association for Computing Machinery, New York, NY, USA, 22–25 October 2009; pp. 21–28. [Google Scholar] [CrossRef]
- Bobadilla, J.; Ortega, F.; Hernando, A.; Bernal, J. A collaborative filtering approach to mitigate the new user cold start problem. Knowl.-Based Syst.
**2012**, 26, 225–238. [Google Scholar] [CrossRef][Green Version] - Huang, C.; Yin, J. Effective association clusters filtering to cold-start recommendations. In Proceedings of the 2010 Seventh International Conference on Fuzzy Systems and Knowledge Discovery, Yantai, China, 10–12 August 2010. [Google Scholar] [CrossRef]
- Sedhain, S.; Sanner, S.; Braziunas, D.; Xie, L.; Christensen, J. Social collaborative filtering for cold-start recommendations. In Proceedings of the 8th ACM Conference on Recommender Systems, Association for Computing Machinery, Foster City, Silicon Valley, CA, USA, 6–10 October 2014; pp. 345–348. [Google Scholar] [CrossRef][Green Version]
- Lika, B.; Kolomvatsos, K.; Hadjiefthymiades, S. Facing the cold start problem in recommender systems. Expert Syst. Appl.
**2014**, 41, 2065–2073. [Google Scholar] [CrossRef] - Wei, J.; He, J.; Chen, K.; Zhou, Y.; Tang, Z. Collaborative filtering and deep learning based recommendation system for cold start items. Expert Syst. Appl.
**2017**, 69, 29–39. [Google Scholar] [CrossRef][Green Version] - Zhang, D.; Hsu, C.-H.; Chen, M.; Chen, Q.; Xiong, N.; Lloret, J. Cold-start recommendation using bi-clustering and fusion for large-scale social recommender systems. IEEE Trans. Emerg. Top. Comput.
**2013**, 2, 239–250. [Google Scholar] [CrossRef] - Kabbur, S.; Ning, X.; Karypis, G. FISM: Factored item similarity models for top-N recommender systems. In Proceedings of the 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Association for Computing Machinery, Chicago, IL, USA, 11–14 August 2013; pp. 659–667. [Google Scholar] [CrossRef]
- Li, J.; Jing, M.; Lu, K.; Zhu, L.; Yang, Y.; Huang, Z. From Zero-Shot Learning to Cold-Start Recommendation. In Proceedings of the Thirty-Third AAAI Conference on Artificial Intelligence, Honolulu, Hawaii, USA, 27 January–1 February 2019; Volume 33, pp. 4189–4196. [Google Scholar] [CrossRef][Green Version]
- Kawai, M.; Sato, H.; Shiohama, T. Topic model-based recommender systems and their applications to cold-start problems. Expert Syst. Appl.
**2022**, 202, 117129. [Google Scholar] [CrossRef] - Pereira, A.L.V.; Hruschka, E.R. Simultaneous co-clustering and learning to address the cold start problem in recommender systems. Knowl.-Based Syst.
**2015**, 82, 11–19. [Google Scholar] [CrossRef] - Victor, P.; Cornelis, C.; Teredesai, A.M.; Cock, M.D. Whom should I trust? the impact of key figures on cold start recommendations. In Proceedings of the 2008 ACM Symposium on Applied Computing, Association for Computing Machinery, Fortaleza, Ceara, Brazil, 16–20 March 2008; pp. 2014–2018. [Google Scholar] [CrossRef]
- Camacho, L.A.G.; Alves-Souza, S.N. Social network data to alleviate cold-start in recommender system: A systematic review. Inf. Process. Manag.
**2018**, 54, 529–544. [Google Scholar] [CrossRef] - Herce-Zelaya, J.; Porcel, C.; Bernabé-Moreno, J.; Tejeda-Lorente, A.; Herrera-Viedma, E. New technique to alleviate the cold start problem in recommender systems using information from social media and random decision forests. Inf. Sci.
**2020**, 536, 156–170. [Google Scholar] [CrossRef] - Zhao, W.X.; Li, S.; He, Y.; Chang, E.Y.; Wen, J.R.; Li, X. Connecting Social Media to E-Commerce: Cold-Start Product Recommendation Using Microblogging Information. IEEE Trans. Knowl. Data Eng.
**2016**, 28, 1147–1159. [Google Scholar] [CrossRef][Green Version] - Shah, A.A.; Venkateshwara, H. Sparsity Regularization for Cold-Start Recommendation. arXiv
**2022**. [Google Scholar] [CrossRef] - Rashid, A.M.; Karypis, G.; Riedl, J. Learning preferences of new users in recommender systems: An information theoretic approach. ACM SIGKDD Explor. Newsl.
**2008**, 10, 90–100. [Google Scholar] [CrossRef] - Goldberg, K.; Roeder, T.; Gupta, D.; Perkins, C. Eigentaste: A Constant Time Collaborative Filtering Algorithm. Inf. Retr.
**2001**, 4, 133–151. [Google Scholar] [CrossRef] - Sarwar, B.; Karypis, G.; Konstan, J.; Riedl, J. Application of Dimensionality Reduction in Recommender System—A Case Study; Minnesota University Minneapolis Department of Computer Science: Minneapolis, MN, USA, 2000. [Google Scholar]
- Huang, Z.; Chen, H.; Zeng, D. Applying associative retrieval techniques to alleviate the sparsity problem in collaborative filtering. J. ACM Trans. Inf. Syst.
**2004**, 22, 116–142. [Google Scholar] [CrossRef][Green Version] - Qing, L.; Byeong Man, K. Clustering approach for hybrid recommender system. In Proceedings of the IEEE/WIC International Conference on Web Intelligence (WI 2003), Halifax, NS, Canada, 13–17 October 2003. [Google Scholar] [CrossRef]
- Ma, X.; Lu, H.; Gan, Z.; Zhao, Q. An exploration of improving prediction accuracy by constructing a multi-type clustering based recommendation framework. Neurocomputing
**2016**, 191, 388–397. [Google Scholar] [CrossRef] - Shaw, G.; Xu, Y.; Geva, S. Using Association Rules to Solve the Cold-Start Problem in Recommender Systems. In Advances in Knowledge Discovery and Data Mining; Springer: Berlin/Heidelberg, Germany, 2010. [Google Scholar] [CrossRef][Green Version]
- Leung, C.W.-K.; Chan, S.C.-F.; Chung, F.-L. An empirical study of a cross-level association rule mining approach to cold-start recommendations. Knowl.-Based Syst.
**2008**, 21, 515–529. [Google Scholar] [CrossRef] - Lin, W.; Alvarez, S.A.; Ruiz, C. Efficient Adaptive-Support Association Rule Mining for Recommender Systems. Data Min. Knowl. Discov.
**2002**, 6, 83–105. [Google Scholar] [CrossRef] - Viktoratos, I.; Tsadiras, A.; Bassiliades, N. Combining community-based knowledge with association rule mining to alleviate the cold start problem in context-aware recommender systems. Expert Syst. Appl.
**2018**, 101, 78–90. [Google Scholar] [CrossRef] - Mohamed, M.H.; Khafagy, M.H.; Elbeh, H.; Abdalla, A.M. Technology. Sparsity and cold start recommendation system challenges solved by hybrid feedback. Int. J. Eng. Res. Technol.
**2019**, 12, 2734–2741. [Google Scholar] - Hernando, A.; Bobadilla, J.; Ortega, F.; Gutiérrez, A. A probabilistic model for recommending to new cold-start non-registered users. Inf. Sci.
**2017**, 376, 216–232. [Google Scholar] [CrossRef] - Sobhanam, H.; Mariappan, A.K. Addressing cold start problem in recommender systems using association rules and clustering technique. In Proceedings of the 2013 International Conference on Computer Communication and Informatics, Coimbatore, India, 4–6 January 2013. [Google Scholar] [CrossRef]
- Rahman, M.M.; Islam, M.S.; Richi, R.R.; Chakraborty, A. Course Recommendation System for Students Using K-Means and Association Rule Mining. In Proceedings of the 2022 International Symposium on Multidisciplinary Studies and Innovative Technologies (ISMSIT), Ankara, Turkey, 20–22 October 2022. [Google Scholar] [CrossRef]
- Bellini, P.; Palesi, L.A.I.; Nesi, P.; Pantaleo, G. Multi Clustering Recommendation System for Fashion Retail. Multimed. Tools Appl.
**2023**, 82, 9989–10016. [Google Scholar] [CrossRef] - Hasan, S.N.; Khatwal, R. Cold Start Problem in Recommendation System: A Solution Model Based on Clustering and Association Rule Techniques. In Proceedings of the 2022 5th International Conference on Multimedia, Signal Processing and Communication Technologies (IMPACT), Aligarh, India, 26–27 November 2022. [Google Scholar] [CrossRef]
- Kannout, E.; Grodzki, M.; Grzegorowski, M. Utilizing Frequent Pattern Mining for Solving Cold-Start Problem in Recommender Systems. In Proceedings of the 2022 17th Conference on Computer Science and Intelligence Systems (FedCSIS), Sofia, Bulgaria, 4–7 September 2022. [Google Scholar] [CrossRef]
- Seyfi, M.; Nayak, R.; Xu, Y.; Geva, S. Efficient mining of discriminative itemsets. In Proceedings of the International Conference on Web Intelligence, Association for Computing Machinery, Leipzig, Germany, 23–26 August 2017; pp. 451–459. [Google Scholar] [CrossRef][Green Version]
- Harper, F.M.; Konstan, J.A. The MovieLens Datasets: History and Context. J. ACM Trans. Interact. Intell. Syst.
**2015**, 5, 1–19. [Google Scholar] [CrossRef] - Huang, Z. Extensions to the k-Means Algorithm for Clustering Large Data Sets with Categorical Values. Data Min. Knowl. Discov.
**1998**, 2, 283–304. [Google Scholar] [CrossRef] - Muiño, D.P.; Borgelt, C. Frequent item set mining for sequential data: Synchrony in neuronal spike trains. Intell. Data Anal.
**2014**, 18, 997–1012. [Google Scholar] [CrossRef] - Wong, T.T.; Yeh, P.Y. Reliable Accuracy Estimates from k-Fold Cross Validation. IEEE Trans. Knowl. Data Eng.
**2020**, 32, 1586–1594. [Google Scholar] [CrossRef] - Wong, H.B.; Lim, G.H. Measures of Diagnostic Accuracy: Sensitivity, Specificity, PPV and NPV. Proc. Singap. Healthc.
**2011**, 20, 316–318. [Google Scholar] [CrossRef] - Parikh, R.; Mathai, A.; Parikh, S.; Chandra Sekhar, G.; Thomas, R. Understanding and using sensitivity, specificity and predictive values. Indian J. Ophthalmol.
**2008**, 56, 45–50. [Google Scholar] [CrossRef] - Akobeng, A.K. Understanding diagnostic tests 1: Sensitivity, specificity and predictive values. Acta Paediatr.
**2007**, 96, 338–341. [Google Scholar] [CrossRef] - Boutsinas, B.; Tasoulis, D.K.; Vrahatis, M.N. Estimating the number of clusters using a windowing technique. Pattern Recognit. Image Anal.
**2006**, 16, 143–154. [Google Scholar] [CrossRef] - Guo, G.; Zhang, J.; Yorke-Smith, N. Leveraging multiviews of trust and similarity to enhance clustering-based recommender systems. Knowl.-Based Syst.
**2015**, 74, 14–27. [Google Scholar] [CrossRef]

Age | Sex | Occupation | Country | Wage (In $) | |
---|---|---|---|---|---|

u_{1} | 25–35 | F | Pharmacist | Germany | 3000 |

u_{2} | 18–24 | M | Teacher | USA | 5000 |

u_{3} | 36–45 | F | Economist | France | 2500 |

u_{4} | 25–35 | F | Doctor | UK | 3000 |

u_{5} | 46–55 | M | Engineer | USA | 8000 |

u_{6} | 18–24 | M | Professor | USA | 6000 |

u_{7} | 25–35 | M | Lawyer | USA | 8000 |

u_{8} | 18–24 | F | Teacher | USA | 6000 |

u_{nofu-1} | Over 55 | F | Teacher | Canada | 6000 |

u_{nofu} | 46–55 | M | Assistant | China | 1500 |

p_{1} | p_{2} | p_{3} | p_{4} | p_{5} | p_{nofp-1} | p_{nofp} | ||
---|---|---|---|---|---|---|---|---|

u_{1} | 1 | 0 | 1 | Not updated | ||||

u_{2} | 1 | 1 | 1 | Updated (*cold user) | ||||

u_{3} | 1 | 1 | 1 | 1 | Updated | |||

u_{4} | 1 | 0 | 1 | 0 | 1 | Not updated | ||

u_{5} | 0 | 1 | 0 | 0 | ||||

u_{6} | 1 | 0 | 1 | Not updated | ||||

u_{7} | 1 | 1 | 1 | Not updated | ||||

u_{8} | 1 | 1 | 1 | Updated | ||||

u_{nofu-1} | 0 | 1 | ||||||

u_{nofu} | 0 | 1 | 0 | 1 |

Author | Precision | Recall | F1-Measure |
---|---|---|---|

Proposed method (out-of-matrix) | 0.895 | 0.463 | 0.61 |

Proposed method (in-matrix, 30–90%) | 0.903–0.900 | 0.444–0.445 | 0.595–0.595 |

de Carvalho et al. (2020) [8] | 0.035 | 0.121 | 0.052 |

Yanxiang, L. et al. (2013) [9] | 0.651 | - | - |

Peng, Lu et al. (2016) [11] | 0.786 | 0.287 | 0.367 |

Park and Chu (2009) [12] | 0.666 | 0.239 | 0.311 |

Bobadilla, Ortega [13] | 0.37–0.57 | 0.25–1 | - |

Huang and Yin (2010) [14] | - | - | 0.308 |

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |

© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Panteli, A.; Boutsinas, B. Addressing the Cold-Start Problem in Recommender Systems Based on Frequent Patterns. *Algorithms* **2023**, *16*, 182.
https://doi.org/10.3390/a16040182

**AMA Style**

Panteli A, Boutsinas B. Addressing the Cold-Start Problem in Recommender Systems Based on Frequent Patterns. *Algorithms*. 2023; 16(4):182.
https://doi.org/10.3390/a16040182

**Chicago/Turabian Style**

Panteli, Antiopi, and Basilis Boutsinas. 2023. "Addressing the Cold-Start Problem in Recommender Systems Based on Frequent Patterns" *Algorithms* 16, no. 4: 182.
https://doi.org/10.3390/a16040182