Document Type : Research Paper

Authors

1 Department of Computer Engineering, Mashhad Branch, Islamic Azad University, Mashhad, Iran

2 Assistant Professor - Department of Computer Engineering - Islamic Azad University of Mashhad

Abstract

Among the various applications of recommender systems, their use in estimating and suggesting points of interest (POIs) for tourists has expanded significantly in recent years. A common approach to identify user interests is to use collaborative filtering (CF) technique. However, the accuracy and efficiency of CF can be improved by applying different parameters and complementary approaches. In this paper, a new solution for promoting POI offers to tourists is presented, which uses a five-dimensional time model including the dimensions of day and night hours, days of the week, days of the month, months of the year, and occasions, and by calculating the Euclidean distance between the time of recommendation and the time of previous experiences of the active user and his similar users identifies and suggests suitable venues. The proposed solution also uses the trust parameter to increase the accuracy of POI suggestion. To improve the accuracy of trust evaluation, a new criterion based on a similarity tree structure between contexts is introduced. The results of experiments conducted on three well-known datasets show that the proposed model outperforms the state-of-the-art methods in term of efficiency and accuracy.

Introduction

Recommender systems estimate the interests and preferences of each user and suggest items and services to them, thus helping users to make a quick and favorable choice. Among the various applications of these systems, their use in estimating and suggesting points of interest (POIs) for tourists has expanded significantly in recent years. A common approach to identifying user interests is to use the collaborative filtering (CF) technique. However, the accuracy and efficiency of CF can be improved by applying different parameters and complementary approaches. In this research, a new solution for promoting POI offers to tourists is presented, which uses a five-dimensional time model including the dimensions of day and night hours, days of the week, days of the month, months of the year, and occasions, and by calculating the Euclidean distance between the time of recommendation and the time of previous experiences of the active user and his similar users identifies and suggests suitable venues. The proposed solution also uses the trust parameter to increase the accuracy of POI suggestions. To improve the accuracy of trust evaluation, a new criterion based on a similarity tree structure between contexts is introduced. The results of experiments conducted on three well-known datasets show that the proposed model outperforms the state-of-the-art methods in terms of efficiency and accuracy.
Research Question(s)
The main question of the current research is whether considering the different dimensions of the time parameter in touristic place recommendation systems, along with the trust parameter between users, can significantly increase the accuracy of the system's recommendations.

Literature Review

Various research works have been done with the aim of investigating the impact of social relations, time, place, and context on the efficiency of recommender systems. Savage et al. (2012) presented a location-based recommendation algorithm to improve the accuracy of recommended items based on learning according to the analysis of the user's profile in social networks and his location. Bedi (2020) presents a cross-domain approach for group recommender systems. In this approach, the suggestions provided by reliable and well-known users in the group improve the acceptance of recommendations compared to the suggestions of other people in the group. The system is designed in such a way that it takes into account the information of different sub-domains of the tourism domain. El Yebdri et al. (2021) proposed a context-aware trust-based post-refining approach to overcome the problems of data sparsity and cold start in recommender systems. This approach uses the average relative difference between fields. The authors first calculate the average score for each contextual condition and balance all evaluations based on the contextual condition of each tuple.
On the other hand, in the new era, which is known as the post-Fordism era, the supply and demand patterns in the field of tourism have faced significant changes which should be considered in the strategies of tourism service providers (Liasidou, 2022).

Methodology

According to the main goal of the current research, which is to increase the accuracy of systems recommending points of interest to tourists by introducing the influence of time dimensions, the research includes several stages. At first, a new approach to represent time in terms of hours, days of the week, days of the month, months of the year, and occasions is presented. Then, this time representation approach is combined with a trust computing model and a context-aware collaborative filtering technique to build a computational model for extracting and recommending points of interest to tourists. In the next stage of the research, to evaluate the effectiveness of the proposed model in increasing the accuracy of the system's recommendations and the level of user satisfaction, the presented model was implemented on several datasets in the field of tourism.

Results

In this research, several experiments have been performed to evaluate the performance of the proposed model. Experiments have been conducted on three real public datasets in the field of tourism, namely Yelp, Foursquare, and Gowalla. Some common criteria have been used to evaluate the proposed approach and compare its accuracy and efficiency with the existing methods:
Precision: the ratio of the number of relevant items in the list of top N items to N.
Recall: the ratio of the number of relevant items in the list of N suggested items to the total number of relevant items.
The results of the proposed model in this research were compared with three existing similar research works, including USSTC, MEAP-T, and LOCABAL+, which were respectively conducted by Kefalas and Manolopoulos (2017), Ying et al. (2019) and Ardisono and Mauro (2020).
The first experiment was performed to analyze the sensitivity of the proposed model in terms of precision and recall criteria to changes in the value of N for the top N item suggestion. As expected, the precision decreases as the number of suggested venues increases. On the other hand, as N increases, the recall increases as well.
Subsequent experiments were conducted to measure and compare the accuracy and recall criteria and showed that the proposed method provides the best accuracy values ​​for different datasets compared to existing research works.

Discussion

The results of the evaluations based on three well-known data sets in the field of tourism-related recommendation systems showed that the application of these parameters significantly improves the accuracy of the system's recommendations, and therefore they should be considered more seriously in the recommender systems.
It is worth noting that if the absolute values ​​of the results are evaluated, the improvement of the results in the proposed model may seem insignificant compared to the previous models. But if the relative amount of the improvement of the results is considered, for example, in the case of the Yelp dataset, it can be seen that the proposed model has provided a significant increase in precision and recall criteria even compared to its closest competitor, LOCABAL+.

Conclusion

In this research, with the aim of improving the performance of systems recommending venues to tourists, a model based on the estimation of trust between people was presented and evaluated. In the proposed model, the level of trust between two users in choosing their favorite places to visit is estimated based on the similarity level of their feedback and previous comments. In this regard, in the proposed model, parameters of time, location of the tourist, and classification of POIs were considered. In the proposed solution, a five-dimensional time model is used, and suitable venues are identified and suggested by calculating the distance between the time of recommendation and the time of previous experiences of similar tourists. The improvement of the results of this approach, which is evident in the results of this research, shows that systems that apply different dimensions of time in offering places to tourists, provide more accurate recommendations and a higher level of satisfaction for users.
Keywords: Tourism Recommender System, POI, Location-Based Services, Time-Aware Recommendation, Trust-Based Recommendation, Context-Aware Recommendation.
 
 
 

Keywords

Main Subjects

مودودی ارخودی، مهدی و فردوسی، سجاد (1399). سنجش الگوی رفتاری عرضه و تقاضای گردشگری مبتنی بر تحولات نظام سرمایه‏داری، مطالعه موردی: شهر کرج و روستاهای پیرامونی. فصلنامه مطالعات شهری، 9(36)، 100-85. doi: 10.34785/J011.2021.869
شایان، حمید.، مودودی ارخودی، مهدی، فردوسی، سجاد. (1401). رفتارشناسی عرضه و تقاضای گردشگری مبتنی بر تحولات نظام سرمایه‏داری. دوفصلنامه مطالعات اجتماعی گردشگری، 10(19)، 1-28. doi:  10.52547/journalitor.36268.10.19.0
سقایی، مهدی و پاپلی یزدی، محمدحسین (1393). گردشگری: ماهیت و مفاهیم. تهران: انتشارات سمت.
 
References
Ardissono, L., & Mauro, N. (2020). A compositional model of multi-faceted trust for personalized item recommendation. Expert Systems with Applications, 140, 112880. https://doi.org/10.1016/j.eswa.2019 .112880
Bedi, P. (2020). Combining trust and reputation as user influence in cross domain group recommender system (CDGRS). Journal of Intelligent & Fuzzy Systems, 38 (5), 6235-6246. https://doi.org/10.3233/JIFS-179705
Deshpande, M., & Karypis, G. (2004), Item-based top-n recommendation algorithms. ACM Transactions on Information Systems, 22, 143–177. https://doi.org/10.1145/3545796
El Yebdri, Z., Benslimane, S. M., Lahfa, F., Barhamgi, M., & Benslimane, D. (2021). Context-aware recommender system using trust network. Computing, 103, 1919–1937. https://doi.org/10.1007/s00607-020-00876-9
Hosseini, S., Yin, H., Zhou, X., Sadiq, S., Kangavari, M. R., & Cheung, N.-M. (2019). Leveraging multi-aspect time-related influence in location recommendation. World Wide Web, 22(3), 1001-1028. https://doi.org/10.1007/s11280-018-0573-2
Ineson, E., Yap, H., & Niţă, V. (2017). International Case Studies for Hospitality, Tourism and Event Management Students and Trainees. Technopress.
Kefalas, P., & Manolopoulos, Y. (2017). A time-aware spatio-textual recommender system. Expert Systems with Applications, 78, 396-406. https://doi.org/10.1016/j.eswa.2017.01.060
Khazaei, E., & Alimohammadi, A. (2018). An automatic user grouping model for a group recommender system in location-based social networks. ISPRS International Journal of Geo-Information, 7(67), 1-18. https://doi.org/10.3390/ijgi7020067
Liasidou, S. (2022). Reviewing the Content of European Countries’ Official Tourism Websites: A Neo/Post-Fordist Perspective. Tourism and Hospitality, 3(2), 380-398.  https://doi.org/10.3390/tourhosp3020025
Moradi, P., Ahmadian, S., & Akhlaghian, F. (2015). An effective trust-based recommendation method using a novel graph clustering algorithm. Physica A: Statistical mechanics and its applications, 436, 462-481. https://doi.org/10.1016/j.physa.2015.05.008
Musto, C., Gemmis, M., Semeraro, G., & Lops, P. (2017). A multi-criteria recommender system exploiting aspect-based sentiment analysis of users' reviews. Proceedings of the eleventh ACM conference on recommender systems. https://doi.org/10.1145/3109859.3109905
Nobahari, V., Jalali, M., & Mahdavi, S. J. S. (2019). ISoTrustSeq: a social recommender system based on implicit interest, trust and sequential behaviors of users using matrix factorization. Journal of Intelligent Information Systems, 52(2), 239-268. https://doi.org/10.1007/s10844-018-0513-8
Ozsoy, M. G., Polat, F., & Alhajj, R. (2016). Time preference aware dynamic recommendation enhanced with location, social network and temporal information. 2016 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), 909-916. https://doi.org/10.1109/ASONAM.2016.7752347
Rafailidis, D., & Nanopoulos, A. (2015). Modeling user preference dynamics and side information in recommender systems. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 46(6), 782-792. https://doi.org/10.1109/TSMC.2015.2460691
Richa, & Bedi, P. (2021). Trust and Distrust based Cross-domain Recommender System. Applied Artificial Intelligence, 35(4), 326-351. https://doi.org/10.1080/08839514.2021.1881297
Roy, F. (2020). A Comparative Analysis of Different Trust Metrics in User-User Trust-Based Recommender System, Preprint. https://doi.org/10.20944/preprints202011.0466.v1
Sani, N. S., & Tabriz, F. N. (2017). A new strategy in trust-based recommender system using k-means clustering. International Journal of Advanced Computer Science And Applications, 8(9), 152-156. https://doi.org/10.14569/IJACSA.2017.080922
Savage, N. S., Baranski, M., Chavez, N. E., & Höllerer, T. (2012). I’m feeling loco: A location-based context aware recommendation system. Advances in Location-Based Services, 37-54. https://doi.org/ 10.1007/978-3-642-24198-73
Tahmasbi, H., Jalali, M., & Shakeri, H. (2018). Modeling temporal dynamics of user preferences in movie recommendation. 2018 8th International Conference on Computer and Knowledge Engineering (ICCKE), 194-199. https://doi.org/10.1109/ICCKE.2018.8566316
Tuan, C.-C., Hung, C.-F., & Wu, Z.-H. (2017). Collaborative location recommendations with dynamic time periods. Pervasive and Mobile Computing, 35, 1-14. https://doi.org/10.1016/j.pmcj.2016.07.008
Wu, Z., & Palmer, M. (1994). Verbs semantics and lexical selection. Proceedings of the 32nd annual meeting on Association for Computational Linguistics. https://doi.org/10.3115/981732.981751
Ying, H., Wu, J., Xu, G., Liu, Y., Liang, T., Zhang, X., & Xiong, H. (2019). Time-aware metric embedding with asymmetric projection for successive POI recommendation. World Wide Web, 22(5), 2209-2224. https://doi.org/10.1007/s11280-018-0596-8
Zhang, J.-D., & Chow, C.-Y. (2015). TICRec: A probabilistic framework to utilize temporal influence correlations for time-aware location recommendations. IEEE Transactions on Services Computing, 9(4), 633-646. https://doi.org/10.1109/TSC.2015.2413783
Zhao, X., Ma, Z., & Zhang, Z. (2018). A novel recommendation system in location-based social networks using distributed ELM. Memetic computing, 10(3), 321-331. https://doi.org/10.1007/s12293-017-0227-4
Zheng, X.-L., Chen, C.-C., Hung, J.-L., He, W., Hong, F.-X., & Lin, Z. (2015). A hybrid trust-based recommender system for online communities of practice. IEEE Transactions on Learning Technologies, 8(4), 345-356. https://doi.org/10.1109/TLT.2015.2419262
Zhou, Y., Yang, G., Yan, B., Cai, Y., & Zhu, Z. (2022), Point-of-interest recommendation model considering strength of user relationship for location-based social networks. Expert Systems with Applications. 199, 117147. https://doi.org/10.1016/j.eswa.2022.117147
 
References [In Persian]
Arkhudi, M., & Ferdowsi, S. (2020), Assessing the behavioral pattern of tourism supply and demand based on the changes of the capitalist system (Case: Karaj city and surrounding villages). Motaleate Shahri, 9(36), 85-100. [In Persian].
Shayan, H., Arkhudi, M., & Ferdowsi, S. (2021), Behavior of tourism supply and demand based on the changes of the capitalist system. Social Studies in Tourism, 10(19), 1-28. [In Persian].
Papoli Yazdi, M.H., & Saghaei, M. (2004). Tourism: Nature and Concepts. Tehran: SAMT Publication. [In Persian].