Location recommendation by combining geographical, categorical, and social preferences with location popularity
Introduction
A large number of location-based social networks (LBSNs), such as Yelp and Foursquare, have attracted millions of users to share points of interest (POIs, i.e., locations) and social friendship via check-ins (Cheng, Yang, King & Lyu, 2016). Users would check in at locations such as restaurants, bars, home service stores, as well as museums, art galleries, and parks to comment about the services or share their experiences. By mining these data, user behaviors can be widely applied to recommend locations according to user-derived preferences (Bao, Zheng, Wilkie & Mokbel, 2015). As a crucial feature of LBSNs, location recommendation has received extensive attention from industry and academia. By mining users’ check-in records and social relationships, location recommendation is useful for suggesting new places for users to visit. It can be defined as recommending an unvisited location list to a user based on his/her check-in sequences, including visited locations and check-in time stamp (Zhao, King & Lyu, 2016). This function can not only increase the visibility of locations and attract more customers, but also improve user experience and customer satisfaction (Wu, Kao, Wu & Huang, 2015; Zhu, Chang, Luo & Li, 2014). One classic application is to free users from information overload dilemma and help them discover unvisited locations that satisfy their personalized preferences.
Location recommendation aims to predict users’ future movement by mining their check-in histories or consuming behaviors. In this process, massive volume of location data and high sparsity of user-visit data render significant challenges for improving the recommendation quality. For example, the Yelp dataset1 used in this study involves 393 cities with an average of 353 locations. Las Vegas owns the most venues, i.e., 28,670 in total. However, the average number of user check-ins is only 3.75, involving 1.48 cities. As shown in Fig. 1, more than half of the users (53%) only have one check-in record, and those who have less than five check-ins constituted 86.75% of the users. Most users (73.43%) have visited only one city, while only 1.44% of the users have visited more than five cities. Data sparsity limits the accuracy in inferring personalized user preferences (Liu, Wu & Liu, 2013), rendering it difficult to discover potential locations of interest from massive venues. Therefore, the crucial problem to be solved by location recommendation is to precisely model preferences from exiguous visit records.
Recent studies have applied collaborative filtering (CF), content-based methods, and hybrid methods to extract useful information from multiple types of data for modeling user preferences in location recommendation (Rehman, Khalid & Madani, 2016). CF infers the similarities between users and venues by identifying similar visiting patterns for users and venues from user behaviors. Some important properties, such as business category and geographical characteristics, are not considered in CF but are exploited in content-based methods. In modeling users’ preferences, the importance of geographical, categorical, and social information have been demonstrated (Zhao et al., 2016). Geographical information defines the physical range of user activity, categorical information reflects the types of venues favored by users, while social information reveals the social influence of locations on users. Simultaneously combining these three aspects may help alleviate problems caused by sparse data; however, this has been proven challenging (Geng, Jiao, Gong, Li & Wu, 2019). Early studies only adopted one or two aspects, while a few recent methods attempted to exploit all three aspects to model user preferences from a broader perspective (Ference, Ye & Lee, 2013; Leung, Lee & Lee, 2011; Noulas, Scellato, Lathia & Mascolo, 2012). In addition to the three aspects above, word-of-mouth opinions measured as location popularity from the public can influence users’ decision making (Liu, Fu, Yao & Xiong, 2013).
In this study, we propose a new location recommendation approach that measures users’ personalized preferences to locations by geographical, categorical, and social associations, as well as the popularity of locations. Although neither the three aspects of preferences nor location popularity are novel, we introduce a new method to integrate them into the model to improve the accuracy of location recommendation. Specifically, the main contributions of this study can be summarized as follows:
- •
A unified location recommendation framework that combines geographical, categorical, and social preferences with location popularity is proposed to measure users’ check-in probability of unvisited locations.
- •
A novel method is developed based on category hierarchy to capture categorical preference by calculating the semantic similarity between location tags.
- •
Unlike existing studies, location popularity is exploited as a global attribute of location in this study. The priority ranks of locations derived by all three types of preferences are further adjusted by location popularity.
- •
The proposed recommendation approach is experimentally evaluated on two large-scale datasets collected from Weeplaces and Yelp; subsequently, it is compared with two state-of-the-art location recommendation baselines.
The remainder of this paper is organized as follows. Related studies are reviewed in Section 2. The proposed method is introduced in Section 3 and compared with two state-of-the-art baselines on two datasets in Section 4. Subsequently, experimental results are discussed in Section 5, followed by conclusions and future studies in Section 6.
Section snippets
Related studies
The primary aim of location recommendation is to model user preference by exploiting useful information of users and locations from multiple perspectives. Geographical, social, and categorical information, as well as location popularity have been used in many recent recommendation methods.
Both geographical and social information are the most exploited to profile user preference. In general, these information model user preferences indirectly by mining check-in patterns, rather than revealing
Methodology
In this section, we propose a location recommendation approach by comprehensively considering geographical and categorical aspects of locations, social relations among users, as well as the popularity of locations (GCS-P).
Experimental results
In this section, the experiment settings are described for evaluating the performance of the proposed method against two state-of-the-art baselines.
Discussion
The experimental results show that the performance of GCS-P is significantly better than those of the baselines. In this section, we investigate how the recommendation accuracy is improved by analyzing the effects of information selection, method optimization, and data sparsity. Furthermore, a case study was conducted to compare the specific recommended locations of different techniques to comprehensively study factors affecting location recommendation quality.
Conclusion and future studies
Location recommendation is crucial in both increasing the visibility of locations and breaking the information overload dilemma for users. Multiple types of information have been exploited to solve the crucial problem of location recommendation, which is to precisely model preferences from exiguous check-in records. In this study, a unified location recommendation framework was proposed by incorporating geographical, categorical, and social preferences with location popularity to improve
CRediT authorship contribution statement
Yaxue Ma: Methodology, Software, Writing - original draft, Formal analysis. Jin Mao: Methodology, Investigation, Writing - review & editing. Zhichao Ba: Investigation, Visualization. Gang Li: Supervision, Funding acquisition.
Declaration of Competing Interest
We have no conflict of interest.
Acknowledgements
We thank the anonymous reviewers for their constructive comments. This study was funded by the National Natural Science Foundation of China (NSFC) Grant Nos. 71790612, 71603189, and 71804135. Jin Mao is sponsored by the China Postdoctoral Science Foundation Project (No. 2018M630885).
References (58)
- et al.
Discovering socially similar users in social media datasets based on their socially important locations
Information Processing and Management
(2018) - et al.
A personalized point-of-interest recommendation model via fusion of geo-social information
Neurocomputing
(2018) - et al.
A two-step personalized location recommendation based on multi-objective immune algorithm
Information Sciences
(2019) - et al.
Bayesian probabilistic matrix factorization with social relations and item contents for recommendation
Decision Support Systems
(2013) - et al.
Computers in human behavior role of fairness, accountability, and transparency in algorithmic affordance
Computers in Human Behavior
(2019) - et al.
The role of location and social strength for friendship prediction in location-based social networks
Information Processing and Management
(2018) - et al.
Location-aware service applied to mobile short message advertising: Design, development, and evaluation
Information Processing and Management
(2015) - et al.
Semantic trajectory-based high utility item recommendation system
Expert Systems with Applications
(2014) - et al.
Friend recommendation with content spread enhancement in social networks
Information Sciences
(2015) - et al.
Understanding the adoption of location-based recommendation agents among active users of social networking sites
Information Processing and Management
(2014)
Implementation of haversine formula and best first search method in searching of tsunami evacuation route
IOP Conference Series: Earth and Environmental Science
Random walk based context-aware activity recommendation for location based social networks
Location-based and preference-aware recommendation using sparse geo-social networking data
Recommendations in location-based social networks: A survey
GeoInformatica
GeoTeCS: Exploiting geographical, temporal, categorical and social aspects for personalized poi recommendation
Sensing the urban: Using location-based social network data in urban analysis
Who, what, when, and where: Multi-dimensional collaborative recommendations using tensor factorization on sparse user-generated data
Statistical analysis of separation distance between equatorial plasma bubbles near suvarnabhumi international airport, Thailand
Journal of Geophysical Research-Space Physics
On information coverage for location category based point-of-interest recommendation
A unified point-of-interest recommendation framework in location-based social networks
ACM Transactions on Intelligent Systems and Technology
Friendship and mobility: User movement in location-based social networks
RecNet: A deep neural network for personalized POI recommendation in location-based social networks
International Journal of Geographical Information Science
Integrating geographical and temporal influences into location recommendation: A method based on check-ins
Information Technology and Management
Collaborative filtering is not enough? Experiments with a mixed-model recommender for leisure activities
Location recommendation for out-of-town users in location-based social networks
Addressing the cold-start problem in location recommendation using geo-social correlations
Data Mining and Knowledge Discovery
Using location for personalized POI recommendations in mobile environments
Your neighbors affect your ratings: On geographical neighborhood influence to rating prediction
Point-of-interest recommendation in location-based social networks with personalized geo-social influence
China Communications
Cited by (23)
Sequence recommendations for groups: A dynamic approach to balance preferences
2022, Information SystemsCitation Excerpt :If different activities are related together, e.g., visiting points of interest in a city [34], listening to songs [35], it is interesting to provide a recommendation for the whole sequence of such activities. The vast majority of the works on this topic (e.g., [12,36–40]) focus on a single user, also when they take into consideration the influence of context, in particular relatively to the geographical dimension, neglecting the influence of the group composition on the satisfaction maximization. In our work, we consider the recommendation of a sequence of activities to a group of users.
Leveraging social influence based on users activity centers for point-of-interest recommendation
2022, Information Processing and ManagementCitation Excerpt :Considering these two influence factors, they suggest the POI to the user using the bookmark-coloring algorithm. Ma, Mao, Ba, and Li (2020) proposed a novel POI recommendation method that includes geographical, categorical, and social information with POI popularity. Geographical preferences are related to the probability of the user visiting close to their activity area.
Combining Non-sampling and Self-attention for Sequential Recommendation
2022, Information Processing and ManagementCitation Excerpt :However, these models only model static user preferences and cannot capture user dynamic interest changes. Recently, due to the successful application of deep learning in the field of recommendation systems, a large number of POI recommendation methods based on neural networks have also been proposed (Ma et al., 2020b; Yang, Bai, Zhang, Yuan, & Han, 2017; Yin, Wang, Wang, Chen, & Zhou, 2017). The task of next POI recommendation is to use the user’s previous historical check-in data to predict the next point of interest.
Adaptive time series prediction and recommendation
2021, Information Processing and ManagementCitation Excerpt :Chen, Li, Zhang, and Ma (2018) proposed a tunable temporal recommendation method based on the age and degree of nodes. Ma, Mao, Ba, and Li (2020) integrated the location popularity with geographical, categorical, and social features to make a location recommendation. Ding, Chen, Dong, and Herrera (2019) discussed the influence of nodes degree on consensus opinion formation in social networks.
A robust personalized location recommendation based on ensemble learning
2021, Expert Systems with ApplicationsCitation Excerpt :However, none of those studies was conducted within the context of location recommendations, and their conclusions might not be applicable to other contexts. Researchers (Lin et al., 2015; Ma, Mao, Ba, & Li, 2020; Yuan et al., 2013; Zhang & Chow, 2013, 2015) applied an ensemble framework to generate location recommendation lists. The final score of a location was combination of the results of individual recommenders, such as UBCF, SCF, FCF, power-law distribution, KDE, etc. (Table 1).
Multi-context embedding based personalized place semantics recognition
2021, Information Processing and Management