, pp.421-426 http://dx.doi.org/10.14257/astl.2017.147.60 Study and Analysis of Recommendation Systems for Location Based Social Network (LBSN) N. Ganesh 1, K. SaiShirini 1, Ch. AlekhyaSri 1 and Venkata Naresh Mandhala 1 1 Department of Computer Science and Engineering, KL University, Vaddeswaram, Guntur {ganeshchandu1996, sshirini4, alekhya.dimmy, mvnaresh.mca}@gmail.com Abstract. This paper describes the overview of recommending new venues to users who participate in location based social network. As an increasingly larger number of users participate in LBSN, the recommendation problem in this area has attracted attention in research and in practical applications. The detailed information about past user behavior that is traced by LBSN differentiates the problem significantly from its traditional settings. In this paper, we are going to analyze different types of recommendation systems and at last we will recommend which type of recommendation system best suits for LBSN. Keywords: Location based social network, recommender system, content based filtering, collaborative filtering, hybrid filtering. 1 Introduction Recommender systems are information filtering systems that deal with the problem of information overload by filtering dynamic information It has the ability to predict whether a particular user would desire an item or not based on the user s profile. In location based social network, user share information about their locations, places their visit with other users. Visits are reported explicitly by user check-ins or implicitly by allowing smart phone applications to report visited locations to the LBSN. This information is then shared with the other users who are socially related. Recommending new locations is an important issue; it allows to efficiently advertise companies with physical presence such as theatres, bars, restaurants etc. Recommender systems are useful to both service providers and users. They reduce transaction costs of finding and selecting items in an online shopping environment. Recommendation systems have also showed to improve decision making process and quality. Recommendation system is used to generate meaningful suggestions to users. Recommending new location to users is done by recommendation algorithms. The Process of recommending location is done with a help of keyword. By studying real LBSN data we can confirm that the social links of a user and the distance of a location to his/her previous visits are important factors in predicting whether he/she will visit a new location or not. This is the era of the web, lots of web data are accessible on the web. On the web, where the number of choices are irresistible, so there is need of the ISSN: 2287-1233 ASTL Copyright 2017 SERSC
information filtering on the web. Although many dissimilar approaches to recommender systems have been established in past few years, the interest in this area is still high because of growing demand to practical applications, which can deal with the personalized recommendation and deal with large amount of overloaded data. A personalized information filtering technology used to either predict whether a certain user will like a certain product or item or to identify a set of N items products that will be the interest to a particular user. Further many other applications also used in the recommendation system for different various purposes like make more profit in industry, make an real and efficient personalized result in the system. In this paper describes Section 2 describes the Existing system. Section 3. describes the proposed system. Section 4 describes the comparative study analysis and Section 5 gives the Conclusion. 2 Existing System There are many real world applications available that use recommendation systems to help their users to find more appropriate products. Microsoft word is one best example of RS, in which words are recommended to users. Mobile keypads, in which messaging application, users type words are learned intelligently and in the future the wrong words are recommending by the correct words. Amazon, movie lens, LIBRA, Pandora, Google etc are using recommending systems. The Amazon uses collaborative filtering to recommend items to users. 3 Proposed System The spatial nature in the past user behavior and also the information about the user social interaction with other users, provide a richer background to build a more accurate and expressive recommendation model. Although there have been extensive studies on recommender systems working with user-item ratings, GPS trajectories, and other types of data, there are very few approaches that exploit the unique properties of the LBSN user check-in data. we propose algorithms that create recommendations based on four factors: a) past user behavior(visited places), b) the location of each venue, c) the social relationships among the users, and d) the similarity between users. 4 Comparative Study There are two approaches in the recommendation system. a) Personalized-RS b) Non-Personalized-RS 422 Copyright 2017 SERSC
Non-Personalized RS:-The system gives same recommendation of items to all the users. This type of approach is done in you tube for most popular video by aggregation method like average, mean etc. Personalized RS: - It considers each user interest or preference, so more effectively recommend particular item to the user or the group of the users. Mainly there are three types of Recommendation systems:- 1) Content based recommendation system. 2) Collaborative based recommendation system. 3) User based recommendation system. 4) Hybrid based recommendation system. 4.1 Content Based Recommendation System Content based recommendation system is a domain-dependent technique and it emphasizes more in the analysis of the attributes of items in order to produce recommendations. It recommend based on similarities between items or things. It recommends items based on user profile. This type of profile is created at the beginning, when the user creates the account and starts using the system. As user increases interaction between system the more possibly it will recommend the items to the user. The idea is if the user likes the item in the past then user probably like. After some time, CBF compare the user items profile with the current items profile and try to recommend similar type of items in the future. For example:- If there are set of movies. If we select one movie then at the bottom it will show recommendation of other movies to the user. It is based on type of movie. It internally categorize based whether it is an animated, action or comedy. Problems of content based recommendation systems a) New user problem b) Over specialization problem a) New user problem:- When new user enter into system, system does not have sufficient information. b) Over specialization problem:- System does not recommend items that are different from anything that the user has seen before. This may be problem because user wants to try something new and the system would never recommend new items. Copyright 2017 SERSC 423
4.2 Collaborative Based Recommendation System Collaborative recommendation system is a domain-independent prediction model for the content that cannot easily and satisfactorily be analyzed by metadata. It recommend not only by similarities but also by other user preferences. It is based on productions. The system recommends items based on past rating of all users collectively. It works by collecting users feedback in the form of rating for items in a given domain. It exploits similarities in rating behavior. For example:- If there are user A and user B, if A likes movie X and Y, then if user B likes movie X then recommendation system recommends movie Y to B based on recommendation system algorithm. Problems in collaborative recommendation system For this recommendation system initially some data is required. If there is no data collected initially then we cannot recommend it to the users. This problem is called start up problem. 4.3 Hybrid Based Recommendation System It is a combination of both content based and collaborative based recommended system. It picks recommendation based on similarities and user preferences. Most of the sites use this recommendation system. 4.4 User Based Recommendation System In user based recommendation system recommendations are done by user browsing history like purchases and rating taken into the account. This algorithm makes proposals by considering users having similar interest. It relates user as per the rating is allocating to the product. 5 Four Square Dataset We use four square dataset in order to store the data. This data set is used for recommendation of new places to the user. This data set contains user Check-In information, location where the place is there. The places latitude and longitude information is stored. Recommendation algorithm checks the user id and user check-in and based on that it will recommend places for the user when he want to visit new location. If it is a new place where no user went, then it will collect data from the experts. So, based on user behavior it will recommend places for this case. It is purely based on predictions. 424 Copyright 2017 SERSC
6 Comparison There are different methods in Recommendation Systems. Each of the methods has their advantages and some disadvantages, where some of the new approaches are created to overcome the disadvantages. In the content-based method, there are some problems like new user problem and overspecialization. So, to overcome these disadvantages, new method is created that are collaborative filtering. In collaborative based method, it overcomes the disadvantages of content-based filtering. As we know every method has its pros and cons collaborative method also has its cons like start-up problem, where it can be avoided by analyzing expert review. The comparison of both the methods is described below in Table 1. Table 1. Comparison of Content and Collaborative Recommendation Systems. Content based recommendation system Collaborative based recommendation system Recommendations are done Recommendations are done based on similarities based on predictions. between items. Main problem is over Main problem is start-up specialization. It cannot be problem. But this can be overcome avoided by analyzing expert review. 7 Conclusion In this paper, we discussed various real world existing recommendation systems. Also we discussed their strengths and limitations. So, collaborative recommendation system best suits for LBSN rather than remaining. We can use hybrid based recommendation system as it contains both features of RS. In future, to design our recommendation algorithm we study real time LBSNs like Bright kite and Gowalla, and analyze the relation between user and visited location. References 1. Shah, L., Gaudani, H., & Balani, P. (2016).: Survey on recommendation system. system, 137(7). 2. Thomas, A., & Sujatha, A. K. (2016, March).: Comparative study of recommender systems. In Circuit, Power and Computing Technologies (ICCPCT), 2016 International Conference on (pp. 1-6). IEEE. 3. Wang, H., Terrovitis, M., & Mamoulis, N. (2013, November).: Location recommendation in location-based social networks using user check-in data. In Proceedings of the 21st ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (pp. 374-383). ACM. Copyright 2017 SERSC 425
4. Xiang, L. (2012).: Recommendation system practice. People s Posts and Telecommunications Press, Beijing, 179-193. 5. Asabere, N. Y. (2012).: Review of Recommender Systems for Learners in Mobile Social/Collaborative Learning. International Journal of Information, 2(5). 6. Núñez-Valdéz, E. R., Lovelle, J. M. C., Martínez, O. S., García-Díaz, V., De Pablos, P. O., & Marín, C. E. M. (2012).: Implicit feedback techniques on recommender systems applied to electronic books. Computers in Human Behavior, 28(4), 1186-1193. 7. Lops, P., De Gemmis, M., & Semeraro, G. (2011).: Content-based recommender systems: State of the art and trends. In Recommender systems handbook (pp. 73-105). Springer US. 8. Zhao, Z. D., & Shang, M. S. (2010, January).: User-based collaborative-filtering recommendation algorithms on hadoop. In Knowledge Discovery and Data Mining, 2010. WKDD'10. Third International Conference on (pp. 478-481). IEEE. 9. Sarwar, B., Karypis, G., Konstan, J., & Riedl, J. (2001, April).: Item-based collaborative filtering recommendation algorithms. In Proceedings of the 10th international conference on World Wide Web (pp. 285-295). ACM. 10. Ramaswamy, L., Deepak, P., Polavarapu, R., Gunasekera, K., Garg, D., Visweswariah, K., & Kalyanaraman, S. (2009, May).: Caesar: A context-aware, social recommender system for low-end mobile devices. In Mobile Data Management: Systems, Services and Middleware, 2009. MDM'09. Tenth International Conference on (pp. 338-347). IEEE. 11. Goldberg, K., Roeder, T., Gupta, D., & Perkins, C. (2001).: Eigentaste: A constant time collaborative filtering algorithm. Information Retrieval, 4(2), 133-151. 12. Sarwar, B., Karypis, G., Konstan, J., & Riedl, J. (2001, April).: Item-based collaborative filtering recommendation algorithms. In Proceedings of the 10th international conference on World Wide Web (pp. 285-295). ACM. 13. Pronk, V., Verhaegh, W., Proidl, A., & Tiemann, M. (2007, October).: Incorporating user control into recommender systems based on naive bayesian classification. In Proceedings of the 2007 ACM conference on Recommender systems (pp. 73-80). ACM. 14. Amatriain, X., Pujol, J. M., & Oliver, N. (2009, June).: I like it... i like it not: Evaluating user ratings noise in recommender systems. In International Conference on User Modeling, Adaptation, and Personalization (pp. 247-258). Springer, Berlin, Heidelberg. 15. Balabanović, M., & Shoham, Y. (1997).: Fab: content-based, collaborative recommendation. Communications of the ACM, 40(3), 66-72 426 Copyright 2017 SERSC