Location Guidance

DOI : 10.17577/IJERTCONV5IS20039

Download Full-Text PDF Cite this Publication

Text Only Version

Location Guidance

Find a Way by using POI Recommendations

Aishwarya M.K

Aishwarya Satish Pai

Kavyashree C

Dept. of CSE

Dept. of CSE

Dept. of CSE




Likitha B Asha Rani M

Dept. of CSE Dept. of CSE


Abstract Point-of-Interest recommendation is an essential means to help people to discover an attractive locations, especially when people travel out of town or to unfamiliar regions.

While a growing line of research has focu sed on modelling user geographical preferences for POI recommendation, they ignore the phenomenon of user interest drift across geographical regions, i.e., users tend to have different interests when they travel in different regions, So we propose a latent class probabilistic generative model Spatial-Temporal LDA (ST-LDA) to learn region- dependent personal interests according to the contents of their checked-in POIs at each region. In another concept called LBSNs (Location Based Social Network), users can post their physical locations in the form of check-in, and share their life experiences in the physical world. It is crucial to utilize user check-in data to make personalized POI recommendation in LBSNs, which helps users know new POIs and explore new regions (e.g., cities),and POI approaches for u ser to make the travel package plans to travel a city , find the places between the sou rce and destination and find the places according to user privileges.

Keywords Location-based social networks (LBS Ns), point-of-interest (POI), types of recommendation, location recommendation, area recommendation, travel route recommendation,travel package recommendation, travel route planning (TRP) algorithm.

  1. Introduc tion

    Point-of-Interest recommendation is an essential means to help people discover attractive locations, especially when people travel out of town or to unfamilia r regions.

    When people visit different places they can upload the photos of visited locations, post their comments, share their experiences with heir friends, share their location via check-in using this application. This check-in info rmation is used for point of interests like historical places, temp les, restaurants etc. This task of recommending unexplo red new p laces is referred to as Point of Interest (POI).

    POI reco mmender system considers frequency of check-in data at a particula r location, social friends interest. Generally POI is recommended in ma jority of cases based on check-in information wh ich is obtained when users share their location using this application. Reco mmendation is done taking into consideration user preferences extracted fro m location based service such as at which time user visit which location and travelling interest.

    Crowdness is calculated based on the users check- in information. Crowdness calculation will help the people to know about the places, whether the place is crowded or not. For the crowdness to be calculated the users must use the application and check-in to the

    particular location. Based on the network used by user crowd-ness will be calculated.

    Fig1: How to discover new locations.

    Fig 2: T wo major challenges for personalized travel recommendation.

    There are two ma jor challenges (see Fig. 2). The first is dynamically obtaining users travel preference. User study is a conventional investigation strategy, but it is not scalable for a large number of users. It also cannot reflect the transfer of travelers preferences, unless a survey is periodically conducted, which is infeasible. The wide use of LBSNs results in the possibility of leveraging the crowd sourced data generated to obtain user preferences. The second one is that POIs in a travel package are constrained by physical conditions and are interre lated with each other. In contrast with single POI reco mmendations, reco mmendation of mu lti-POIs is more like ly to be a constrained optimization proble m. The selection and order of POIs are determined by a set of factors, including user preference to POIs, distance between POIs.

    The ma in contributions of this study are as follows.

    1. User profiling and location modeling based on LBSNs: The check-in behavior of LBSNs indicates the actual visit of locations, while traditional online behaviors, like web search or browsing, do not. Therefore, the spatiotemporal trajectories from LBSNs can be regarded as the actual connections between users and locations. Thus, we propose constructing the user profile and location model based on crowdsourced check-in records.

    2. Generating modules with respect to POI recommendation:

      Search location Place near me Plan city ravel

      Find places between source and destination.

      History Location ala rm Point-of interest

      Search location: Locations are added to the database by admin istrator. The user can select the places from location list and should enter the current location. The application will route the best path from current location to the selected location.

      Place near me: the users current location is fetched and near- by privileges are shown on the map within the radius of 3 kms. For e xa mp le if the user selects ATM privilege, a ll ATMs within the specified radius will be shown the map with markers on it. When particular ATM is selected its name and address will displayed.

      Privileges can be ATMs, Hospitals, Hotels, Te mples and so on.

      Plan city travel: The user enters city name, start time and end time. The list of locations that can be visited in the entered location will be shown. The user selects the places. For selected places priorities will be assigned and best route is calculated.

      Find places between source and destination: the user here enters the source name and destination name. The places between specified source and destination will be listed. User can select his/her interested places. Map of selected places will be generated. On the way to trip privileges are shown. Trip is planned on a time ly manner. Rev iew of any places can be checked.

      History: history module will contain the informat ion about the places previously visited by the user. This helps him / her to know or vis it the same place again.

      Location alarm: Alarms can be set when particular p lace/location is reached by the user. It is users interest to set alarm to any particula r location.

      Point-of-interest: places listed according to the point-of-interest of the user.

    3. System prototyping and performance evaluation: We implemented a prototype system with a mobile client and a recommendation server. We conducted a trace-driven simulation to evaluate system performance. Results suggest that the proposed approach can improve the recommendation accuracy with moderate

  2. Related Work

There is lot of study conducted on the travel package re co mm e n datio n over spatial data and check in h istory.

Zhiwen u et.a l [1] proposes a system that predict locations as per users interest and generate travel package. Users preference changes with time and hence is dyn a mi cal ly e xt racted fro m LBSN .Thus user profiles are continuously updated. Location popularity is considered and rating is a lso given importance. Package reco mmendation consist of sequence of POIs useful for t ravel p lanning. First the system finds POIs which are near to current loation of the user. Then it calculates prefe rences for these new POIs depending on what user likes based on information obtained fro m user profile . This process is repeated for different time slots. Thus the systemdetermine route to take and the travel package.

H. Yin et. al[2] uses Spatial Te mporal LDA model to recommend POIs at each region which are region d ep e nd e nt and ac c or din g to user interest as well. It uses the fact that users tend to have different interest when they travel to different i.e . out of town regions. Chec k- in records of local users is mined to learn local c rowd preferences. Check-in records fro m outsiders will be used to learn tourist preferences. Also an algorith m is developed to speed up the recommendation process. Daily act ivity done at different t ime is also considered.

Jihang Ye[3] try to predict most like ly category of user activity to be done next using check-in category informat ion. By using mixed hidden Markov model this prediction is done. Category level modeling reduces the huge prediction space which is result of millions of check in information. The system addresses the problem as two sub-problems viz. pred icting category of user activity at next step and predicting location depending on category distribution.

  1. Lu [4] propose a system wh ich predicts sequence of travel routes to be taken using geotagged photos

    .These geotagged photos uploaded by various people

    are the aggregated to recover possible travel routes. Also user choice like duration and visiting time and destination type preferences etc are considered.

    H.Yin et. a l[5] proposes recommender system that recommend a set of locations and event by considering both individual users interest and as well as preferences of local crowd . Helps the user while visiting unknown places. Offline module captures co- occurrence patterns and explo its Item contents. Online module takes user query to predict top recommendations for POIs but taking into consideration the interest of tourist and local crowd. It uses threshold algorithm for speeding up online process. It integrates collaborative and content into a probabilistic generative model.

    Yonghong Yu et. al [6] surveys POIs reco mmendation in LBSN. It finds that POI reco mmendation uses Tobblers Law that everything is related to everything else but near things are more re lated than distant things. This means that people prefer to visit nearby location w.r.t. their current locations. Users preferences are reflected through check- in frequency for locations. Lots of Check-in info rmation c reates sp arsity pr o ble m for POI re co m m e n d ation . It shown by Ye, Yin in their study that social influence has limited contribution on users check-in behaviour. It identifies 4 categories of POIs viz. pure check-in , geographical influence enhanced POI ,social influence enhanced POI and temporal influence enhanced POI .The survey concludes that although all kinds of information is used, still check-in data, geographical influence and temporal influence have significant impact on recommendation quality.

    Yin et.al [7] proposes to use probabilistic model na med TRM for pred iction of POIs taking into consideration semantic, te mporal and spatial patterns of users check-in act ivities. It is used for home town as well as unknown new place recommendation. The proposed system can effect ively solve sparsity and cold start problems .Se mantic patterns minimizes the effect of data sparsity when analyzing unknown place recommendation and temporal patterns are used in hometown reco mmendation. Its limitation is it assumes users interest are stable across geographical regions.

    Zhiwen Yu et. al [8] proposes a system that recommend a travel package by ma king use of crowd sourced data fro m LBSNs. Considering user choice of places to visit , it mines the check in records to find peculiar points of interest (POIs) characteristics. Constraints such as travel season, time period and starting location are also considered while recommending the travel package.

    Qi Liu et. al[9] built a system based on the characteristics of existing travel packages. The topic e xtraction is done considering the constraints like locations, seasons, tourist. These topics are then used for personalized travel pac kage reco mmendation. The model is modified by min ing patterns that exist in relationships among tourists in each group. Through e xperiments it is proved that the proposed model identify the features of the travel data which helps in more p roper reco mmendation for travel packages.

    Chen Cheng et. al[10] propose to find successive personalized POI reco mmendation in LBSNs. Two prominent properties in the check-in sequence: personalized Markov chain and region localization are used for this purpose. This system does not consider the temporal re lations.

    Gregory Fe rence et. al[11] proposes location recommendation system for out of town users by

    ma king use of user preference, social friends influence and geographical c loseness to the current location. It uses the fact that similar users will like similar places .This fact is used for in-town users while recommendation is made for out of town users based more impo rtantly on social influence

    Bo Hu et. al[12] proposes a system model capturing both the social interact ion and topic aspects of user check-ins. It e xp lores areas of social network-based recommender systems. Based on friends interests and their check-in behavior, location reco mmendation is made.

    Yan-Ying Chen et. al[13] propose to conduct personalized travel reco mmendation not only by using community contributed geo tagged photos but also specific user attributes like age, gender , cultural background, profession etc. and type of people travelling with like fa mily, friends, couple etc. Personalized reco mmendation is made with respect to users interest and attributes.

    III. Pr oposed Work System arc hitecture:

    Fig3: system architecture

    We aim to build a location guidance for user to find places according to the poit-of-interest of the user to the places in travelling.

    The above figure(fig3) shows the requirement components for the system .GPS satellite system provides an goggle map ,internet provides the goggle APIs and online map facilities, web application is used by admin and mobile application and its modules are listed out.


    The figure (fig4) shows the flow of the main modules of the location guidance system.

    Fig4:flowchart fo r location guidance ap plication.

    Conclusion and Future En ha nc e m en t

    1. We can find the loc ation a c co rdi ng to point of interest of the user.

    2. We can plan the c ity travel ac c ordi ng to user timing and his interest and places availability.

    3. We did location search for d istrict wise for states, find the places according to privileges beet win source

      Mysore and destination Bengaluru.

    4. We can save the location and set the alarm for that location, we can share the reviews of a physical location by check in to that place.

    5. We can discover the new locations.

    6. We can save the check-in location to history of the user.

Future enhanceme nt:

  1. Improve the crowd-ness calculation for a region by using artificial intelligence

  2. Increase the places list in location search.

  3. Find the places between any source and any destination.

  4. Share the images in revie w share for the location.

    Re fe re nces

    1. Zhiwen Yu, Huang Xu, Zhe Yang, and Bin Guo, Personalized Travel Package With Multi-Point-of-Interest Recommendation Based on Crowdsourced User Footprints IEEE Trans. On Human machine sytems, Feb 2016.

    2. Hongzhi Yin, Xiaofang Zhou, Bin Cui, Hao Wang, Kai Zheng and Quoc Viet Hung Nguyen Adapting to User Interest Drift for POI Recommendation IEEE Trans on Data and Knowledge, 2016.


    3. H. Cheng, J. Ye, and Z. Zhu. Whats your next move: User activity prediction in location-based social networks. In SDM, pages 171179, 2013.

    4. X. Lu, C. Wang, J.-M. Yang, Y. Pang, and

      L. Zhang, Photo2trip: Generating travel routes from geo-tagged photos for trip planning, in Proc. Int. Conf. Multimedia, 2010, pp. 143152.

    5. H. Yin, B. Cui, Y. Sun, Z. Hu, and L. Chen. Lcars: A spati al item recommender system, ACM Trans. Inf. Syst., 32(3):11:1 11:37, July 2014.

    6. Yonghong Yu,Xingguo Chen A survey of Point of Interest Recommendation in Location Based Social Networks,AAAI Workshop, 2015.

    7. H. Yin, B. Cui, X. Zhou, W. Wang, Z. Huang, and S. Sadiq. Joint mode ling of user check-in behaviors for real-time point-of-interest recommendation., ACM Trans. Inf. Syst., 2016.

    8. Z. Yu, Y. Feng, H. Xu, and X. Zhou, Recommending travel packages based on mobile crowdsourced data, IEEE Commun. Mag., vol. 52,no. 8, pp. 5662, Aug. 2014.

    9. Q. Liu, E. Chen, H. Xiong, Y. Ge, Z. Li, and X. Wu, A cocktail approach for travel package recommendation, IEEE Trans. Knowl. Data Eng., vol. 26, no. 2 , pp. 278293, Feb. 2014.

    10. BC. Cheng,H. Yang, M. R. Lyu, and I. King. Where you like to go next: Successive point-of-interest recommendation. In IJCAI, pages 2605 2611, 2013.

    11. G. Ference, M. Ye, and W.-C. Lee. Location recommendation for out-of town users in location-based social networks. ,In CIKM, pages 721726, 2013.

    12. B. Hu and M. Ester. Social topic modeling for point-of-interest recommendation in location-based social networks. ,In ICDM, pages 845850, 2014.

    13. Yan-Ying Chen, An-Jung Cheng, and Winston H. Hsu Travel Recommendation by Mining People Attributes and Travel Group Types From CommunityContributed Photos IEEE Transc. Oct.

Leave a Reply