This article proposes a new approach to predict user location that combines spatial and temporal check-in data with social network information. It improves upon existing models like PMM and PSMM by applying a radiation model to calculate location probabilities. An evaluation using check-in data from 90 users found the radiation model approach achieved better prediction accuracy than baseline methods. The research contributes an enhanced method for predicting user locations in location-based social networks.
This article proposes a new approach to predict user location that combines spatial and temporal check-in data with social network information. It improves upon existing models like PMM and PSMM by applying a radiation model to calculate location probabilities. An evaluation using check-in data from 90 users found the radiation model approach achieved better prediction accuracy than baseline methods. The research contributes an enhanced method for predicting user locations in location-based social networks.
This article proposes a new approach to predict user location that combines spatial and temporal check-in data with social network information. It improves upon existing models like PMM and PSMM by applying a radiation model to calculate location probabilities. An evaluation using check-in data from 90 users found the radiation model approach achieved better prediction accuracy than baseline methods. The research contributes an enhanced method for predicting user locations in location-based social networks.
Radiation Model and Social Check-lns Brief description of the work Predicting the location of a person at a certain time is very important for many computer applications. For instance, venue recommender systems, city traffic modeling systems, urban planning systems
This research article proposes a new approach for the
prediction of user location at a particular time. Continue The approach has two components. These are Spatio-temporal component which uses coordinates and times of user check-ins and social component which use of the social interaction of a user. The research is aimed to improve the performance of a state-of the art model using the radiation model. Continue The performance enhancement is through increasing the accuracy of the prediction ability of the model as compared to the existing models such as the periodic mobility model (PMM) and the periodic and social mobility model (PSMM). Periodic mobility model (PMM) PMM consists of two parts spatial and temporal. The spatial part of the PMM represents both clusters as 2D Gaussian distribution. The temporal part model the probability of the individual being in some where represents as 1D Gaussian distribution and PMM does not take time into account. Periodic and social mobility model (PSMM) PSMM an extension of PMM. It consists of two components spatio-temporal and social. PSMM uses only social connections of a given user instead of use knowledge about categories of venues (cinema, coffee shop, football stadium etc.) as well as global patterns of mobility, i.e. number of check-ins at venues performed by all users. Radation model Radiation model approach uses Spatio-temporal information form user check-ins by applying the radiation model instead of Gaussians used by PSMM and PMM. Moreover it is based on coordinate and time the user check-ins and social interaction b/n different users. Continue The radiation model can be written as follows. Pij = m.n/[(m+ Sij) (m+n + Sij)]
Where Pij is the probability of an individual moving
from location i to location j and m is the attractiveness of i and j is the attractiveness of j. Sij the attractiveness of the places within a cirlcle of radius Rij centered at i. Continue The research gets a data from the social networks as a basic ingredient to find out the location of a particular user at a point in time. Examples of such social networks include face book and foursquare. It takes 90 users in the data set check-ins that users made on Foursquare between 13/07/2012 and 24/07/2012. Social network communication Continue The results of our experiment strongly suggest that both the spatio-temporal and social components proposed in this paper perform better than the state- of-the-art baseline. The radiation model proved to be better than PMM and PSMM. Objective of the research work General objective The general objective of the research is to come up with a predictive model for the location of a certain user at a certain time of Location based social networks with better performance and accuracy than the current-state of art model(PSMM). Specific objective Answering the following questions. What is the best combination of spatio-temporal and social components? What is the best social component? What is the best spatio-temporal component? Contribution of the work The research contribution is the introduction of the Radiation model to improve the performance of the Periodic and social Mobility model (PSMM). Critiques Strong Points The overall document organization of the article is good.
The abstract represents the whole document very well.
The problem is identified and stated properly.
Related works have been discussed properly.
Continue The research article uses recent reference. Out of 17 references 16 of them published within the last 5 years.
The conclusion and future work section of the
research paper gives general finalization in the article and an indication for future work.
The document is citied properly.
Weak Points The title of the research, more or less describe the content but it would have been better if the title included the word improving/enhancing.
There are few spelling errors. For example, on page 3
the writers have misspelled the word than as then. Moreover, there are few unnecessary punctuation marks on the same page. For instance, the formula on page 3 contain the “‘” symbol. Continue Introduction tries to introduce the topic but the discussion is more on the related works.
In experimental setup section of the article lacks
clarity. For example, it uses the Stanford without telling the reader what it represents. Continue The sample data taken for the experiment is moderate. It uses data of only 90 users. But it would have been better if the sample size were bigger than this.
The research does not make any distinctions between
the training and test data. It seems, it is using the same sample data for both purposes. Continue The research states that people movements are the result of periodic movements and social influences with out considering impact of culture and religion.
The methodology of research meaning the data
capture is clear but the process of approaches and statistical method are complicate to understandable. Summary Prediction of user location at certain time is now much better predicted due to the contribution of this research paper. It increases the accuracy of prediction of user’s locations due to the incorporation of Radiation model. It tried to fills the gaps left out by the previously applied approaches. Thank You The End