The emergence of Location-based Social Network (LBSN) services provides a wonderful opportunity to build personalized Point-of-Interest (POI) recommender systems. Although a personalized POI recommender system can significantly facilitate users' outdoor activities, it faces many challenging problems, such as the hardness to model user's POI decision making process and the difficulty to address data sparsity and user/location cold-start problem. To cope with these challenges, we define three types of friends (i.e., social friends, location friends, and neighboring friends) in LBSN, and develop a two-step framework to leverage the information of friends to improve POI recommendation accuracy and address cold-start problem. Specifically, we first propose to learn a set of potential locations that each individual's friends have checked-in before and this individual is most interested in. Then we incorporate three types of check-ins (i.e., observed check-ins, potential check-ins and other unobserved check-ins) into matrix factorization model using two different loss functions (i.e., the square error based loss and the ranking error based loss). To evaluate the proposed model, we conduct extensive experiments with many state-of-the-art baseline methods and evaluation metrics on two real-world data sets. The experimental results demonstrate the effectiveness of our methods.

Point-of-Interest Recommendations: Learning Potential Check-ins from Friends

The online car-hailing service has gained great popularity all over the world. As more passengers and more drivers use the service, it becomes increasingly more important for the the car-hailing service providers to effectively schedule the drivers to minimize the waiting time of passengers and maximize the driver utilization, thus to improve the overall user experience. In this paper, we study the problem of predicting the real-time car-hailing supply-demand, which is one of the most important component of an effective scheduling system. Our objective is to predict the gap between the car-hailing supply and demand in a certain area in the next few minutes. Based on the prediction, we can balance the supply-demands by scheduling the drivers in advance. We present an end-to-end framework called Deep Supply-Demand (DeepSD) using a novel deep neural network structure. Our approach can automatically discover complicated supply-demand patterns from the car-hailing service data while only requires a minimal amount hand-crafted features. Moreover, our framework is highly flexible and extendable. Based on our framework, it is very easy to utilize multiple data sources (e.g., car-hailing orders, weather and traffic data) to achieve a high accuracy. We conduct extensive experimental evaluations, which show that our framework provides more accurate prediction results than the existing methods.

DeepSD: Supply-Demand Prediction for Online Car-Hailing Services Using Deep Neural Networks

Theories of exploratory behavior suggest that inertia and variety-seeking tendencies may coexist within the individual, implying that the same individual may exhibit inertia and variety-seeking at different times depending on his/her choice history. Past research has not allowed for such within-consumer variability in these tendencies. The purpose of this study is to present a choice model that allows us to identify such "hybrid" behavior i.e., a mixture of inertia and variety-seeking, and to distinguish hybrid behavior from simpler types of behavior such as pure inertia, pure variety-seeking and zero-order behavior. The model is estimated at the household level using panel data for three product classes, covering a total of 1069 households. The empirical analysis shows that more than half the households analyzed exhibited hybrid behavior rather than simpler types of behavior.

Modeling inertia and variety-seeking tendencies in brand choice behavior (advertising, brand switching)

Point-of-Interest (POI) recommendation has become an important means to help people discover attractive and interesting places, especially when users travel out of town. However, the extreme sparsity of a user-POI matrix creates a severe challenge. To cope with this challenge, we propose a unified probabilistic generative model, the Topic-Region Model (TRM), to simultaneously discover the semantic, temporal, and spatial patterns of users’ check-in activities, and to model their joint effect on users’ decision making for selection of POIs to visit. To demonstrate the applicability and flexibility of TRM, we investigate how it supports two recommendation scenarios in a unified way, that is, hometown recommendation and out-of-town recommendation. TRM effectively overcomes data sparsity by the complementarity and mutual enhancement of the diverse information associated with users’ check-in activities (e.g., check-in content, time, and location) in the processes of discovering heterogeneous patterns and producing recommendations. To support real-time POI recommendations, we further extend the TRM model to an online learning model, TRM-Online, to track changing user interests and speed up the model training. In addition, based on the learned model, we propose a clustering-based branch and bound algorithm (CBB) to prune the POI search space and facilitate fast retrieval of the top-k recommendations. We conduct extensive experiments to evaluate the performance of our proposals on two real-world datasets, including recommendation effectiveness, overcoming the cold-start problem, recommendation efficiency, and model-training efficiency. The experimental results demonstrate the superiority of our TRM models, especially TRM-Online, compared with state-of-the-art competitive methods, by making more effective and efficient mobile recommendations. In addition, we study the importance of each type of pattern in the two recommendation scenarios, respectively, and find that exploiting temporal patterns is most important for the hometown recommendation scenario, while the semantic patterns play a dominant role in improving the recommendation effectiveness for out-of-town users.

Joint Modeling of User Check-in Behaviors for Real-time Point-of-Interest Recommendation

Social community detection is a growing field of interest in the area of social network applications, and many approaches have been developed, including graph partitioning, latent space model, block model and spectral clustering. Most existing work purely focuses on network structure information which is, however, often sparse, noisy and lack of interpretability. To improve the accuracy and interpretability of community discovery, we propose to infer users' social communities by incorporating their spatiotemporal data and semantic information. Technically, we propose a unified probabilistic generative model, User-Community-Geo-Topic (UCGT), to simulate the generative process of communities as a result of network proximities, spatiotemporal co-occurrences and semantic similarity. With a well-designed multi-component model structure and a parallel inference implementation to leverage the power of multicores and clusters, our UCGT model is expressive while remaining efficient and scalable to growing large-scale geo-social networking data. We deploy UCGT to two application scenarios of user behavior predictions: check-in prediction and social interaction prediction. Extensive experiments on two large-scale geo-social networking datasets show that UCGT achieves better performance than existing state-of-the-art comparison methods.

/pdf/discovering-interpretable-geo-social-communities-for-user-3xkl6uamtg.pdf

Discovering interpretable geo-social communities for user behavior prediction

Location-based data is increasingly prevalent with the rapid increase and adoption of mobile devices. In this paper we address the problem of learning spatial density models, focusing specifically on individual-level data. Modeling and predicting a spatial distribution for an individual is a challenging problem given both (a) the typical sparsity of data at the individual level and (b) the heterogeneity of spatial mobility patterns across individuals. We investigate the application of kernel density estimation (KDE) to this problem using a mixture model approach that can interpolate between an individual's data and broader patterns in the population as a whole. The mixture-KDE approach is evaluated on two large geolocation/check-in data sets, from Twitter and Gowalla, with comparisons to non-KDE baselines, using both log-likelihood and detection of simulated identity theft as evaluation metrics. Our experimental results indicate that the mixture-KDE method provides a useful and accurate methodology for capturing and predicting individual-level spatial patterns in the presence of noisy and sparse data.

/pdf/modeling-human-location-data-with-mixtures-of-kernel-1nqfp7or5p.pdf

Modeling human location data with mixtures of kernel densities

A challenge for studies assessing routine activities theory is accounting for the spatial and temporal confluence of offenders and targets given that people move about during the daytime and nightt...

/pdf/using-social-media-to-measure-temporal-ambient-population-1bovggo4tv.pdf

Using Social Media to Measure Temporal Ambient Population: Does it Help Explain Local Crime Rates?

There are numerous contexts where individuals typically consume a few items from a large selection of possible items. Examples include purchasing products, listening to music, visiting locations in physical or virtual environments, and so on. There has been significant prior work in such contexts on developing predictive modeling techniques for recommending new items to individuals, often using techniques such as matrix factorization. There are many situations, however, where making predictions for both previously-consumed and new items for an individual is important, rather than just recommending new items. We investigate this problem and find that widely-used matrix factorization methods are limited in their ability to capture important details in historical behavior, resulting in relatively low predictive accuracy for these types of problems. As an alternative we propose an interpretable and scalable mixture model framework that balances individual preferences in terms of exploration and exploitation. We evaluate our model in terms of accuracy in user consumption predictions using several real-world datasets, including location data, social media data, and music listening data. Experimental results show that the mixture model approach is systematically more accurate and more efficient for these problems compared to a variety of state-of-the-art matrix factorization methods.

Predicting Consumption Patterns with Repeated and Novel Events

In this paper we address the problem of building user models that can predict the rate at which individuals consume items from a finite set, including items they have consumed in the past and items that are new. This combination of repeat and new item consumption is common in applications such as listening to music, visiting web sites, and purchasing products. We use zero-inflated Poisson (ZIP) regression models as the basis for our modeling approach, leading to a general framework for modeling user-item consumption rates over time. We show that these models are more flexible in capturing user behavior than alternatives such as well-known latent factor models based on matrix factorization. We compare the performance of ZIP regression and latent factor models on three different data sets involving music, restaurant reviews, and social media. The ZIP regression models are systematically more accurate across all three data sets and across different prediction metrics.

/pdf/prediction-of-sparse-user-item-consumption-rates-with-zero-3mp7vrhm91.pdf

Prediction of Sparse User-Item Consumption Rates with Zero-Inflated Poisson Regression

Personalization is increasingly important for a range of applications that rely on location-based modeling. A key aspect in building personalized models is using population-level information to smooth noisy sparse data at the individual level. In this paper we develop a general mixture model framework for learning individual-level location models where the model adaptively combines different types of smoothing information. In a series of experiments with Twitter geolocation data and Gowalla check-in data we demonstrate that the proposed approach can be significantly more accurate than more traditional smoothing and matrix factorization techniques. The improvement in performance over matrix factorization is pronounced and may be explained by the tendency of dimensionality reduction methods to over-smooth and not retain enough detail at the individual level.

https://dl.acm.org/doi/pdf/10.1145/2996913.2996953

Moshe Lichman

Papers

Modeling human location data with mixtures of kernel densities

Using Social Media to Measure Temporal Ambient Population: Does it Help Explain Local Crime Rates?

Predicting Consumption Patterns with Repeated and Novel Events

Prediction of Sparse User-Item Consumption Rates with Zero-Inflated Poisson Regression

Personalized location models with adaptive mixtures