DeepStore: An Interaction-Aware Wide&Deep Model for Store Site Recommendation With Attentional Spatial Embeddings

MetaStore: A Task-adaptative Meta-learning Model for Optimal Store Placement with Multi-city Knowledge Transfer

“…Recently, with the rapid development of DNNs [26,37], more and more works propose DNNbased approaches to improve the performance of store placement by characterizing consumption behavior based on multi-source data [23,48]. Liu et al [23] propose a model named DeepStore, including the cross network, the deep network, and the linear component, thus, it can learn low-and high-order feature interactions explicitly and implicitly from dense and sparse features simultaneously to model complex user behavior. Xu et al [48] propose an attentive neural method to select promising business locations by fusing the discriminative features extracted from urban data and satellite data.…”

“…Intuitively, identifying whether the candidate location is appropriate to place a new store in the long term mainly depends on the nearby users. Following previous work [23], we associate each user with a location-based community, which is a group of homes and other buildings built together. To characterize potential customers in different communities, we consider some demographic profiles, such as gender, age, profession, income level, and so on, and then make statistics on the number of people with different profiles in each community as user features.…”

“…Nevertheless, these methods rely on expertise feature engineering to characterize sophisticated influences and extract features from a single data source, which fail to learn complex feature interactions from multi-source data. Recently, with the rapid development of deep neural networks (DNNs), more and more works propose DNN-based approaches [23,48] to improve the performance of site selection by characterizing consumption behavior based on multi-source data. The method proposed in Reference [23] learns low-and highorder feature interactions simultaneously to model complex user behaviors.…”

“…Recently, with the rapid development of deep neural networks (DNNs), more and more works propose DNN-based approaches [23,48] to improve the performance of site selection by characterizing consumption behavior based on multi-source data. The method proposed in Reference [23] learns low-and highorder feature interactions simultaneously to model complex user behaviors. However, the superior performance of existing models is conditioned on large-scale training data, and most of them cannot work in some new cities with a few historical consumption data where a chain enterprise just develops its market at the initial phase.…”

Knowledge Transfer in Commercial Feature Extraction for the Retail Store Location Problem

“…These two metrics are used to compute a global index (Jensen's Quality Index) that can be employed to assess the quality of a potential location in terms of the attractive or repulsive forces that neighboring stores exert on it. Jensen's Quality Index is very popular in the retail research field, where it is often used as an input for supervised learning models [18][19][20] and for location recommendation systems [20].…”