PROJECT TITLE :

NCF: A Neural Context Fusion Approach to Raw Mobility Annotation

ABSTRACT:

Improving business intelligence in mobile environments requires a thorough comprehension of human mobility patterns on a point-of-interest (POI) scale. This is an extremely important factor. The majority of studies just use POI check-ins to mine the concerned mobility patterns, which is an approach whose efficacy is typically hindered due to the fact that there is a lack of data. This is despite the fact that significant efforts have been made in this direction. In this paper, our goal is to directly annotate the POIs that are associated with raw user-generated mobility records so that we can obtain more accurate POI-based human mobility data for mining. We propose a neural context fusion approach that integrates a variety of context factors into people's patterns of point-of-interest (POI) visits. Through the use of representation learning, our method analyzes both the preferences and the transitional factors. To deal with the randomized changes in raw mobility, one notable aspect of our approach is the incorporation of an attention mechanism. The domain knowledge factors, such as distance, time, and popularity, continue to be effective, and our method incorporates them further from a data-driven point of view. A feed-forward neural network is used to automatically fuse together the various factors. In addition to this, we make use of a multi-head architecture so that the model can express itself more fully. Our experimental study was conducted by utilizing two different sets of real-world data, and the results showed that our method consistently outperformed the state-of-the-art baselines in terms of accuracy by at least 32 percent. In addition, using a POI recommendation example, we illustrate the utility of the obtained POI-based human mobility.