The prevalence of Predicting Point of Interest (POI) has been extremely important to location-based applications, such as reviews on POIs. Many current approaches are rarely able to achieve adequate efficiency due to the shortage of POI knowledge. This tendentious restricts the advice to famous locations and lacks equally important qualities in unlikely attractions. This paper introduces a novel method to forecasting the performance of POIs, dubbed Hierarchical Multi-Clue Fusion (HMCF). In general, to address sparsity issues, it is proposed that POIs be defined in a simple way usage different method of User-Generated Content (UGC) By different origin. And there is construct a hierarchically powerful POI modeling framework that concurrently injects semantonal Awareness and multiple layer representation regulation of POIs. Users are building a multi-view POI database for assessment by compiling both text and visual information from four conventional tourism channels from many separate provinces in China during 2006 to 2017. Extensive experimental findings indicate that the new technique will substantially improve the output of forecasting the success of attractions relative to a variety of reference methodologies.
Read full abstract