The Research On Yelp Catering Community Data Mining

In recent years,the e-commerce in catering industry is vigorous and developing rapidly.On-line meal ordering platform not only provides convenience for patrons,but also leads to the problem of information overload.For patrons,filtering useful information is not a small task.For platform managers,it is also a great challenge to provide retrieval services for billions of patrons and operate millions of restaurants.In view of the above reasons,more and more enterprises and research institutes seek the solutions by data mining and machine learning methods.Therefore,this thesis has finished a series of research works based on human dining dataset.The main contributions and results of this thesis are as follows:1.Foraging Pattern: We propose three kinds of entropies to characterize foraging pattern,with respect to both geography and cuisine.We show that modern foraging patterns of restaurant patrons in both geography and cuisine are of high regularity,indicating that their behaviors are rather predictable.What's more,we find that the relation between foraging patterns and individual social status in the community.2.Foraging Contagion: In order to estimate the effects of gender and network factors on foraging contagion separately,we use instrumental variable estimation to measure these contagions between patrons.The results show that foraging behavior is socially contagious which is associated with gender relationships between friends.Considering network motif theories of social contagion,the results suggest that network motifs in the patron's ego network express heterogeneous contagions.3.Restaurant Recommendation: In order to improve traditional recommendation algorithms which only use rating data,we propose a costing index to characterize a patron's foraging cost that is reconstructed from her foraging behavior,then construct recommendation model in both rating and costing indexes.Comparing with other recommendation algorithms,our method has improved significantly.Based on the Yelp empirical data,our work shows the effective applications of data mining technology in catering community,catering to the different needs of restaurant owners,dining patrons and platform managers.The results in our work could provide important references for owners to discern consumer preferences,for patrons to find appropriate friends,and to helping managers better operate catering community platform.