Analysis on Yelp Friend Network

EasyChair Preprint no. 10986

Abstract

Social networking functions like word of mouth when it comes to reviews and ratings. Given the data of users and reviews from the Yelp dataset, we first build a network of friends to evaluate the influence of social networking on user reviews and stars, and then another network of reviews on businesses to try to identify characteristics of successful businesses in their corresponding network of user reviews. For milestone 2, we percept the influence of friend relations among users given the context. Meanwhile, we rebuild the second network mentioned above by taking temporal information into account. We mined the new network and extracted feature embeddings for nodes before combining them with common network metrics and building the final feature vectors for the new networks of businesses. Finally, we trained several classifiers for the task of determining whether a business is successful or not and made evaluations of the results.

Keyphrases: Natural Language Processing, Network Feature Mining, online user behavior, social networking, Yelp data analysis

@Booklet{EasyChair:10986,
  author = {Chuang Ke and Minghao Zhu},
  title = {Analysis on Yelp Friend Network},
  howpublished = {EasyChair Preprint no. 10986},

  year = {EasyChair, 2023}}