A Hybrid Recommendation Approach for Viral Food Based on Online Reviews

Abstract

Nowadays, there are many types of viral foods and consumers expect to be able to quickly find foods that meet their own tastes. Traditional recommendation systems make recommendations based on the popularity of viral foods or user ratings. However, because of the different sentimental levels of users, deviations occur and it is difficult to meet the user’s specific needs. Based on the characteristics of viral food, this paper constructs a hybrid recommendation approach based on viral food reviews and label attribute data. A user-based recommendation approach is combined with a content-based recommendation approach in a weighted combination. Compared with the traditional recommendation approaches, it is found that the hybrid recommendation approach performs more accurately in identifying the sentiments of user evaluations, and takes into account the similarities between users and foods. We can conclude that the proposed hybrid recommendation approach combined with the sentimental value of food reviews provides novel insights into improving the existing recommendation system used by e-commerce platforms.

Integrating Sentiment Analysis in Book Recommender System by using Rating Prediction and DBSCAN Algorithm with Hybrid Filtering Technique

“…When combined with Cen Song et al [17] . 's hybrid recommendation approach, the emotional value of food reviews shows how the current recommendation system used by e-commerce platforms is improved.…”