Recommender systems based on opinion mining and deep neural networks

Recommender systems based on opinion mining and deep neural networks

Blog Article

To address rating sparsity problem, various review-based recommender systems have been developed in recent years.Most m1917a1 kelly helmet of them extract topics, opinions, and emotional polarity from the reviews by using the techniques of text analysis and opinion mining.According to existing researches, review-based recommendation methods utilize review elements in rating prediction model, but underuse the actual ratings provided by users.

In this paper, we adopt one lexicon-based opinion mining method to extract opinions hidden in reviews, and also, we combine opinions with actual ratings.In addition, we embed deep neural networks model which breaks through the limitation of traditional collaborative perdieme lingerie filtering.The experimental results based on two public datasets indicate that this personalized model provides an effective recommendation performance.