Negative ratings play a positive role in information filtering

Abstract

The explosive growth of information asks for advanced information filtering techniques to solve the so-called information overload problem. A promising way is the recommender system which analyzes the historical records of users' activities and accordingly provides personalized recommendations. Most recommender systems can be represented by userobject bipartite networks where users can evaluate and vote for objects, and ratings such as ''dislike'' and ''I hate it'' are treated straightforwardly as negative factors or are completely ignored in traditional approaches. Applying a local diffusion algorithm on three benchmark data sets, MovieLens, Netflix and Amazon, our study arrives at a very surprising result, namely the negative ratings may play a positive role especially for very sparse data sets. In-depth analysis at the microscopic level indicates that the negative ratings from less active users to less popular objects could probably have positive impacts on the recommendations, while the ones connecting active users and popular objects mostly should be treated negatively. We finally outline the significant relevance of our results to the two long-term challenges in information filtering: the sparsity problem and the coldstart problem.

Recommender systems

“…Different from user similarity, an algorithm based on item similarity recommends a user the objects that are similar to what this user has collected before. Note that, sometimes the opinions from dissimilar users [117] or the negative ratings [118,119] can play a significant (even positive) role in determining the recommendation, especially when the data set is very sparse and thus the information about relevance is more important than that about correlation [120]. For additional information see the recent review articles [121,122], and [123] is a nice survey that contains a number of similarity indices.…”

Recommender Systems

“…Different from user similarity, an algorithm based on item similarity recommends a user the objects that are similar to what this user has collected before. Note that, sometimes the opinions from dissimilar users [117] or the negative ratings [118,119] can play a significant (even positive) role in determining the recommendation, especially when the data set is very sparse and thus the information about relevance is more important than that about correlation [120]. For additional information see the recent review articles [121,122], and [123] is a nice survey that contains a number of similarity indices.…”

A review of a recommendation filtering system approach based on reliable sustainable opinion mining

“…Zuva et al (2012) use the same parameters of Massa (2006) and Zeng et al (2011) to define trust and adds that trust exists despite a possible negative outcome.…”

“…The definition of Yang et al (2020) is similar to that of Zeng et al (2011). Kohavi et al (2009) show that trust allows to decrease the complexity of the environment as people make daily choices that help them trust and adapt to their environment.…”