Trust-based recommendation for the social Web

The social web presents new opportunities for e-commerce and services provided to the consumer. This paper presents an online recommendation system that eases the matching of a user with the most relevant products and services. The designed system consists of three recommendation algorithms; two of them are already widely discussed in the literature, while the third one is new and was proposed to consider the trust relationship between users and their peers. This work also presents an experiment with real users that intends to evaluate the proposed new algorithm in comparison with the other two and also with direct, human-to-human recommendations. The paper presents the results gathered on experimental data analysis and the statistical hypotheses tests that were performed, which allowed concluding in which circumstances trust-based recommendation is advantageous.