Trust No One: Evaluating Trust-Based Filtering for Recommenders
Abstract
To be successful recommender systems must gain the trust of users. To do this they must demonstrate their ability to make reliable predictions. We argue that collaborative filtering recommendation algorithms can benefit from explicit models of trust to inform their predictions. We present one such model of trust along with a cost-benefit analysis that focuses on the classical trade-off that exists between recommendation coverage and prediction accuracy.
Cite
Text
O'Donovan and Smyth. "Trust No One: Evaluating Trust-Based Filtering for Recommenders." International Joint Conference on Artificial Intelligence, 2005.Markdown
[O'Donovan and Smyth. "Trust No One: Evaluating Trust-Based Filtering for Recommenders." International Joint Conference on Artificial Intelligence, 2005.](https://mlanthology.org/ijcai/2005/oaposdonovan2005ijcai-trust/)BibTeX
@inproceedings{oaposdonovan2005ijcai-trust,
title = {{Trust No One: Evaluating Trust-Based Filtering for Recommenders}},
author = {O'Donovan, John and Smyth, Barry},
booktitle = {International Joint Conference on Artificial Intelligence},
year = {2005},
pages = {1663-1665},
url = {https://mlanthology.org/ijcai/2005/oaposdonovan2005ijcai-trust/}
}