Recommendations with minimum exposure guarantees: a post-processing framework. Lopes, R., Alves, R., Ledent, A., Santos, R. L. T., & Kloft, M. Expert Systems with Applications, 236:121164, February, 2024. ![Recommendations with minimum exposure guarantees: a post-processing framework [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex Relevance-based ranking is a popular ingredient in recommenders, but it frequently struggles to meet fairness criteria because social and cultural norms may favor some item groups over others. For instance, some items might receive lower ratings due to some sort of bias (e.g. gender bias). A fair ranking should balance the exposure of items from advantaged and disadvantaged groups. To this end, we propose a novel post-processing framework to produce fair, exposure-aware recommendations. Our approach is based on an integer linear programming model maximizing the expected utility while satisfying a minimum exposure constraint. The model has fewer variables than previous work and thus can be deployed to larger datasets and allows the organization to define a minimum level of exposure for groups of items. We conduct an extensive empirical evaluation indicating that our new framework can increase the exposure of items from disadvantaged groups at a small cost of recommendation accuracy.
@article{lopes_recommendations_2024-1,
title = {Recommendations with minimum exposure guarantees: a post-processing framework},
volume = {236},
issn = {0957-4174},
shorttitle = {Recommendations with minimum exposure guarantees},
url = {https://www.sciencedirect.com/science/article/pii/S0957417423016664},
doi = {10.1016/j.eswa.2023.121164},
abstract = {Relevance-based ranking is a popular ingredient in recommenders, but it frequently struggles to meet fairness criteria because social and cultural norms may favor some item groups over others. For instance, some items might receive lower ratings due to some sort of bias (e.g. gender bias). A fair ranking should balance the exposure of items from advantaged and disadvantaged groups. To this end, we propose a novel post-processing framework to produce fair, exposure-aware recommendations. Our approach is based on an integer linear programming model maximizing the expected utility while satisfying a minimum exposure constraint. The model has fewer variables than previous work and thus can be deployed to larger datasets and allows the organization to define a minimum level of exposure for groups of items. We conduct an extensive empirical evaluation indicating that our new framework can increase the exposure of items from disadvantaged groups at a small cost of recommendation accuracy.},
urldate = {2023-09-19},
journal = {Expert Systems with Applications},
author = {Lopes, Ramon and Alves, Rodrigo and Ledent, Antoine and Santos, Rodrygo L. T. and Kloft, Marius},
month = feb,
year = {2024},
keywords = {Exposure, Fairness, Integer linear programming, Recommender systems, to-read},
pages = {121164},
}
Downloads: 0
{"_id":"ziWpjgqYEycPADrNA","bibbaseid":"lopes-alves-ledent-santos-kloft-recommendationswithminimumexposureguaranteesapostprocessingframework-2024","author_short":["Lopes, R.","Alves, R.","Ledent, A.","Santos, R. L. T.","Kloft, M."],"bibdata":{"bibtype":"article","type":"article","title":"Recommendations with minimum exposure guarantees: a post-processing framework","volume":"236","issn":"0957-4174","shorttitle":"Recommendations with minimum exposure guarantees","url":"https://www.sciencedirect.com/science/article/pii/S0957417423016664","doi":"10.1016/j.eswa.2023.121164","abstract":"Relevance-based ranking is a popular ingredient in recommenders, but it frequently struggles to meet fairness criteria because social and cultural norms may favor some item groups over others. For instance, some items might receive lower ratings due to some sort of bias (e.g. gender bias). A fair ranking should balance the exposure of items from advantaged and disadvantaged groups. To this end, we propose a novel post-processing framework to produce fair, exposure-aware recommendations. Our approach is based on an integer linear programming model maximizing the expected utility while satisfying a minimum exposure constraint. The model has fewer variables than previous work and thus can be deployed to larger datasets and allows the organization to define a minimum level of exposure for groups of items. We conduct an extensive empirical evaluation indicating that our new framework can increase the exposure of items from disadvantaged groups at a small cost of recommendation accuracy.","urldate":"2023-09-19","journal":"Expert Systems with Applications","author":[{"propositions":[],"lastnames":["Lopes"],"firstnames":["Ramon"],"suffixes":[]},{"propositions":[],"lastnames":["Alves"],"firstnames":["Rodrigo"],"suffixes":[]},{"propositions":[],"lastnames":["Ledent"],"firstnames":["Antoine"],"suffixes":[]},{"propositions":[],"lastnames":["Santos"],"firstnames":["Rodrygo","L.","T."],"suffixes":[]},{"propositions":[],"lastnames":["Kloft"],"firstnames":["Marius"],"suffixes":[]}],"month":"February","year":"2024","keywords":"Exposure, Fairness, Integer linear programming, Recommender systems, to-read","pages":"121164","bibtex":"@article{lopes_recommendations_2024-1,\n\ttitle = {Recommendations with minimum exposure guarantees: a post-processing framework},\n\tvolume = {236},\n\tissn = {0957-4174},\n\tshorttitle = {Recommendations with minimum exposure guarantees},\n\turl = {https://www.sciencedirect.com/science/article/pii/S0957417423016664},\n\tdoi = {10.1016/j.eswa.2023.121164},\n\tabstract = {Relevance-based ranking is a popular ingredient in recommenders, but it frequently struggles to meet fairness criteria because social and cultural norms may favor some item groups over others. For instance, some items might receive lower ratings due to some sort of bias (e.g. gender bias). A fair ranking should balance the exposure of items from advantaged and disadvantaged groups. To this end, we propose a novel post-processing framework to produce fair, exposure-aware recommendations. Our approach is based on an integer linear programming model maximizing the expected utility while satisfying a minimum exposure constraint. The model has fewer variables than previous work and thus can be deployed to larger datasets and allows the organization to define a minimum level of exposure for groups of items. We conduct an extensive empirical evaluation indicating that our new framework can increase the exposure of items from disadvantaged groups at a small cost of recommendation accuracy.},\n\turldate = {2023-09-19},\n\tjournal = {Expert Systems with Applications},\n\tauthor = {Lopes, Ramon and Alves, Rodrigo and Ledent, Antoine and Santos, Rodrygo L. T. and Kloft, Marius},\n\tmonth = feb,\n\tyear = {2024},\n\tkeywords = {Exposure, Fairness, Integer linear programming, Recommender systems, to-read},\n\tpages = {121164},\n}\n\n","author_short":["Lopes, R.","Alves, R.","Ledent, A.","Santos, R. L. T.","Kloft, M."],"key":"lopes_recommendations_2024-1","id":"lopes_recommendations_2024-1","bibbaseid":"lopes-alves-ledent-santos-kloft-recommendationswithminimumexposureguaranteesapostprocessingframework-2024","role":"author","urls":{"Paper":"https://www.sciencedirect.com/science/article/pii/S0957417423016664"},"keyword":["Exposure","Fairness","Integer linear programming","Recommender systems","to-read"],"metadata":{"authorlinks":{}}},"bibtype":"article","biburl":"https://api.zotero.org/users/6655/collections/3TB3KT36/items?key=VFvZhZXIoHNBbzoLZ1IM2zgf&format=bibtex&limit=100","dataSources":["7KNAjxiv2tsagmbgY"],"keywords":["exposure","fairness","integer linear programming","recommender systems","to-read"],"search_terms":["recommendations","minimum","exposure","guarantees","post","processing","framework","lopes","alves","ledent","santos","kloft"],"title":"Recommendations with minimum exposure guarantees: a post-processing framework","year":2024}