Filters








141,075 Hits in 3.5 sec

Deconfounded Causal Collaborative Filtering [article]

Shuyuan Xu and Juntao Tan and Shelby Heinecke and Jia Li and Yongfeng Zhang
2021 arXiv   pre-print
We first frame user behaviors with unobserved confounders into a causal graph, and then we design a front-door adjustment model carefully fused with machine learning to deconfound the influence of unobserved  ...  Recommender systems may be confounded by various types of confounding factors (also called confounders) that may lead to inaccurate recommendations and sacrificed recommendation performance.  ...  Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect those of the sponsors.  ... 
arXiv:2110.07122v1 fatcat:2ytfb42y6bcndopzvs4gp3455m

Collaborative Filtering with Social Exposure: A Modular Approach to Social Recommendation [article]

Menghan Wang, Xiaolin Zheng, Yang Yang, Kun Zhang
2017 arXiv   pre-print
Inspired by recent causal process based recommendations that first model user exposures towards items and then use these exposures to guide rating prediction, we utilize social information to capture user  ...  exposures rather than user preferences.  ...  So we assume that social information influences users on the exposure level, and then influences rating predictions by exposures.  ... 
arXiv:1711.11458v1 fatcat:msw6zwnyirbbxcwt2jdhza7eji

Causal Disentanglement with Network Information for Debiased Recommendations [article]

Paras Sheth, Ruocheng Guo, Lu Cheng, Huan Liu, K. Selçuk Candan
2022 arXiv   pre-print
., user-social and user-item networks), which are shown to influence how users discover and interact with an item.  ...  Considering these entangled representations for inferring user preference may lead to biased recommendations (e.g., when the recommender model recommends popular items even if they do not align with the  ...  For instance, the mechanism with which a user is either exposed or not exposed to an item can influence the exposure value and the rating value for that user-item pair.  ... 
arXiv:2204.07221v1 fatcat:t7tkuh3425hcxgazt3siuzygty

Improving Compliance with Password Guidelines: How User Perceptions of Passwords and Security Threats Affect Compliance with Guidelines

Florence Mwagwabi, Tanya McGill, Michael Dixon
2014 2014 47th Hawaii International Conference on System Sciences  
The study described in this paper investigates how user perceptions of passwords and security threats affect intended compliance with guidelines and explores how these perceptions might be altered in order  ...  It tests a research model based on protection motivation theory [24] . Two groups of internet users were surveyed, one of which received password security information and an exercise to reinforce it.  ...  The proposed model has been shown to be an acceptable model for explaining user compliance with password guidelines.  ... 
doi:10.1109/hicss.2014.396 dblp:conf/hicss/MwagwabiMD14 fatcat:2z27uqeugrh6hmi5qegn76b2se

Modeling and Counteracting Exposure Bias in Recommender Systems [article]

Sami Khenissi, Olfa Nasraoui
2020 arXiv   pre-print
Then we model the exposure that is borne from the interaction between the user and the recommender system and propose new debiasing strategies for these systems.  ...  Our research findings show the importance of understanding the nature of and dealing with bias in machine learning models such as recommender systems that interact directly with humans, and are thus causing  ...  CONCLUSION In this paper, we provided a personalized model to model the user exposure along with the experimental protocol used to confirm its effectiveness.  ... 
arXiv:2001.04832v1 fatcat:4tcc56lcrjflhjbibjknxljdv4

Bias and Debias in Recommender System: A Survey and Future Directions [article]

Jiawei Chen, Hande Dong, Xiang Wang, Fuli Feng, Meng Wang, Xiangnan He
2021 arXiv   pre-print
While recent years have witnessed a rapid growth of research papers on recommender system (RS), most of the papers focus on inventing machine learning models to better fit user behavior data.  ...  In this paper, we first summarize seven types of biases in recommendation, along with their definitions and characteristics.  ...  [67] Meta learning [3] Conformity Bias Modeling popularity influence [23] Modeling social influence [25] , [26] , [27] , [28] Exposure Bias Evaluator Propensity Score [68] Training Heuristic  ... 
arXiv:2010.03240v2 fatcat:6fticc3otndsra2whs5e4nrdpi

Considering aesthetics and usability temporalities in a model based development process

Sophie Dupuy-Chessa, Yann Laurillau, Eric Céret
2016 Actes de la 28ième conférence francophone sur l'Interaction Homme-Machine on - IHM '16  
ACKNOWLEDGMENTS Realised in collaboration with the CERAG laboratory, this work has been funded by the UPMF University and the INNOVACS research federation.  ...  Regarding user interface usability, the results show that usability does not influence the participants' evaluation when the users visually inspect the website (exposure stage).  ...  As recommended in the Cameleon reference framework [2] , the process ( Fig. 1 ) is based on four kinds of models: Tasks, Domain, Abstract User Interface (AUI), Concrete User Interface" (CUI).  ... 
doi:10.1145/3004107.3004122 dblp:conf/ihm/Dupuy-ChessaLC16 fatcat:4kokqsdxvjat5h5u2h5m6lqehq

Using Exploration to Alleviate Closed Loop Effects in Recommender Systems

Amir H. Jadidinejad, Craig Macdonald, Iadh Ounis
2020 Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval  
Hence the deployed recommendation systems will recommend some items and not others, and items will have varying levels of exposure to users.  ...  Moreover, with the aid of exploration we can decrease the effect of closed loop feedback and obtain new and better generalizable models.  ...  CONCLUSIONS Recommendation systems are an instance of dynamic systems where a simple causal reasoning about the users' preferences is difficult because both the recommender and the user have a direct influence  ... 
doi:10.1145/3397271.3401230 dblp:conf/sigir/JadidinejadMO20 fatcat:p6bkgio42zaahb6kjkewjntd5i

Methodological aspects of the implementation of the new ICRP recommendations

W. Raskob, C. Landman, Tatiana Duranova, Wolfgang Raskob, Raimo Mustonen, Thierry Schneider
2013 Radioprotection - Revue de la Societé Francaise de Radioprotection  
Vol. 48, n o 5, pages S43 à S47 Methodological aspects and updates of computational models RadiopRotection -VoL. 48 -© edp Sciences, 2013 S43 AbstRACt With the ICRP recommendations Publications 103, 109  ...  sure they will influence countermeasure simulation approaches: 1. the concept of a "reference level" for emergency and existing controllable exposure situations that represents the level of dose or risk  ...  Therefore, the design of the new model has to be performed together with their users to assure that the current usability will not be traded against functionality.  ... 
doi:10.1051/radiopro/20139907 fatcat:6lofphqt5bfknbjqadvaau442q

SamWalker++: recommendation with informative sampling strategy [article]

Can Wang, Jiawei Chen, Sheng Zhou, Qihao Shi, Yan Feng, Chun Chen
2021 arXiv   pre-print
SamWalker models data confidence with a social network-aware function, which can adaptively specify different weights to different data according to users' social contexts.  ...  Thus, we further propose SamWalker++, which does not require any side information and models data confidence with a constructed pseudo-social network.  ...  by connected friends are ranked higher than those not. • SERec-Bo [42] : A probabilistic model that extends the EXMF model with social influence on user's exposure.  ... 
arXiv:2011.07734v2 fatcat:viztewi3hvhpnoklxvbqvd62we

Model-Agnostic Counterfactual Reasoning for Eliminating Popularity Bias in Recommender System [article]

Tianxin Wei, Fuli Feng, Jiawei Chen, Chufeng Shi, Ziwei Wu, Jinfeng Yi, Xiangnan He
2020 arXiv   pre-print
., fitting a recommender model to recover the user behavior data with pointwise or pairwise loss, makes the model biased towards popular items.  ...  The general aim of the recommender system is to provide personalized suggestions to users, which is opposed to suggesting popular items.  ...  Once the exposure model is estimated, the preference model is fit with weighted click data, where each click is weighted by the inverse of exposure estimated in the first model and thus be used to alleviate  ... 
arXiv:2010.15363v1 fatcat:oqazsv33qres5kssf6lsb4fply

Exposure Inequality in People Recommender Systems: The Long-Term Effects [article]

Francesco Fabbri, Maria Luisa Croci, Francesco Bonchi, Carlos Castillo
2021 arXiv   pre-print
People recommender systems may affect the exposure that users receive in social networking platforms, influencing attention dynamics and potentially strengthening pre-existing inequalities that disproportionately  ...  Our extensive experimentation with the proposed model shows that a minority group, if homophilic enough, can get a disproportionate advantage in exposure from all link recommenders.  ...  Observation 2 Different recommenders exhibit different influence on exposure along time. ALS increases exposure inequality in the first iterations, then stabilizing in a steady state.  ... 
arXiv:2112.08237v1 fatcat:6zo37ujzafeulplmxd4zlfmvju

Unbiased metrics of friends' influence in multi-level networks

Alexandre Vidmer, Matúš Medo, Yi-Cheng Zhang
2015 EPJ Data Science  
This calls for metrics to measure the influence of users on the behavior of their friends.  ...  We use a simple network model based on the influence of friends and preferential attachment to illustrate the performance of our metrics at different levels of friends' influence.  ...  The influence of users sending purchase recommendations to their friends has been studied in [] .  ... 
doi:10.1140/epjds/s13688-015-0057-x fatcat:y7knnbdrwngq3mp4siv3cb5eoy

Fast Adaptively Weighted Matrix Factorization for Recommendation with Implicit Feedback

Jiawei Chen, Can Wang, Sheng Zhou, Qihao Shi, Jingbang Chen, Yan Feng, Chun Chen
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
Naturally, how to assign confidence weights and how to handle the large number of the unobserved data are two key problems for implicit recommendation models.  ...  The personalized data confidence weights are adaptively assigned with a parameterized neural network (function) and the network can be inferred from the data.  ...  Different users may have different influence strength on the communities .  ... 
doi:10.1609/aaai.v34i04.5751 fatcat:jzwi7d2en5affoglyhfxbm7noa

Constrained User Exposure Matrix Factorization in Recommendation System

Zhi-cai HUANG, Shun-zhi ZHU, Weng WEI, Ying ZHONG, Nan QIN
2018 DEStech Transactions on Engineering and Technology Research  
collaborative filter model and apply it to deal with the matrix sparsity problem, which it makes a better recommendation to users who have very few ratings.  ...  The experimental results show that the proposed model gets higher prediction precision than the state-of-the-art models in providing recommendation for user having very few ratings.  ...  [2] propose a generative model called Exposure matrix factorization model (ExpoMF), which adding an user exposure variable and considering another items information to influence the prior of user exposure  ... 
doi:10.12783/dtetr/icmeit2018/23462 fatcat:6lrbf65omjcdddllif64roaekq
« Previous Showing results 1 — 15 out of 141,075 results