Toward Fair Recommendation in Two-sided Platforms

Many online platforms today (such as Amazon, Netflix, Spotify, LinkedIn, and AirBnB) can be thought of as two-sided markets with producers and customers of goods and services. Traditionally, recommendation services in these platforms have focused on maximizing customer satisfaction by tailoring the results according to the personalized preferences of individual customers. However, our investigation reinforces the fact that such customer-centric design of these services may lead to unfair distribution of exposure to the producers, which may adversely impact their well-being. However, a pure producer-centric design might become unfair to the customers. As more and more people are depending on such platforms to earn a living, it is important to ensure fairness to both producers and customers. In this work, by mapping a fair personalized recommendation problem to a constrained version of the problem of fairly allocating indivisible goods, we propose to provide fairness guarantees for both sides. Formally, our proposed FairRec algorithm guarantees Maxi-Min Share of exposure for the producers, and Envy-Free up to One Item fairness for the customers. Extensive evaluations over multiple real-world datasets show the effectiveness of FairRec in ensuring two-sided fairness while incurring a marginal loss in overall recommendation quality. Finally, we present a modification of FairRec (named as FairRecPlus ) that at the cost of additional computation time, improves the recommendation performance for the customers, while maintaining the same fairness guarantees.

Download Full-text

Incremental Fairness in Two-Sided Market Platforms: On Smoothly Updating Recommendations

Proceedings of the AAAI Conference on Artificial Intelligence ◽

10.1609/aaai.v34i01.5349 ◽

2020 ◽

Vol 34 (01) ◽

pp. 181-188 ◽

Cited By ~ 6

Author(s):

Gourab K. Patro ◽

Abhijnan Chakraborty ◽

Niloy Ganguly ◽

Krishna Gummadi

Keyword(s):

Customer Satisfaction ◽

Real World ◽

Well Being ◽

Online Optimization ◽

Smooth Transition ◽

Goods And Services ◽

Online Platforms ◽

Customer Evaluations ◽

Abrupt Changes ◽

Real World Datasets

Major online platforms today can be thought of as two-sided markets with producers and customers of goods and services. There have been concerns that over-emphasis on customer satisfaction by the platforms may affect the well-being of the producers. To counter such issues, few recent works have attempted to incorporate fairness for the producers. However, these studies have overlooked an important issue in such platforms -- to supposedly improve customer utility, the underlying algorithms are frequently updated, causing abrupt changes in the exposure of producers. In this work, we focus on the fairness issues arising out of such frequent updates, and argue for incremental updates of the platform algorithms so that the producers have enough time to adjust (both logistically and mentally) to the change. However, naive incremental updates may become unfair to the customers. Thus focusing on recommendations deployed on two-sided platforms, we formulate an ILP based online optimization to deploy changes incrementally in η steps, where we can ensure smooth transition of the exposure of items while guaranteeing a minimum utility for every customer. Evaluations over multiple real world datasets show that our proposed mechanism for platform updates can be efficient and fair to both the producers and the customers in two-sided platforms.

Download Full-text

EXTENSIONS TO THE K-AMH ALGORITHM FOR NUMERICAL CLUSTERING

Journal of Information and Communication Technology ◽

10.32890/jict2018.17.4.8272 ◽

2018 ◽

Author(s):

Ali Seman ◽

Azizian Mohd Sapawi

Keyword(s):

Real World ◽

Confidence Level ◽

Categorical Data ◽

Euclidean Distance ◽

P Value ◽

Accuracy Score ◽

Original Algorithm ◽

Real World Datasets ◽

The Cost ◽

Numerical Clustering

The k-AMH algorithm has been proven efficient in clustering categorical datasets. It can also be used to cluster numerical values with minimum modification to the original algorithm. In this paper, we present two algorithms that extend the k-AMH algorithm to the clustering of numerical values. The original k-AMH algorithm for categorical values uses a simple matching dissimilarity measure, but for numerical values it uses Euclidean distance. The first extension to the k-AMH algorithm, denoted k-AMH Numeric I, enables it to cluster numerical values in a fashion similar to k-AMH for categorical data. The second extension, k-AMH Numeric II, adopts the cost function of the fuzzy k-Means algorithm together with Euclidean distance, and has demonstrated performance similar to that of k-AMH Numeric I. The clustering performance of the two algorithms was evaluated on six real-world datasets against a benchmark algorithm, the fuzzy k-Means algorithm. The results obtained indicate that the two algorithms are as efficient as the fuzzy k-Means algorithm when clustering numerical values. Further, on an ANOVA test, k-AMH Numeric I obtained the highest accuracy score of 0.69 for the six datasets combined with p-value less than 0.01, indicating a 95% confidence level. The experimental results prove that the k-AMH Numeric I and k-AMH Numeric II algorithms can be effectively used for numerical clustering. The significance of this study lies in that the k-AMH numeric algorithms have been demonstrated as potential solutions for clustering numerical objects.

Download Full-text

Real-world comparative economics of a 12-gene assay for prognosis in stage II colon cancer.

Journal of Clinical Oncology ◽

10.1200/jco.2013.31.15_suppl.3640 ◽

2013 ◽

Vol 31 (15_suppl) ◽

pp. 3640-3640

Author(s):

Tiffany Yu ◽

Steven R. Alberts ◽

Robert J. Behrens ◽

Lindsay A. Renfro ◽

Geetika Srivastava ◽

...

Keyword(s):

Colon Cancer ◽

Adverse Events ◽

Real World ◽

Well Being ◽

Medical Costs ◽

Stage Ii ◽

Direct Medical Costs ◽

Gene Assay ◽

The Cost ◽

Improve Patient

3640 Background: Prior economic analysis of a 12-gene assay (Oncotype DX), compared with patterns of care reported in the NCCN database of patients with stage II, T3, DNA mismatch repair proficient (MMR-P) colon cancer, predicted that the assay would save medical costs and improve patient well-being (Hornberger et al. Value Health 2012). This study assessed the validity of those findings with actual adjuvant chemotherapy (aCT) recommendations. Methods: Outcomes and costs were estimated for patients with stage II, T3, MMR-P colon cancer using a Markov model. A study of 141 patients from 17 sites in the Mayo Clinic Cancer Research Consortium collected data on aCT recommended before and after knowledge of the 12-gene assay results (Srivastava et al. abstract). Quality-adjusted life years (QALY) and medical resource use after recurrence were computed using guideline-validated state-transition probability estimation methods. Risk of progression and incidence of adverse events with different aCT regimens were based on published literature. Drug and administration costs for aCT were obtained from 2012 Medicare fee schedules. One-way sensitivity analyses were conducted to evaluate parameter influence on economic impact. Results: After receiving the 12-gene assay results, physician recommendations in favor of aCT decreased 22% (95% CI 11%-32%; McNemar test p<0.001) from 73 (52%) to 42 (30%) patients. Oxaliplatin aCT and 5-FU monotherapy recommendations each declined 11%. Average aCT costs decreased $3,228 for drugs, $750 for administration, and $3,168 for adverse events management. Overall, average total direct medical costs decreased $1,683. The net effect on average patient well-being was a gain of 0.102 QALYs. Total change in medical costs is most influenced by the cost of death due to colon cancer, time-preference discount rate, and the change in aCT recommendations. Savings are expected to persist even if the cost of oxaliplatin dropped by >75% due to generic substitution. Conclusions: The 12-gene assayhas been shown to alter aCT recommendations for patients with stage II, T3, MMR-P colon cancer. This study provides real-world confirmation that these aCT changes reduce direct medical costs and improve patient well-being.

Download Full-text

A Bandit Approach to Maximum Inner Product Search

Proceedings of the AAAI Conference on Artificial Intelligence ◽

10.1609/aaai.v33i01.33014376 ◽

2019 ◽

Vol 33 ◽

pp. 4376-4383

Author(s):

Rui Liu ◽

Tianyi Wu ◽

Barzan Mozafari

Keyword(s):

Real World ◽

State Of The Art ◽

Linear Time ◽

Identification Problem ◽

Approximate Algorithm ◽

Inner Product ◽

Approximate Algorithms ◽

Product Search ◽

Real World Datasets ◽

The Cost

There has been substantial research on sub-linear time approximate algorithms for Maximum Inner Product Search (MIPS). To achieve fast query time, state-of-the-art techniques require significant preprocessing, which can be a burden when the number of subsequent queries is not sufficiently large to amortize the cost. Furthermore, existing methods do not have the ability to directly control the suboptimality of their approximate results with theoretical guarantees. In this paper, we propose the first approximate algorithm for MIPS that does not require any preprocessing, and allows users to control and bound the suboptimality of the results. We cast MIPS as a Best Arm Identification problem, and introduce a new bandit setting that can fully exploit the special structure of MIPS. Our approach outperforms state-of-the-art methods on both synthetic and real-world datasets.

Download Full-text

Efficient Heuristic Hypothesis Ranking

Journal of Artificial Intelligence Research ◽

10.1613/jair.615 ◽

1999 ◽

Vol 10 ◽

pp. 375-397 ◽

Cited By ~ 3

Author(s):

S. Chien ◽

A. Stechert ◽

D. Mutz

Keyword(s):

Incomplete Information ◽

Real World ◽

Expected Loss ◽

Ranking Problem ◽

Additional Information ◽

Decision Cycle ◽

Ranking Procedures ◽

Probably Approximately Correct ◽

Real World Datasets ◽

The Cost

This paper considers the problem of learning the ranking of a set of stochastic alternatives based upon incomplete information (i.e., a limited number of samples). We describe a system that, at each decision cycle, outputs either a complete ordering on the hypotheses or decides to gather additional information (i.e., observations) at some cost. The ranking problem is a generalization of the previously studied hypothesis selection problem - in selection, an algorithm must select the single best hypothesis, while in ranking, an algorithm must order all the hypotheses. The central problem we address is achieving the desired ranking quality while minimizing the cost of acquiring additional samples. We describe two algorithms for hypothesis ranking and their application for the probably approximately correct (PAC) and expected loss (EL) learning criteria. Empirical results are provided to demonstrate the effectiveness of these ranking procedures on both synthetic and real-world datasets.

Download Full-text

TRANSFER FİYATLAMA UYGULAMASININ İŞLETME PERFORMANSINA ETKİSİ VE VERGİSEL AÇIDAN DEĞERLENDİRİLMESİ

Business And Management Studies An International Journal ◽

10.15295/bmij.v1i2.63 ◽

2014 ◽

Vol 1 (2) ◽

pp. 187

Author(s):

Serdar KUZU

Keyword(s):

Business Performance ◽

Transfer Pricing ◽

Tax Law ◽

Economic Activities ◽

Tax System ◽

Goods And Services ◽

Tax Legislation ◽

International Transactions ◽

The Cost ◽

The Impact

The size of international trade continues to extend rapidly from day to day as a result of the globalization process. This situation causes an increase in the economic activities of businesses in the trading area. One of the main objectives of the cost system applied in businesses is to be able to monitor the competitors and the changes that can be occured as a result of the developments in the sector. Thus, making cost accounting that is proper according to IAS / IFRS and tax legislation has become one of the strategic targets of the companies in most countries. In this respect, businesses should form their cost and pricing systems according to new regulations. Transfer pricing practice is usefull in setting the most proper price for goods that are subject to the transaction, in evaluating the performance of the responsibility centers of business, and in determining if the inter-departmental pricing system is consistent with targets of the business. The taxing powers of different countries and also the taxing powers of different institutions in a country did not overlap. Because of this reason, bringing new regulations to the tax system has become essential. The transfer pricing practice that has been incorporated into the Turkish Tax System is one of the these regulations. The transfer pricing practice which includes national and international transactions has been included in the Corporate Tax Law and Income Tax Law. The aim of this study is to analyse the impact of goods and services transfer that will occur between departments of businesses on the responsibility center and business performance, and also the impact of transfer pricing practice on the business performance on the basis of tax-related matters. As a result of the study, it can be said that transfer pricing practice has an impact on business performance in terms of both price and tax-related matters.

Download Full-text

PREREQUISITES FOR THE FORMATION OF A MECHANISM FOR STATE REGULATION OF SOCIAL GUARANTEES IN THE REGION

International scientific journal Internauka Series Economical Sciences ◽

10.25313/2520-2294-2020-9-6287 ◽

2017 ◽

Author(s):

Vitaly Lobas ◽

◽

Elena Petryaeva ◽

Keyword(s):

Social Protection ◽

Well Being ◽

State Regulation ◽

The State ◽

Point Of View ◽

Periodic Review ◽

Cost Of Living ◽

Economic Relations ◽

State Management ◽

The Cost

The article deals with modern mechanisms for managing social protection of the population by the state and the private sector. From the point of view of forms of state regulation of the sphere of social protection, system indicators usually include the state and dynamics of growth in the standard of living of the population, material goods, services and social guarantees for the poorly provided segments of the population. The main indicator among the above is the state of the consumer market, as one of the main factors in the development of the state. Priority areas of public administration with the use of various forms of social security have been identified. It should be emphasized that, despite the legislative conflicts that exist today in Ukraine, mandatory indexation of the cost of living is established, which is associated with inflation. Various scientists note that although the definition of the cost of living index has a well-established methodology, there are quite a lot of regional features in the structure of consumption. All this is due to restrictions that are included in the consumer basket of goods and different levels of socio-economic development of regions. The analysis of the establishment and periodic review of the minimum consumer budgets of the subsistence minimum and wages of the working population and the need to form state insurance funds for unforeseen circumstances is carried out. Considering in this context the levers of state management of social guarantees of the population, we drew attention to the crisis periods that are associated with the market transformation of the regional economy. In these conditions, there is a need to develop and implement new mechanisms and clusters in the system of socio-economic relations. The components of the mechanisms ofstate regulation ofsocial guarantees of the population are proposed. The deepening of market relations in the process of reforming the system of social protection of the population should be aimed at social well-being.

Download Full-text

The impact of algorithmic decision-making processes on young people’s well-being (Preprint)

10.2196/preprints.17529 ◽

2019 ◽

Author(s):

Elvira Perez Vallejos ◽

Liz Dowthwaite ◽

Helen Creswich ◽

Virginia Portillo ◽

Ansgar Koene ◽

...

Keyword(s):

Young People ◽

Significant Proportion ◽

Well Being ◽

Early Years ◽

Negative Experiences ◽

Children And Young People ◽

Online Platforms ◽

Internet Users ◽

Decision Making Processes ◽

The Impact

BACKGROUND Algorithms rule the online environments and are essential for performing data processing, filtering, personalisation and other tasks. Research has shown that children and young people make up a significant proportion of Internet users, however little attention has been given to their experiences of algorithmically-mediated online platforms, or the impact of them on their mental health and well-being. The algorithms that govern online platforms are often obfuscated by a lack of transparency in their online Terms and Conditions and user agreements. This lack of transparency speaks to the need for protecting the most vulnerable users from potential online harms. OBJECTIVE To capture young people's experiences when being online and perceived impact on their well-being. METHODS In this paper, we draw on qualitative and quantitative data from a total of 260 children and young people who took part in a ‘Youth Jury’ to bring their opinions to the forefront, elicit discussion of their experiences of using online platforms, and perceived psychosocial impact on users. RESULTS The results of the study revealed the young people’s positive as well as negative experiences of using online platforms. Benefits such as being convenient and providing entertainment and personalised search results were identified. However, the data also reveals participants’ concerns for their privacy, safety and trust when online, which can have a significant impact on their well-being. CONCLUSIONS We conclude by making recommendations that online platforms acknowledge and enact on their responsibility to protect the privacy of their young users, recognising the significant developmental milestones that this group experience during these early years, and the impact that technology may have on them. We argue that governments need to incorporate policies that require technologists and others to embed the safeguarding of users’ well-being within the core of the design of Internet products and services to improve the user experiences and psychological well-being of all, but especially those of children and young people. CLINICALTRIAL N/A

Download Full-text

Social Protection, Adolescent Well-Being, and Development in Low- and Middle-Income Countries

10.1093/oso/9780190847128.003.0016 ◽

2018 ◽

Author(s):

Suman Verma

Keyword(s):

Social Inclusion ◽

Social Protection ◽

Well Being ◽

Middle Income ◽

Developmental Potential ◽

Goods And Services ◽

Middle Income Countries ◽

Social Protection Systems ◽

Cycle Of Poverty ◽

Low And Middle Income

Effective social protection policies are crucial to realizing adolescents’ rights, ensuring their well-being, breaking the cycle of poverty and vulnerability, and helping them realize their full developmental potential. Low- and middle-income countries (LMICs) have extended social security coverage to ensure basic protections—while continuing to develop social protection systems. Social protection for LMIC adolescents in the context of gross violations of their basic rights is examined. Prevalence, consequences of protection rights violations, and the role and impact of social protection programs in ensuring enhanced opportunities for development and well-being among young people are discussed. Results demonstrate direct impacts (e.g., increased income, consumption, goods and services access; greater social inclusion; reduced household stress). LMICs need integrated social protection policy and program expansion if the 2030 Agenda for Sustainable Development is to be realized. With adolescent-centered policies and investments, governments can help adolescents realize their rights to a fulfilling and productive life.

Download Full-text