We investigated the differential diffusion of all of the verified true and false news stories distributed on Twitter from 2006 to 2017. The data comprise ~126,000 stories tweeted by ~3 million people more than 4.5 million times. We classified news as true or false using information from six independent fact-checking organizations that exhibited 95 to 98% agreement on the classifications. Falsehood diffused significantly farther, faster, deeper, and more broadly than the truth in all categories of information, and the effects were more pronounced for false political news than for false news about terrorism, natural disasters, science, urban legends, or financial information. We found that false news was more novel than true news, which suggests that people were more likely to share novel information. Whereas false stories inspired fear, disgust, and surprise in replies, true stories inspired anticipation, sadness, joy, and trust. Contrary to conventional wisdom, robots accelerated the spread of true and false news at the same rate, implying that false news spreads more than the truth because humans, not robots, are more likely to spread it.

The spread of true and false news online

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

Social Network Analysis

The paper “Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification” by Joy Buolamwini and Timnit Gebru, that will be presented at the Conference on Fairness, Accountability, and Transparency (FAT*) in February 2018, evaluates three commercial API-based classifiers of gender from facial images, including IBM Watson Visual Recognition. The study finds these services to have recognition capabilities that are not balanced over genders and skin tones [1]. In particular, the authors show that the highest error involves images of dark-skinned women, while the most accurate result is for light-skinned men.

/pdf/gender-shades-intersectional-accuracy-disparities-in-4qgeu0c1i3.pdf

Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification

With the widespread use of AI systems and applications in our everyday lives, it is important to take fairness issues into consideration while designing and engineering these types of systems. Such systems can be used in many sensitive environments to make important and life-changing decisions; thus, it is crucial to ensure that the decisions do not reflect discriminatory behavior toward certain groups or populations. We have recently seen work in machine learning, natural language processing, and deep learning that addresses such challenges in different subdomains. With the commercialization of these systems, researchers are becoming aware of the biases that these applications can contain and have attempted to address them. In this survey we investigated different real-world applications that have shown biases in various ways, and we listed different sources of biases that can affect AI applications. We then created a taxonomy for fairness definitions that machine learning researchers have defined in order to avoid the existing bias in AI systems. In addition to that, we examined different domains and subdomains in AI showing what researchers have observed with regard to unfair outcomes in the state-of-the-art methods and how they have tried to address them. There are still many future directions and solutions that can be taken to mitigate the problem of bias in AI systems. We are hoping that this survey will motivate researchers to tackle these issues in the near future by observing existing work in their respective fields.

A Survey on Bias and Fairness in Machine Learning

Data validation is one of the most important and, possibly, most under-valued task in an organization. Without clean data, an organization cannot employ sophisticated analysis and optimization tools to strive for excellence in operations, delivery or planning. Organizations have started to realize the value of data and its impact on their efficiency. Typically, they either develop in-house solutions or purchase industry standard solutions. In this work, we propose an alternative of data validation as a service offering. We argue that such a service would be a profitable proposition for both the parties, provider as well as consumer. We present a general framework to enable such an offering. We provide details on one such implementation that we carried to showcase the viability of such an approach. We propose multiple variants of the offering to handle privacy concerns of the consumer. Finally, we present a set of initial results comparing the different variants.

Towards Providing Data Validation as a Service

In today's services driven economic environment, it is imperative for organizations to provide better quality service experience to differentiate and grow their business. Customer satisfaction (C-SAT) is the key driver for retention and growth in Retail Banking. Wait time, the time spent by a customer at the branch before getting serviced, contributes significantly to C-SAT. Due to high footfall, it is improbable to improve the wait time of every customer walking in the branch. Therefore, banks in developing countries are strategically looking to segment its customers and services and offer differentiated QoS based service delivery. In this work, we present a system for customer segmentation, and scheduling based on historic value of the customer and characteristics of current service request. We describe the system and give mathematical formulation of the scheduling problem and the associated heuristics. We present results and experience of deployment of this solution in multiple branches of a leading bank in India.

A system for providing differentiated QoS in retail banking

Providing the history and context(s) of a news article that emerges in the middle of an evolving news story--sometimes multiple news stories--is a complex task. The complexity of the task is compounded by the fact that different users are interested in different contexts of the article, and it is impossible to guess what a particular user is most interested in. In this paper, we introduce ESTHETE, a system that provides rich context(s) (through what we call personalized flexible context extraction), by preprocessing and storing articles in a structured representation (directed graphs) that makes it easy for the user to explore different contexts. The advantage of this approach is that the incremental computational expense in incorporating new articles as they are published is minimal. Our system is available at: http://konfrap.com/esthete.

ESTHETE: a news browsing system to visualize the context and evolution of news stories

In this paper we present a model for delivering differentiated quality of service based on the value for the provider. The value for the provider is captured by taking into account the i) client value, ii) service value, and certain exogenous factors. We enumerate and articulate different dimensions which will affect these components. We present a couple of case studies to demonstrate the usefulness of the proposed model. The first case study is from Retail Banking based on our real life experience of deploying a service delivery system in a leading bank in India. The second case study is from a Business Process Outsourcing industry. We also define the notion of value plot to visualize the relative importance of customers from value perspective and highlight how the value plot can be used for both operational as well as strategic planning in service delivery.

Towards Providing Value Based Differentiated QoS

Service Interaction Networks [7] are abstractions used in modeling and analyzing people-centric aspects of service delivery. They pose interesting challenges from modeling, algorithmicand knowledge mining viewpoints. Typically, the data needed for the construction of serviceinteraction networks is sensitive in nature and organizations may not readily share the data.Therefore, it is imperative that we build realistic simulations of service interaction networks soas to enable not only the development of novel analyses, but also to validate their applicability.Lack of benchmark datasets to compare against poses challnges in validating simulations. Wepresent a simulation tool designed to take into account the people-centric eﬀects of servicedelivery and an interesting approach to validate it. We hightlight some novel analyses of serviceinteraction networks that require new modeling and algorithmic insights. 1 Introduction One of the distinguishing features of the services industry is the high emphasis on people interactingwith each other and serving customers rather than transforming physical goods in the process.This aspect of service delivery results in several unique characteristics of services as opposed toother industrial settings. Naturally, modeling and analyzing services and their delivery, taking intoaccount people-centric aspects is an important research direction.In a typical service delivery scenario, an organization consisting of a set of trained actors receivesa time stamped stream of service requests. Each request is serviced by a subset of the actors as perits requirements. It is natural to view this process in terms of an abstract network. The notion ofService Interaction Networks was introduced by Kameshwaran et al. [7] to help model and analyzewide range of problems that arise in such a setting. However, several reasons (elaborated later) makeit diﬃcult to obtain the necessary real-life data, especially for the purpose of publicly disseminatedresearch. In this paper, we present a simulation platform that captures various people-centric issuesofservice delivery and can be used to generate instances that are likely toreﬂect real-life observations.We present some example analyses of service interaction networks that have interesting applicationsin service delivery.The central abstract construct in service delivery scenarios is the so-called process model. Itconsists of many atomic work units (specialized and well-speciﬁed). The process model connects thevarious diﬀerent atomic units associated with the process so as to reﬂect the logical and precedenceconstraints, and communication requirements. Typically, skilled individuals or a team possessingan appropriate combination of skills is assigned to an atomic unit and we call them as agents oractors. As we conceive of the service requests as tasks to be performed, throughout the paper,we will refer to the process model of the request as task-graph. Abstractly, it consists of nodesthat represent speciﬁc, atomic work units that can be assigned to actors and edges that captureinteraction requirements between the actors assigned to the corresponding work units. In general,the edges could connect multiple work units, i.e, they could be hyperedges. A request, representedby its task-graph is executed by assigning actors to each of the nodes. This assignment takes intoaccount required skill sets, capacity and other speciﬁcations of the tasks. If we consider the set of allthe actors in the system and the interactions between them during the course of executing a largenumber of service requests, we get networks that capture the overall interaction patterns amongthem. Each request also has an overall outcome associated with it. Combinging the outcomeswith the interaction patterns gives rise to an enriched Service Interaction Network [7] that couldpotentially reveal critical insights on the service delivery.Ideally, instead of monitoring request level outcomes, one would like to further micro-monitorthe execution of task-graphs. But, there are several issues that make micro-monitoring diﬃcult.Chief among them is that a subset of the actors (functionalities in other words) assigned to therequests could be independent service vecndors (ISVs) whose actions may not be monitored. Evenwithin an organization, such micro-monitoring could result in issues related to morale, trust betweencolleagues, and unfair competitions etc. Therefore, in most scenarios, only high-level monitoring,such as measuring the overall eﬀectiveness of a request resolution, is carried out. We now discusssome scenarios in which notion of service interaction networks is helpful.

Sameep Mehta

Papers

Towards Providing Data Validation as a Service

A system for providing differentiated QoS in retail banking

ESTHETE: a news browsing system to visualize the context and evolution of news stories

Towards Providing Value Based Differentiated QoS

Service Interaction Networks: Simulation and Applications