When: Wednesday, November 7, 2018
Where: PGH 563
Time: 11:00 AM
Detecting Nastiness in Social Media
Speaker: Niloofar Safi
Although social media has made it easy for people to connect on a virtually unlimited basis, it has also opened doors to people who misuse it to undermine, harass, humiliate, threaten and bully others. There is a lack of adequate resources to detect and hinder its occurrence. In this research, we present our initial NLP approach to detect invective posts as a first step to eventually detect and deter cyberbullying. We crawl data containing profanities and then determine whether or not it contains invective. Annotations on this data are improved iteratively by in-lab annotations and crowdsourcing. We pursue different NLP approaches to distinguish the use of swear words in a neutral way from those instances in which they are used in an insulting way. We also show that this model not only works for our dataset, but also can be successfully applied to different datasets.
Niloofar Safi Samghabadi received her Bachelor's degree in Computer Science from Sharif University of Technology in 2014. Currently, she is a fourth-year PhD Student at the University of Houston, working with Dr. Thamar Solorio. Her research focuses on Natural Language Processing and Applied Machine Learning.
Improved Probabilistic Guarantees for Influence Maximization
Speaker: Nguyen Dinh Pham
The standard Influence maximization problem involves choosing a seed set of a given size, which maximizes the expected influence. However, such solutions might have a significant probability of achieving low influence, which might not be suitable in many applications. In this paper, we consider a different approach: find a seed set which maximizes the influence set size, which can be achieved with a given probability. We show that this objective is not submodular, and design two algorithms for this problem, one of which gives rigorous approximation bounds. We evaluate our algorithms on multiple datasets, and show that they have similar or better performance as the ones optimizing the expected influence, but with additional guarantees on the probability. Keywords: social network influence, influence maximization, multi-criteria approximation, Monte-Carlo sampling.
Nguyen Dinh Pham is a fifth year PhD student in the Department of Computer Science, working with Dr. Gopal Pandurangan. His main research topic is Graph influence optimization, and approximation with theoritical guarantees. The side topics, which is the theme in Dr. Pandurangan lab, is theoritical distributed algorithms. His other interests are functional programming and practical asynchronous distributed system.
An Adaptive Approach for Demand-Response and Latency Control in Distributed Web Services
Speaker: Gandhimathi Velusamy
In this ever-connected digital era, the ubiquitous dependence on the internet mandates massive number of servers to be deployed at Data Centers across the globe. In addition, the increasing trend in Cloud adaptation by the businesses increase the usage of servers at the Data Centers of the Cloud service providers such as Amazon, Google, etc. The Data Centers are the large consumers of electricity and impacts the temporal variations in electricity prices in the Demand Response program due to its volatile workloads. The Peak-based pricing method further increases the electricity bills for the maximum power utilization at some places. However, the Demand-Response program helps the large consumers to regulate their usage patterns by knowing the Real Time Prices through the two-way communications offered by the Smart Grid. We propose an energy-cost and peak-usage conscious request re-routing method to distribute the web-based workloads to servers in an Internet Data Center system to reduce the energy expenses for the Cloud operators. Our experimental results from a Cloud testbed have proved with improved performance over the existing methods.
Gandhimathi Velusamy is a fourth year PhD student in the Department of Computer Science, at University of Houston. She is being advised by Dr. Ricardo Lent and Co-advised by Dr. Jaspal Subhlok. Her research focuses on the design, the analysis and the experimental evaluation of learning algorithms applied to load balancing with energy and latency considerations. Her interests include Software Defined Networking and Machine Learning.