Online social networks are an important tool for communication and information exchange. Twitter is one of the most utilized social networks to propagate information or ideas. The engagement that a user has will determine how fast and far its tweet will propagate though the network. In this work, a model to quantify the influence of a tweet as a function of its replies is proposed. The model provides a tool to measure and visualize the propagation of a tweet according to its energy by relating basic physics concepts with tweet metrics. To demonstrate the effectiveness of the work, a group of public figures twitter accounts were tracked and analyzed
This project tests whether human input obtained through crowd computing can help in finding solutions to problems that are hard for computers or single people to solve, be it by reducing the convergence time of known methods, or by increasing the quality of the solutions. We use
videogames as the means to create our own crowd computing platform: by attracting people to participate, providing the interface through which they can give input, and motivating players to continue participating.
Our game is designed in such a way that, after being taught the game mechanics by example, people perform computation that is useful for our purpose simply by playing it. In particular, this game was designed to find solutions to a combinatorial optimization problem: the Robust
Facility Location Problem. We designed and implemented a visualization for this problem, a user interface for manipulating it and exploring the solution space, and a means to gather the results of a player's playthrough from their devices for analysis.
Author: Luis Perez Estrada
Social media is not only creating new opportunities for companies to interact with customers through online campaigns, but also offers businesses plentiful data that can be mined to better inform marketing initiatives and support a variety of business intelligence applications.
In this project the main objective was developing a novel method to extract and visualize actionable information from streams of social media messages, analyzed as conversational elements.
Our method has been applied to over 4 million messages related to more than 35 different events, demonstrating good results identifying conversational patterns.
Author: Dante Gama
Home visits have risen sharply at many private Medicare health plans, which treat close to 16 million elderly and disabled people under contracts with the federal government. Healthcare cost reducing programs are programs that aim to save money on treating costs of patients with diverse medical conditions, where medical staff monitors patients in different settings including primary care clinics, specialty clinics, local hospital, the patient’s home, or in other agreed upon locations.
Given that the main objective of these programs is cost reduction, the optimization of resources in all of the components is one of the main considerations. This work focuses on the optimization of the scheduling of home visits, and its effects on cost savings and patient readmission rates.
Author: Dante Gama
Seconds are critical to survivability of out-hospital Sudden Cardiac Arrestpatients since it A typically causes death if proper care is not administered rapidly. Use of public-access Automated External Defibrillators (AEDs) is showing promising results in decreasing collapse-to-shock times amongpatients, which is associated with improved patient outcomes. Bystander access to these medical devices ensure that the necessary care to victims is provided prior to arrival of emergency responders. Prior studies have suggested methods for deploying AEDs for public use by implementing mathematical optimization based on historic incidences of SCA. The purpose of this project was to improve upon these studies by developing a novel method for generating placement plans in urban environments. The novelty of this study is: (1) the use of route-based walking time instead of straight-line approximations; (2) introduction of temporal availability in deployed devices to account for location hours-of-operation; (3) use of a multi-objective optimization to balance decision-maker objectives; and (4) the implementation of an interactive decision-maker tool for observing effects on benefits and costs.
Author: Dante Gama
Bicycle sharing programs provide modern cities with an alternative mode of transportation to residents and visitors alike. As such, it is important to analyze the behavior of these programs and assess their reliability. This work presents an approach to perform a behavioral analysis and to assess the reliability of bicycle sharing programs, taking New York City's Citibike program as a case study. The approach take two errors into account: (i) the error of having excess bicycles and the user not being able to return a bicycle, and (ii) the error of not having bicycles available at for a user to take. The probability of occurrence of each error is calculated according to the behavior of each station and a certain time. This work could help decision makers plan for rebalancing and docking expansions for bicycle sharing systems.
NYC Subway :
The annual ridership for New York City’s Subway system is of more than 1.7 billion during the past year (7th in world rank). Statistics have shown an increase of ridership over the last years but a decrease in the number of trains that are on time. This work enables an understanding of NYCs subway system, creating a behavioral characterization per station and route. This understanding can help forecast the the inflows and outflows of people per periods of time and per station. This work also plans to gain understanding of repairs and micro events that can affect public transportation systems.
Services in a smart city are assessed to see if they contribute to the improvement of quality of live within a city. The evaluations are done on a technical level, but do they match the perception of the people? Do people really think they live in a smarter city and that these services make their lives better? This work explores this concept.
Author: Gabriela Gongora
Drawing on psychological theory, we created a new approach to classify negative sentiment tweets and presented a subset of unclassified tweets to humans for categorization. With these results, a tweet classification distribution was built to visualize how the tweets can fit in different categories. As a final step, we used unsupervised learning to help in the understanding of this new classification, understanding and validating the human factor. The approach developed through visualization, classification and clustering of data could be an important base to measure the efficiency of a machine classifier with psychological diagnostic criteria as the base . Nonetheless, this proposed system used to identify red flags in at risk population for further intervention, due the need to be validated through therapy with an expert.