Social Computing meets every other week to share knowledge and be productive in a friendly co-working environment. Like other 'hacking' groups on campus, we focus on practical skills, emerging technologies, and implementations of good ideas. Our focus is on the "social", in two senses: the social sciences, meaning we work on socially generated data (whether it be social networks, demographic information, text, or anything else), and the conviviality of our own group.
We host presentations of computational social scientific work in progress. We ask folks to present: What am I working on? What have I done, and how did I do it? What are my next steps and where do I need help? We balance tutorials, lightning talks, free-flowing discussion, and coworking.
** NEXT MEETING **
5/14, Daniel Aranki, a PhD student in EECS, will give a talk on Private Disclosure of Information.
We present a novel framework, called Private Disclosure of Information (PDI), which is aimed to prevent an adversary from inferring certain sensitive information about subjects using the data that they disclosed during communication with an intended recipient. We show cases where it is possible to achieve perfect privacy regardless of the adversary's auxiliary knowledge while preserving full utility of the information to the intended recipient and provide sufficient conditions for such cases. We also demonstrate the applicability of PDI on a real-world data set that simulates a health tele-monitoring scenario.
In the second part of the talk, we present a general-purpose framework for health tele-monitoring, which is the motivation behind PDI. We start by briefly presenting the results from a pilot study that featured 15 patients with heart failure from which we learned lessons towards the design of the general-purpose framework. We then present the features and components of the framework as they relate to the lessons learned from the aforementioned pilot study, and close by describing directions for future work.
** PREVIOUS MEETINGS **
5/1 Quantifying academic research: A case study. -- Till Bergmann
In this talk, I will present some preliminary research on a meta-analysis of different subfields of cognitive science. One part of the analysis looks at thematic overlap between different journals to deduce how similar subfields are, while a different part focuses on collaboration and citation networks. Apart from presenting some interesting tendencies I have discovered so far, I hope we can brainstorm about a) potential measures to be applied and b) what these measures mean for the field of cognitive science (or any other field).
4/17 Causal Inference in Networks
Seb Benthall will be leading a discussion of causal inference in networked data. We will discuss applications and limitations of the potential outcomes framework in networked data sets for particular research problems. In particular, we will inquire into how to think rigorously about causal mechanisms involving emergent properties of networks.
3/13 Double Feature:
3:30-4:30 - Collecting and displaying data with with the Web (D. Clark)
and
4:30-5:30 - nframe: experiments in text featurization (S. Benthall)
In “Collecting and displaying data with with the Web,” Dav Clark will provide an overview of web architecture for social scientists. He will focus on the challenges of coordinating two computational environments - the server and the browser. Topics include static and dynamic web servers, HTML, CSS, and JavaScript, including an explanation of what some of the major JS frameworks do. This is a great opportunity to find some colleagues to learn with!
In “nframe: experiments in text featurization,” Sebastian Benthall will present recent work-in-progress on nframe, an implementation of an unconventional approach to text featurization. Designed for analyzing conceptual change in large amounts of communications data, nframe tries to walk a middle path between n-gram ‘bag of words’ text models and the full grammatical parsing used in feature extraction. If Seb can get his act together in time, this talk may include an interactive demo.
~~~~~~
2/27 -- Life cycle of a text analysis project
Rochelle Terman will present on the life cycle of a text analysis project in the context of her dissertation on human rights shaming. She will discuss web scraping, preprocessing, and visualization, and cover myriad techniques such as topic modeling, clustering, and named entity recognition. (2 hours)
~~~~~~
2/13 Double feature: Network analysis and visualization AND Managing your research like a boss:
Brooks Ambrose will discuss his experiences studying network clustering, cohesion, and visualization. He will present his work in progress working with citation network data in R. He asks: can I parallelize my code if I use I Python? With what tools should I implement my novel ideas for visualizing network data on the web? (30 minutes)
Discussion and coworking (30 minutes)
Sebastian Benthall will discuss "Managing your research project like a boss." He will present how he has organized the technical part of his dissertation as an open source project, BigBang, and how that has opened up opportunities for collaboration with other grad students and attracted undergraduate apprentices. Will present strategies for project management with GitHub and project promotion. We'll also discuss the challenges of adapting software project management best practices to an academic setting. (30 minutes)
Discussion and coworking (30 minutes)
~~~~~~
