Data Science Applications Using Social Network Data: Presentations
In this presentation, we’ll dive into how BuzzFeed data science quantifies and studies viral content like “#thedress” (Black and blue? White and gold?) This talk will focus on large-scale graph computations, and present the results, applications, and challenges of such analysis across multiple social media platforms using proprietary technology. Ultimately, it will provide the audience with a window into BuzzFeed’s ability to rapidly disseminate information across the world.
In the wake of a disaster, it can take weeks or months to identify, locate, and contact potential victims. In this talk, we’ll address several prospects and challenges of leveraging “open sources,” such as social media, for disaster response. We propose an approach for fusing multiple open sources into a coherent social graph to enable efficient data discovery, increased situation awareness, and application of analytics for performing inference and prioritization.
As social networking ad products have become prominent, the need to prove their efficacy has risen in importance. We utilize a data science methodology to determine the offline impact of an online sales travel campaign. We explain how to use Pig and R to process geo-location information in Hadoop and determine whether a user was motivated to pursue action offline after online campaign exposure, ultimately finding a statistically significant impact.
From romantic relationships to career advancements, data science is rapidly releasing new insights about society. A critical piece of arriving at these conclusions is differentiating between growth of data and growth within society. By controlling for exponential data growth we can reveal the linear growth in society. Here we cover methodologies for handling growth in LinkedIn's data with two case studies: industry migration and the rising glass ceiling.