Fall 2021

NVISION: Network Visualization Interventions Supporting Interpretation of Objective News

“The current fractured media landscape allows individuals to choose confirming over credible information, and information spreads quicker online than interventions like fact-checking. Misinformation can be debiased by identifying gaps in mental representations of the world (mental models) and prompt alerts to be vigilant about assessing information (Lewandowsky et al, 2012).

We aim to develop interventions to make media-balance salient to users to mitigate the spread of misinformation. Social sampling theory describes that our misperceptions of others is explained by the sample of people we encounter (Galesic, Olsson, and Rieskamp, 2012), and we are more likely to link to similar people online (Kossinets & Watts, 2009). Our interventions address limitations in individual views of the media landscape. We aim to attach visualizations of sharing network to news articles in real-time to make these biases explicit.”

NVISION: Network Visualization Interventions Supporting Interpretation of Objective News
Object detection and classification APIs for urban street image analysis
Developing models to analyze images is a demanding task that requires significant time, resources, and effort. Recently, companies such as Amazon and Google are providing services to make the modeling process easier so even users with little machine learning expertise can enjoy deep learning technologies. Based on our prior work in object detection and classification for smart city applications, we would like to compare and evaluate the process and performance of commercial services using our training datasets. This project will be a good practice to understand the image machine learning modeling process and the advantages/limitations of commercial services for customized learning.
Object detection and classification APIs for urban street image analysis
Tracking health and nutrition signals from social media data (begun Spring 2020)

Food environments (the physical spaces where people acquire and consume food) can profoundly impact diet and related diseases. Effective, robust measures of food environment nutritional quality are required by researchers and policymakers investigating their effects on individual dietary behavior and designing targeted public health interventions. The most commonly used indicators of food environment nutritional quality are limited to measuring the binary presence or absence of entire categories of food outlet type, such as ‘fast-food’ outlets, which can range from burger joints to salad chains. There would be great value in a summarizing indicator of restaurant nutritional quality that exists along a continuum, and which can be applied at the scale of large food environments, for example across Los Angeles County, to make distinctions between diverse restaurants within and across categories of food outlets.

This project will explore the ability to track real-life health and nutrition signals from social media data, focusing on data from Foursquare and Yelp. We will investigate the ability to access menu information from the APIs of these social media platforms, and develop measures to assess the nutritional content of these menus. Multiple aims will be investigated in this project, including scraping data from social media; NLP of menu text, tag, and comment data; developing predictive models of obesity; and more. “Ground truth” data on dietary patterns of LA residents will be available, enabling validation of dietary measures and predictive models built from menu data.

Transfer learning for adversarial machine translation
Neural Machine Translation (NMT) is the process of mapping a segment of words from a source language to a target language using neural networks. However, NMT systems rely on large datasets for the source and target languages, and perform poorly on low-resource languages where there is insufficient parallel data. An effective method for improving NMT on low-resource languages is to employ transfer learning, where a model trained on a high-resource language pair is used to initialize training for the low-resource language pair. In this work, we will study the effect of employing transfer learning methods on an adversarial machine translation models based on Long Short-Term Memory Recurrent Neural Networks (LSTM).