- 1/31 – Gabriela Zago (Portland-based independent researcher): social media, news, and social network analysis
- 2/7 – Nicole Dahmen (SOJC): visual communication
- 2/14 – Michael Dreiling (Sociology at UO): social movements, social networks, and neoliberalism
- 2/21 – Damian Radcliffe (SOJC): local journalism
- 2/28 – Charlie Butler (SOJC): long term and narrative journalism
Ontology-based Deep Learning and Applications in Healthcare and Others
January 24, 2018
Dejing Dou, Professor, Department of Computer and Information Science
Ontologies have been well used as formal specification of human knowledge in various domains. In this talk, I will first formally define ontology and explain its relationships with other research areas in AI, such as knowledge base, Semantic Web, and knowledge graph. Then I will show how to apply ontologies in deep learning, which has attracted tremendous enthusiasm in both academia and industry recently. The key breakthrough in deep learning was due to a series of improvements in artificial neural network, big data, and computational power. The design of deep learning models (i.e., representation learning) is flexible but also “flat” in certain scenarios. We argue that a hierarchical design of deep learning systems based on domain knowledge (e.g., ontologies) will gain some advantage in performance and help providing the explanation for the resulted models. I will introduce our current ontology-based deep learning (OBDL) framework and its applications in healthcare datasets and other domains. The research efforts of OBDL also contribute to the establishment of an NSF funded Center for Big Learning (CBL).
An Introduction to Research Development Services
October 25, 2017
Lynn Stearney, Director of Research Development Services
Lynn Stearney, Director of Research Development Services, a unit of the Vice President for Research & Innovation, will provide a brief overview of the services available from her office and also discuss the world of research funding from federal agencies and private foundations. She will leave plenty of time for discussion and questions.
Lynn will be joined by Tony Schaffer, Director of the Corporate Partners program at UO. Tony joined the UO this fall. His prior position was in foundation relations with NPR in Washington, D.C.
Digging for (ratings) gold: The popularity and profitability of television investigative journalism
October 18, 2017
Jesse Abdenour, Assistant Professor
Investigative journalism benefits democratic society and improves a news outlet’s credibility, but its profitability is often questioned. A content analysis of U.S. local television newscasts and an examination of audience ratings found that investigative news quality was highly predictive of viewership. This finding suggests that quality watchdog reporting can help solve the dilemma of the commercial news outlet by serving both the public good and the bottom line. Additionally, stations producing longer investigative stories and investigative stories with concealed information were associated with larger audiences. However, stations that produced greater numbers of investigative stories did not earn higher ratings, suggesting that focusing on investigative quality is a more effective news strategy compared to focusing on quantity. Building on the work of Shoemaker (1996) and others, this study proposes that the popularity of investigative journalism is attributable to evolutionary cognitive mechanisms and allusions to widespread cultural themes.
You’ve Come a Long Way, Hello Kitty: Exploring Celebrity Spokescharacters and the Match-Up Hypothesis in Cosmetics Advertising
October 11, 2017
Kim Sheehan, Professor, Honor Program Coordinator
High-end cosmetics companies use characters from childhood, such as Hello Kitty and the Little Mermaid, to sell cosmetics to adult women. While such images may tap into adults’ positive nostalgic feelings, and advertisements featuring these images may stimulate these feelings, it is unclear how well images from childhood fit with adult women’s notions of the role of cosmetics: that is, to empower women through beauty. This study investigated how adult American women viewed four different advertisements for the cosmetics company MAC featuring three celebrity spokescharacters (Miss Piggy, Hello Kitty and Wonder Woman) and a traditional beautiful model. The research showed evidence for the match-up hypothesis, that is, that better fit was related to more positive evaluations of the ad and the brand. The study also showed that celebrity spokescharacters from a woman’s childhood do not generate feelings of beauty and empowerment as well as a traditional model, and variance among the celebrity spokescharacters exists as well.