Victoria University

Oracle Customer Success

Oracle’s data science offering aids domestic violence research

Summary

Researchers at Victoria University turned to Oracle for Research to leverage Oracle Cloud to try to predict domestic violence incidents reported on social media.

Business challenges

Researchers at Victoria University were creating multiple deep learning models for predictive analysis of domestic violence. They analyze publicly available social media data spanning two years and wished to understand facial micro-expressions in videos, nuances in text, and tone and emotion in voice.

But the available tools and computational resources limited the research team’s progress. It was taking too long to process the large volume of data, analyze it, and discover results. They needed stronger support and resources to perform deeper, more powerful analysis for domestic violence research.

Sudha Subramani applied to Oracle for Research with her proposal, “Extracting Actionable Knowledge from Domestic Violence Discourse on Social Media during COVID-19,” which supported Subramani and her colleagues with a one-year Oracle Cloud credit research grant.

At Victoria University, we need a strong technology foundation so we can continue doing what we do best—providing a world-class educational experience for our students. Oracle has been our partner in making that possible and we have always been able to rely on cutting-edge technology, reliability, and security so we can focus on moving forward with courage, boldness, innovation, and agility.

Zoran Sugarevski

Executive Director, Information Technology Services, Victoria University

Why Victoria University Chose Oracle

Subramani and her colleagues Ayman Ibaida and Wenjie Ye liked that Oracle Cloud Infrastructure Data Science notebooks are easy to provision, CPUs and GPUs can be terminated without incurring billing, code persists even after notebook sessions are shut down, and users can size the block storage volume according to their needs.

The team tried other cloud products, but did not receive very strong support. They had historically worked with Oracle and knew they could continue to count on the excellent support both during and after onboarding. In fact, through Oracle for Research, Oracle assigned specialists who worked directly with the research team to transfer their research to Oracle Cloud. In the process, the researchers found that Oracle’s systems integrated well with the other tools and services they used.

Results

With Oracle Cloud Infrastructure Data Science, Victoria University’s researchers can now move their research forward in groundbreaking ways. They improved video data processing and sentiment analysis time from 15 days to 26 hours, while also reducing speech data processing time from more than 24 hours to just minutes.

The team also reduced the time needed to apply deep learning for text classification from 3 hours to 30 minutes.

In addition, the Victoria University researchers discovered that team collaboration is vastly simpler now. As part of the domestic violence research initiative, there are three initiatives led separately by three researchers.

Previously, sharing code and models was complicated because all information was stored on their local machines. With Cloud Infrastructure Data Science, models and code can easily be stored and shared on the cloud, making collaboration more streamlined.

Partners

Oracle for Research is a global community that is working to address complex problems and drive meaningful change in the world. The program provides scientists, researchers, and university innovators with high-value, cost-effective cloud technologies, participation in Oracle research user community, and access to Oracle’s technical support network. Through the program’s free cloud credits, users can leverage Oracle’s proven technology and infrastructure while keeping research-developed IP private and secure.

Oracle’s cloud and research team has not only provided us impressive support with high-performance computing resources, allowing powerful and scalable management of the computing and data processing resources, but they have also allocated experts to join our research team at different stages. Such a model of close collaboration will be exceptionally beneficial to the research and the relevant communities.

Professor Yuan Miao

Head, Information Technology Discipline, College of Engineering and Science, Victoria University