Developing a Profile-based Website for Disaster Communities

This study will build capacity in the delivery of information about oil spills, tailored to the needs of disaster communities in the Gulf region.

Background

Man typing on keyboard. Photo by devrim_pinar/Fotolia.

Photo by devrim_pinar/Fotolia

The Risk Communications subteam is building a website that will organize and explain information in a manner tailored to the needs of specific users. It will automate personalized content and formats according to users’ priorities and preferences. The team developed priorities and preferences of different types of users by conducting in-person interviews with a range of potential users, such as people working in government and non-government organizations responsible for risk management, elected officials and their staff, business leaders, faith-based organizations, and other community leaders. The interviews explored who the key decisionmakers are, what their decision problems are, and what decision information they need and use.

Creating a State-of-the-Art Website that Feels Personal

Thousands of people visit websites like Amazon.com each day, and each is welcomed into the site by a uniquely customized homepage. Why? The content is personalized so that visitors feel like the page is speaking directly to them. The Consortium for Resilient Gulf Communities (CRGC) is using state-of-the-art techniques to design a website that is personalized for people interested in learning more about how to build community resilience in the face of a major oil spill. The result should be a website that is more relevant, more engaging, and ultimately more useful for diverse end users.

What is web content personalization?

Instead of trying to appeal to a wide range of visitors with one generic home page, we can target different content to different types or groups of visitors. One visitor may be a community health professional who wants to learn more about health impacts of oil spills; they might be most interested in the findings of recent studies on depression or stress. A business owner, on the other hand, may be more concerned about government support during a disaster. A government employee responsible for hazard management planning might want to review community action plans from similar areas. Rather than requiring each visitor to take multiple steps to find the information they want, we can tailor information to meet their needs. Importantly, the tailoring can be done without fundamentally limiting the information that any user has access to.

How does content personalization work?

We can map website content to individual visitors using data available to us about the visitor in two steps:
Step One: Characterizing
Step 1 in personalizing content We will use information we have already gathered from in-depth interviews with a range of potential users from the Gulf region. These are people from a range of backgrounds involved in making decisions related to building community resilience before and after an oil spill. From the interviews we have identified types of decision makers (e.g., government workers, business owners, leaders from community-based organizations) interested in different human dimensions of oil spills (e.g., health, economic, social impacts). We have also identified keywords that capture the topics of interest to decision makers (e.g., community development plan, environmental jobs, emergency fund, community events). We can combine this knowledge from the interviews with more specific information provided by each visitor to our website via an online registration form that requests basic personal information about their background (e.g., seafood industry worker) and special interests (e.g., health impacts on workers exposed to oil).
Step Two: Recommending
Step 2 for personalizing content At this stage, we assign weights to a set of keywords, based on users’ interests, to create user profiles. Then we use a recommender system to match user profiles with items to be recommended. The system relies on rules that affect the content presented to users when their profile satisfies the rule’s conditions.

What are the benefits of content personalization?

  • Fast response: Visitors can find the information and tools they need quickly.
  • Uniquely informative: Visitors will be more highly engaged and more likely to come back to the website and recommend it to others.
  • Less distraction: Visitors won’t be frustrated with content that seems to have nothing to do with their interests.

Learn more

Web Content Personalization: A State-of-the-Art Review (See Full Report)

Creating a State-of-the-Art Website (2-Pager)

Related News

CRGC Student Featured as GoMRI Scholar

1_jacqueline_fiore_headshot-225x175Jacqueline Fiore, a Ph.D. student in Tulane University’s Economic Analysis and Policy program and a GoMRI Scholar with the Consortium for Resilient Gulf Communities (CRGC), uses applied economics to study the Deepwater Horizon oil spill’s socioeconomic impacts on Gulf fisheries, assess their ability to recover, and help inform future oil spill response.

Learn more »

Lead

Kristen Brent Venable, Ph.D.

Associate Professor and Research Scientist

Tulane University, Department of Computer Science

Kristen Brent Venable

Kristen Brent Venable is an associate professor of computer science at Tulane University and a research scientist at the Institute for Human & Machine Cognition. Her research interests are within artificial intelligence and regard, in particular, constraints, preferences, temporal reasoning and computational social choice. Her research is dedicated to providing a solid framework for the design and deployment of intelligent systems able to reason about preferences.

Co-Lead

Melissa Finucane, Ph.D.

Senior Behavioral & Social Scientist

RAND Corporation

Melissa Finucane

Melissa Finucane is a senior social and behavioral scientist at the RAND Corporation. Also a senior fellow at the East-West Center in Honolulu, Hawai'i, her interdisciplinary and policy-oriented research focuses on understanding the human dimensions of environmental health risks.