The goal of this project is to develop techniques to help people find their loved ones during or after a disaster.
Many people can be affected by disasters such as Hurricane Katrina. There are many Web sites available during and after a disaster, with the same goal of helping people find their loved ones, or report they are OK. For a parent who wants to find online information about his missing kid, he needs to go to many such sites to search. This process is very time consuming, especially when information from the sources keeps changing.
Another problem is that, a lot of data at the source has inconsistencies and even errors. For instance, a kid called Michael could be stored as Micheal, since the person who submitted the data mistyped the name. A person called William Smith could also be stored as Bill Smyth. The last name of a kid could be missing. For these reasons, the parent might not be able to find the information, even if the source does include valuable information about his missing kid.
We study these research issues in this project. As a prototype, we crawl and extract information from these Web sources, and put the data into a central repository. We also provide interfaces for people to submit information about missing people, or report they are OK. We allow users to search for information in the repository, instead of going to those individual sources.
Due to the low quality of the data, it is very important for the system to allow users to find persons whose description matches a search query approximately. We are developing research techniques to answer such search queries efficiently. In addition, we also need to study issues such as crawling, information extraction, and answering subscribed queries.
With the NSF award 1030002, we will study how to support powerful keyword search with efficient indexing structures and algorithms in a clouding-computing infrastructure. A main application is supporting family reunification in disasters such as the Haiti Earthquake. Check our portals for the Haiti Earthquake and Chile Earthquake. The main challenge is how to use limited programming primitives in the cloud to implement index structures and search algorithms.here. here.
We crawled around 35,000 entries from the following websites:
For any questions regarding this project please contact Rares Vernica
This project is supported by: