NMSU branding

New Mexico State University

New Mexico State University

News Center

Doctoral student developing system for assessing reliability of Web info

Have you ever wished your computer could judge the reliability of the information it serves up when you do a Web search?

Melanie Martin, NMSU graduate student in computer science, is developing a system for evaluating the reliability of Web information. (NMSU photo by Darren Phillips)

Martin is working on it. The computer science doctoral student at New Mexico State University is developing an interactive software system that will enable someone browsing the World Wide Web to determine if the information he or she finds there is reliable and verifiable.

The phenomenal growth of the Web has brought not just an explosion of information but also an explosion of false, misleading and unsupported information, Martin noted.

"It's amazing how much garbage is out there, and it can be dangerous," she said.

A 1997 study looked at Web sites that describe what to do when a child has a fever, she said, and found that only four out of 41 sites "had accurate information that adhered closely to published guidelines."

Martin's goal is to create a system that will be useful to a wide range of users, from the parent whose child has a fever to the scholar doing extensive professional research. Using advanced statistical techniques, it will provide a measure of the reliability of a Web page's content and cluster pages according to their agreement or disagreement with each other on the topic in question. It also will provide tools for processing queries and retrieving relevant documents.

Martin's project caught the attention of the Association for Computing Machinery's Special Interest Group in Information Retrieval (ACM SIGIR), which selected her to present a research paper and participate in a doctoral consortium at the group's international conference in Sheffield, England, July 25-29. She is one of only nine doctoral students from around the world chosen for the consortium, during which the students will discuss their work with 10 senior researchers in the field.

The ACM is the leading international association for computer science.

Martin's project draws on her background as a mathematician as well as her current research interest in the areas of information retrieval, computational linguistics and information science. With a master's degree in mathematics from the University of Oregon, she was pursuing a mathematics Ph.D. at NMSU when she decided to switch to computer science.

"In mathematics you have definitions and absolutes," she said. "It's like the search for truth and beauty. When you work with natural language, things get messy and ambiguous. I wanted to bring some of my understanding of math into the study of language."

She expects to have at least "a good first pass" on her research project completed by December, when she is scheduled to finish her doctoral degree.

Martin's system will utilize advanced statistical methods, algorithms and computer power to retrieve documents from the Web and analyze their content. But until such systems are available to consumers, she can offer some expert advice on smart Web searching.

When looking at a Web page critically, she said, ask some key questions: "Is there an author present? Can you tell when the page was last updated? Why was the page published? What was the author's motivation? Does it seem slanted or biased to you?"

The popular Google search engine has a feature that allows a user to find out what other Web pages are linked to the page the user is looking at, Martin noted.

"By linking to a page you are sort of validating it," she said. "Looking at who is linking to a page can give you some idea of its credibility."

Library Web sites are a good source of expert help, she added. The NMSU Library Web site, for example, offers a tutorial program called "The Good, the Bad and the Ugly, or Why It's a Good Idea to Evaluate Web Sources." It can be found under "Instruction" on the library's home page, http://lib.nmsu.edu/.