The search query data made available to the winners has been stripped of any user information. It's obviously important to Microsoft to protect user privacy, but equally important to preserve the academic value of the data. Microsoft is encouraging the researchers to publish their findings in journals and make the information freely available to the public.

Winners included researchers from the United States, India, United Kingdom, Germany, and Australia. The proposed research spans a broad range of topics and will undoubtedly yield valuable information for the field of Search.

Abstracts for the various research projects are presented here.

• VISP: Visualizing Information Search Processes

  
  Using health related search queries as a model, these researchers at the University of Michigan will study the relationship between search engine ranking and the completeness of the information provided. In essence, the study should provide information about how relevant the search results truly are and could be used to help search engine developers improve search algorithms.

• Vinegar: Leading Indicators in Query Logs

  This group of researchers from the University of Washington are engaged in the abstract problem of how search queries relate to current events, and how they might be used to actually predict future trends or events. I find this study to be interesting from a sociological point of view. It may shed light on the ways people share information about current events using the internet.

• Entity and Relation Types in Web Search: Annotation, Indexing and Scoring Techniques

  The research being performed at IIT Bombay, India is concerned with the development of improved search algorithms. They are exploring various methods to improve the way a search engine actually works when handling complicated search tasks involving textual data.

• Deepening Search: From the Surface to the Deep Web

  At the University of Illinois at Urbana-Champaign, research is being done to make information stored in databases more readily available to users. The goal is to create a search system that directs users to information based on dynamic links, which is a part of what they refer to as the "MetaQuerier" project.

• Discovering and Using Meta-Terms

  "Meta-terms" are the focus of research being performed at the University of Massachusetts at Amherst. These so called "meta-terms" are search queries performed that are difficult to match up with content on actual web pages. The idea is to replace or expand these terms to increase the performance of the search engine. A database of these meta-term replacements will be developed using search query log analysis.

• Incorporating Trust into Web Authority

  The problem of web spam is the topic of research at Lehigh University. They are developing ways to measure the "authority" of web sites or pages presented in search results, with the desired goal of reducing or perhaps even eliminating web spam.

• Statistical Machine Learning for User Modelling

  The research topic for the group at University of Cambridge, UK is reminiscent of the work being done on the "Vinegar" project (described above). They intend to use statistical techniques to study search query data with the goal of identifying users who set trends, predict the next query or clicked page, and other interesting information.

• Combining Econometric and Text Mining Approaches for Measuring the Effect of Online Information Exchange

  A research group at New York University is investigating the economic value of text comments made by consumers after online transactions take place. It's not clear to me how this relates to search technology, but it is nevertheless an interesting idea. A system that can measure the effect of consumer information exchange might help businesses increase their profit margins by allowing them to more quickly react to negative perceptions about their products or services.

• The Truth is Out There: Aggregating Answers from Multiple Web Sources

  At Rutgers University, a research team is designing a user interface that will allow users to view search results from multiple sources. The idea here is to give users a tool that can be used to validate the information they receive from several sources. This will theoretically lead to more accurate retrieval of information. I'm not sure how their interface will expand what the major metacrawlers already offer, but presumably the interface being developed will have a more sophisticated method of ranking the quality of results returned from the various source engines.

• Predictive Exploitation of Click-Through Knowledge

  Researchers at the University of Melbourne, Australia are attempting to improve search rankings based on an analysis of the behavior of previous users that have performed similar queries. The concept is that the click-through behavior of these users can be used as a way to measure the quality of the results returned. This sounds like an excellent way for search engines to learn from the way users interact with the search results.

• Social Search: Bringing the Social Component to the Web

  At the University of Kassel, Germany, researchers are investigating open bookmark web sites such as del.icio.us. These types of sites recommend other sites based on user opinion rather than ranking algorithms, which is in contrast to the way search engines work. Nonetheless, these sites are growing in popularity and can have a significant effect on the popularity and traffic listed web sites receive.

• Mine Query/Click Log for Collaborative Internet Search

  Researchers at the University of Illinois at Urbana-Champaign are engaged in a project similar in concept to the researchers at the University of Melbourne, Australia: both groups are analyzing click-through information with the goal of improving search results by learning from past user experiences. The concept behind these projects seems very sound, and I would be surprised if search results were not improved from the study of this kind of data.
  

As I'm sure you can appreciate from a review of the above topics, a good deal of exciting research is being conducted in the field of Search! As the research results from these teams become available, I will review the results in some detail in future posts here on MarketPosition.com.

                 ← What is this?