[For the full entry, please see the Oxford Internet Institute’s Summer Doctoral Programme wiki]
During her presentation to our SDP group, Wendy Hall mentioned that if two people type the same thing into Google, regardless of their purpose for the search, they will get the same results. Jim Hendler raised similar concerns when he spoke of efficiency trumping accuracy in Google search results. Certainly, if users had to wait five minutes or longer for a search engine to deliver a correct answer, they would likely not use search engines with the frequency that they currently do. That said, why can’t search engines be fast and accurate?
Through our SDP discussions, I realized that small studies, such as my dissertation research, can contribute to our understanding of why people search, what they expect, and how best to deliver this information. For example, currently, the consequences of inaccurate search results are minor and thus users are willing to modify their terms multiple times to find what they need. While delivery of search results is quite fast, how long do people spend modifying their terms until they happen upon information that answers their question, or completes their task? The human side, then, absorbs much of the work that the system could potentially provide. Understanding the cognitive process of information search and use could thus better inform the development of more accurate information delivery systems.
In addition to improving the accuracy of information delivery, understanding users’ cognitive processes could inform the depth of information provided about the data itself. Online information is currently, for the most part, opaque in terms of reporting authorship and authority, which makes determining credibility difficult. In developing the semantic web, perhaps an additional question to consider is how we can leverage its ontologies to provide more transparency in terms of who is providing the information, when it has been updated, who is linking to it, and other relevant credibility criteria/measures.