As the informational Web continues to increase at exponential rates, tools that manage and simplify this complexity are in high demand. How else would one find out “the effect of chocolate on a person’s emotion?”; this question, in fact, is used in the study. This paper is an excellent addition to the field, describing a new visual search interface that can answer those questions in new and arguably simpler ways. The paper also lays out a foundation for how new tools should be compared to the norm, mainly Google.
The authors incorporated a study group’s flyoff between the two search capabilities, presenting statistics of how their visual digest capability could usually answer a set of 30 obscure queries, like the chocolate one, faster than traditional approaches. Their approach basically extends the normal textual search retrieval results and, through semantic-based document clustering, presents aggregated results, including select document images; their “visual interface is content or semantics oriented rather than physical link or document oriented.” By using various topic clustering approaches and combining them with the visual displays, higher and potentially more accurate results decrease the time needed to answer questions.
The second half of the paper details the user study, the 30 questions, and various statistics comparing their favorable results to the Google benchmark. Overall, this paper will help readers understand the limitations of current Web search approaches. The authors then show how document digest approaches like theirs, if done efficiently and effectively, will form the base of the next-generation visual search interfaces for Web browsing.