A Name Is Worth a Thousand Pictures: Referential Practice in Human Interactions with Internet Search Engines
Today’s Internet search engines are highly effective in returning relevant web pages to users in fractions of seconds. Yet interactions with search engines are far from trouble free. When interacting with search engines, users experience a variety of trou
- PDF / 841,260 Bytes
- 28 Pages / 439.37 x 666.14 pts Page_size
- 23 Downloads / 197 Views
A Name Is Worth a Thousand Pictures: Referential Practice in Human Interactions with Internet Search Engines Robert J. Moore
Abstract Today’s Internet search engines are highly effective in returning relevant web pages to users in fractions of seconds. Yet interactions with search engines are far from trouble free. When interacting with search engines, users experience a variety of troubles, which are still poorly understood. One particular kind of trouble stems from users’ prior knowledge about entities of interest, particularly regarding their names. This study examines how referential practice is organized in the context of search-engine interactions. It finds that, as in conversation, users employ naming in their queries to refer to entities if they can. However, when they do not know the name, or a name fails, they attempt a two-stage search: first they search for the entity name, using generic descriptions combined with image search, and second, if the name is found, they formulate subsequent queries using that name. Computer interaction analysis is used to reveal formal features of users’ referential practices from recordings of screen video with eye tracking and design recommendations for search engines are offered.
Introduction Search engines are among the top-ten most-visited websites on the Internet (Hitwise 2010). On a typical day, more than half of all Internet users accesses a search engine, and this number has been steadily increasing over the past decade (Fallows 2008). Current search engines achieve astounding performance in enabling users to find highly relevant pages in fractions of seconds. However users’ search behaviors and typical rates of success are still poorly understood (Aula et al. 2008; Hassan et al. 2010).
R.J. Moore, Ph.D. (*) Yahoo! Research, 4401 Great America Parkway, Santa Clara, CA 95054, USA e-mail: [email protected]; [email protected] A. Neustein and J.A. Markowitz (eds.), Mobile Speech and Advanced Natural Language Solutions, DOI 10.1007/978-1-4614-6018-3_10, © Springer Science+Business Media New York 2013
259
260
R.J. Moore
Users are generally on their own when deciding on the most effective strategies for searching the Internet, and they must learn through trial and error. One of the most critical parts of a web searcher’s task is the formulation of queries. The particular choice of words and symbols is highly consequential for the particular results that will be returned and for the possibility of success. The work of formulating queries includes decisions about the level of detail to include in the query, the number of words, the inclusion of definite articles, conjunctions and prepositions and more. Perhaps the most critical part of this formulation work is deciding how to refer to entities of interest. As in human conversation there are always multiple ways to refer to people, places and things. For example, “that sword,” “a scimitar,” “a curved sword” or “a sword like the guy in Indiana Jones” can all refer to the same thing. It is the task of the speaker
Data Loading...
