The "Real World" Web Search Problem

author: Eric Glover,
published: Dec. 3, 2007,   recorded: September 2007,   views: 5927


Related Open Educational Resources

Related content

Report a problem or upload files

If you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.
Lecture popularity: You need to login to cast your vote.


There are numerous papers which present methods to address web-search related challenges such as relevance and ranking, query processing, and classi cation. Unfortunately, many of these methods are ine ective in a large-scale commer- cial setting, despite statistically signi cant experimental results. To help bridge this gap between academic and commercial settings, this lecture examines the components of large-scale commercial search engines, then proposes ve classes of problems encountered by researchers in this area - biases; bad or di erent assumptions about statistics, users, queries or web contents; insucient or miss- ing data; inconsistencies related to evaluations and objectives; and policies or external factors, including resource limitations. Using real stories and personal experiences, the lecture illustrates examples of these problems, along with a few proposed approaches to deal with or reduce their consequences or e ects. In addition to the classes of problems, there are several fundamental prop- erties of the web that are often not considered suciently when performing experiments or de ning problems, resulting in unrealistic experiments or ob- jectives. Even within a search engine, overlooking key properties such as the non-stationarity of the users and the web, can result in ine ective evaluations, and may even lead to failed subsystems. Fortunately, very simple approaches can often be highly e ective. This lec- ture helps put context on how commercial search engines work, what problems they face, what e ective solutions require, and how evaluations and problem de nitions could be changed to more e ectively predict success in a commercial setting - while still retaining interest of researchers.

See Also:

Download slides icon Download slides: mmdss07_glover_trw_01.pdf (293.1┬áKB)

Help icon Streaming Video Help

Link this page

Would you like to put a link to this lecture on your homepage?
Go ahead! Copy the HTML snippet !

Write your own review or comment:

make sure you have javascript enabled or clear this field: