Question

You want to improve your search engine by comparing to an competitor. In your database you have over 30million queries and your target is to select only 5 thousands for evaluation. How would you sample? Note that we are not asking for implementation of the sampling, but are asking for the design.

Answer 1

Given :- Database have 30 million Queries

Target:- To select only 5 thousand for evaluation for checking;
The effectiveness of retreival of information of our search engine which is;
with our competitior.

Search engines are one of the important software in the field of internet today.

Everyone is looking for fast and relevant results for their search queries.Retreival
effectiveness is one of the most important feature of evaluation of search engines.Most od the tests for this feature does not take into account the actual user interaction during the search process instead they are based on query response paradigm.Selection of queries is the key part of this kind of evaluation.

Following categories of queries will make a good combination:

• Information queries: These are the kind of queries where user looks for some documents or information on some topic.This is the most common kind of queries.
• Navigational Queries: Here user is somewhat aware of the location of the informaton and tries to follow a particular path form that location.Like going to the home page and then exploring things
• Trabaction query - Here user is looking at some place where further transaction like downloading software etc can be done and he is moving from page to page.

A good mix of such queries can give a better picture of the performance of out search engine.