1. google queries reverse engineering, diff & deltas for search engine result sets

    March 25, 2008 by pierre

    This morning the sentence “I wonder what he googled to find this idea” made Jérôme and I think of a game: given a page of search results, find the closest list of keywords (the “google query”) that leads to it. Today the biggest search engines are based on indexation of content using keywords*: being able to find the correct keywords or combination of keywords that leads you to relevant information on a topic you do not know yet is a quite important skill. This game could be a way to exercise this skill. To program this as an online game would be interesting. You could either play it with two players, one player set the query the other one get the results and have to guess the query, or it could be played with a single player and a programs generates a random query from a list of words and push the results to the player for him to guess the query.

    Now if the person who guess the query picks a not exactly identical but similar query, we would like to tell him: “not exactly but close!”, and “how close”. This could be based on finding that the words in the original query and the words in the guess are synonyms, but more interestingly it could be based on calculating the difference, distance, or delta between the two results sets that the original and the guessed query yields. Is there any code that compares search results and compute deltas? In what other situations this kind of calculus could be useful?

    *there is a demo of an alternative way of searching on the company “the knowledge” homepage. This technology for sure is interesting, but since I saw it I wonder: what kind of non trivial questions would I ask there? and what would be the success rate of the answers then? hu…