Please post your solutions to the Sep 19 In-Class Exercise to this thread. Chris

-- Sep 19 In-Class Exercise Thread

You might enter "San Jose Earthquakes" into a search engine to learn about earthquakes in San Jose, California. Precision on Google was 2/10. One way to estimate relevance is to take the average of the number of relevant results on each page (assuming a paginated result set).

-- Sep 19 In-Class Exercise Thread

* San Jose, CA soccer team * Earthquake natural disaster information in San Jose, CA * Earthquake natural disaster information in San Jose, Costa Rica * Earthquakes experienced by St. Joseph * Earthquakes conjured by St. Joseph * Some random super hero/villain's finishing move called "SAN JOSE EARTHQUAKE!" P@10 = 0.9 from Google for info about soccer team |Rel| could be estimated by considering the popularity and history of your measure of relevance for the query. More popularity and history around your definition of relevancy for the query (i.e. earthquakes conjured by St. Joseph) would lead to more pages and more relevant results, which increases |Rel|.

-- Sep 19 In-Class Exercise Thread

1. To check if earthquake there is any recent geological/ seismic activity/earthquake in my current location - San Jose, CA (other scenarios, non-CA san jose earthquakes" 2. used Bing/yahoo - top 3/10 results(8th, 9th, 10th result) matches my scenario - P= 0.3. All other results are for san jose soccer team called 'earthquakes' 3. Rel can be calculated based on (1) current location settings in browser, (2) if all terms are in doc, (3)Though the term is 'earthquake' I get results for 'earthquakes' (plural). hence relevance should be given low

-- Sep 19 In-Class Exercise Thread

Why I entered the query: To find the most recent earthquake in San Jose region Search engine - Bing P@10 = 1/10 To estimate |REL|, I could enter a very specific query (eg: the details of the most recent earthquake in San Jose) and taking this count

-- Sep 19 In-Class Exercise Thread

 = 34600000 * 0.1 = 3460000

//Yaoyan Xi San Jose is prone to Earthquake and I want to know the chance or timing of next Earthquake. I used google which returned 34.6 million results. Only one among the first 10 met the above rationale. P@10 = |Res and Rel|/|Res| = 1/10 = 0.1 To estimate Rel, I used the percent of relevant results on the first page ( optimistic approach ) Rel = total returned results * (% of relevant result of first page ) = 34600000 * 0.1 = 3460000

-- Sep 19 In-Class Exercise Thread

	Relevant results - 6
	P@10 = 0.6

The reason for query being searched might be a recent occurrence of earthquake in San Jose, or to find out the history of earthquakes in San Jose, California. The search engine I used was ‘duckduckgo’ For top 10 results: Relevant results - 6 P@10 = 0.6 We can estimate |rel| by calculating the relevant results on first n pages and then multiply the result with the number of total pages.

-- Sep 19 In-Class Exercise Thread

 Google
 Info on soccer was P@10 = 0.8
 Info on actual earthquakes was P@10 = 0.2

One would search for this to either research recent earthquakes in the SJ area or follow news of the soccer team. Google Info on soccer was P@10 = 0.8 Info on actual earthquakes was P@10 = 0.2 Relevance can be judged whether or not the words 'soccer', 'team', or 'sport' appear on the document

-- Sep 19 In-Class Exercise Thread

Reason for query: To look up recent earthquakes in San Jose. Search engine: Google P@10: 1/10 Estimate |Rel|: Google returns a total of about 42,900,000 results. Multiply P@10 with this for an estimate of the size of Rel.

-- Sep 19 In-Class Exercise Thread

Query - San Jose Earthquake Reason - To find out how prone San Jose is to Earthquake ? - Bing P@10 = 5/11 = 0.454 |Rel| - Amount of click on the link that needs to be done in order to get desired answer for the query.