Solve the below:
Giving in lecture:
• Recall (R) = Number of relevant documents that are retrieved / Number of relevant documents in the entire database.
• Precision (P) = Number of relevant documents that are retrieved / Total number of documents retrieved (relevant +non-relevant).
• Relative Recall (RR) = Number of relevant documents that are retrieved / Number of relevant documents that are retrieved by a number of searches on the topic using different query terms.
1. A new retrieval algorithm is being evaluated using a test database, constructed for that purpose. A single search (search ‘A') of the database retrieved a total of 6,400 documents, of which 1,129 pertain to the topic being studied. Several additional searches on the topic of interest were also conducted and these additional searches plus search ‘A' retrieved a total of 7,395 distinct documents, of which 1,488 pertained to the topic of interest. The person who constructed the database tells you that there are a total of 10,500 documents in the database, of which 1,750 are pertinent to the topic of your search.
a. Given all of the above information, with respect to search ‘A' what quantity (quantities) (recall, precision, or relative recall) can be calculated?
According to search ‘A' the quantities that can be calculated in order to determine the relevance of each page are:
If the algorithm matches the results to the subject and is categorized as relevant it is given a score of two.
If the result from the database is not closely related to the subject matter but has some relevant concepts to the subject matter of the algorithm and is categorized less relevant is given a score of one.
A document that occurs but other than the above named is given a score of zero.
A document that repeats itself is given a score of zero
A search where the server do not respond is given a score of zero.
b. With respect to search ‘A', what is (are) the value(s) of that quantity (those quantities)?
The quantities are calculated as:
Precision is calculated as the documents that pertain to the topic divided by the total documents retrieved by search A. 1129/6400 =0.17640625.
Recall was calculated as the pertained documents from the addition of other searches divided by the total documents retrieved from the combined search. 1488/7395= 0.201217038
Relative recall is gotten as the total number of pertained documents from the database divided by the total number of documents from the database. 1750/10500=0.167
2. Given:
Set A = (red, orange, yellow, green, blue, violet)
Set B = (red, pink, orange, white, brown, violet)
Set C = (orange, mauve, yellow, lavender, white, green, violet)
Find:
Set D = (A OR B) AND (C NOT A).
(A OR B)= (red, orange, yellow, green, blue, violet) OR ((red, pink, orange, white, brown, violet))
Answer A OR B = (red, orange, yellow, green, blue, violet, pink, white, brown).
C NOT A = (orange, mauve, yellow, lavender, white, green, violet) NOT (red, orange, yellow, green, blue, violet)
Answer (C NOT A) = (mauve, lavender, white)
(mauve, lavender, white).
(A OR B) AND (C NOT A) = (white).
Set D= (white).