UIUC大学之Coursera课程Text Retrieval and Search Engines：Week 3 Quiz

Week 3 QuizHelp Center

Warning: The hard deadline has passed. You can attempt it, but you will not get credit for it. You are welcome to try it as a learning exercise.

In accordance with the Coursera Honor Code, I  certify that the answers here are my own work.

Question 1

Assume you are using a unigram language model to calculate the probabilities of phrases. Then, the probabilities of generating the phrases “study text mining” and “text mining study” are  not equal, i.e., P(“study text mining”)  ≠  P(“text mining study”).

False

True

Question 2

You are given a vocabulary composed of only four words: “the”, “computer”, “science”, and “technology”. Below are the probabilities of three of these four words given by a unigram language model. 

Word	Probability
the	0.4
computer	0.2
science	0.3

What is the probability of generating the phrase “the technology” using this unigram language model?

0.5

0.04

0.0024

0.1

Question 3

You are given the query Q= “online courses” and two documents:
D1 = “online courses search engine”
D2 = “online education is affordable”
Assume you are using the maximum likelihood estimator  without smoothing to calculate the probabilities of words in documents (i.e., the estimated  p(w|D)  is the relative frequency of word  w  in the document  D ). Based on the unigram query likelihood model, which of the following choices is correct?

P(Q|D1) = 1/16       P(Q|D2) = 0

P(Q|D1) = 0            P(Q|D2) = 1/4

P(Q|D1) = 1/16       P(Q|D2) = 1/4

P(Q|D1) = 1/2         P(Q|D2) = 1/2

Question 4

Assume the same scenario as in Question 3, but using linear interpolation (Jelinek-Mercer) smoothing with  λ=0.5 . Furthermore, you are given the following probabilities of  some of the words in the collection language model:

Word	P(w|C)
online	1/4
courses	1/4
education	1/8

Based on the unigram query likelihood model, which of the following choices is correct?

P(Q|D1) = 1/16       P(Q|D2) = 1/16

P(Q|D1) = 1/16       P(Q|D2) = 1/32

P(Q|D1) = 1/32       P(Q|D2) = 1/32

P(Q|D1) = 1/16       P(Q|D2) = 0

Question 5

The BM25 has more free parameters to tune than the ranking function of the Dirichlet Prior smoothing.

True

False

Question 6

Assume you are using Dirichlet Prior smoothing to estimate the probabilities of words in a certain document. What happens to the smoothed probability of the word when the parameter  μ  is  increased?

It becomes closer to the probability of the word in the collection language model

It becomes closer to the maximum likelihood estimate of the probability derived from the document

It does not change

It tends to 1

Question 7

It is possible that pseudo feedback decreases the precision and recall of a certain retrieval system.

True

False

Question 8

Refer to the Rocchio feedback formula in the slides. If you want to eliminate the effect of  non-relevant documents when doing feedback, which of the following parameters must be set to zero?

γ  and  β

γ

α

β

Question 9

Let  q  be the original query vector,  DR={P1,...,Pn}  be the set of positive document vectors, and  DN={N1,...,Nm}  be the set of negative document vectors. Let  q1  be the expanded query vector after applying Rocchio on  DR  and  DN  with positive parameter values  α ,  β , and  γ . Let  q2  be the expanded query vector after applying Rocchio on  DR  and  DN  with the same values for α ,  β , but  γ  being set to zero. Which of the following is correct?

q2  can have greater or equal weights to  q1

q1  has strictly greater weights than  q2

q2  has strictly greater weights than  q1

q1  can have greater or equal weights to  q2

Question 10

Which of the following is  not true about the KL-divergence retrieval model?

It cannot be computed as efficiently as the query likelihood model.

It represents both queries and documents as language models.

It supports relevance feedback.