2 CONSTRUCTION OF TRAINING AND TESTING SETS‘TEST C’, WE USED THE WOR...

5.2 Construction of Training and Testing Sets

‘Test c’, we used the words in the question title part

We made use of the questions crawled from Yahoo!

as the main search query and the other words in the

Answers for the estimating models and evaluation.

information need part as search query expansion to

More specifically, we obtained 2 million questions

retrieve candidate recommended questions from Ya-

under two categories at Yahoo! Answers: ‘travel’

hoo! Answers website. We obtained an average of

154 resolved questions under ‘travel’ or ‘computer-

5

https://traloihay.net

s&internet’ category, and three assessors were in-

6

https://traloihay.net

volved in the manual judgments.

logId=LDC2006T13

7

https://traloihay.net jimmylin/resources.html

Given a question returned by a recommendation

method, two assessors are asked to label it with

‘good’ or ‘bad’. The third assessor will judge the

conflicts. The assessors are also asked to read the in-

formation need and answer parts. If a recommended

Q1: If I want a faster computer should I buy

question is considered to express the same or similar

more memory or storage space?

information need, the assessor will label it ‘good’;

InfoN If I want a faster computer should I buy

otherwise, the assessor will label it as ‘bad’.

more memory or storage space? What-

Three measures for evaluating the recommenda-

s the difference? I edit pictures and

tion performance are utilized. They are Mean Re-

videos so I need them to work quickly....

ciprocal Rank (MRR), top five prediction accura-

RQ1 Would buying 1gb memory upgrade

cy (precision@5) and top ten prediction accuracies

make my computer faster?

(precision@10) (Voorhees and Tice, 2004; Cao et

InfoN I have an inspiron B130. It has 512mb

al., 2008). In MRR the reciprocal rank of a query

memory now. I would add another 1gb

question is the multiplicative inverse of the rank of

into 2nd slot ...

the first ‘good’ recommended question. The top five

RQ2 whats the difference between memory

prediction accuracy for a query question is the num-

and hard drive space on a computer and

ber of ‘good’ recommended questions out of the top

why is...?InfoN see I am starting edit videos on my com-

five ranked questions and the top ten accuracy is cal-

puter but i am running out of space. why

culated out of the top ten ranked questions.

is so expensive to buy memory but notexternal drives? ...