Each line has a guess (page) and a correct answer (answer) given some fraction of the question revealed so far (text). Your goal is to predict whether they match. Each guess is the title of a Wikipedia page. To get you started, you have the following columns

row A unique ID for every guess

body_score A measure of the confidence for how well the text of the Wikipedia page matches the question text

page The guess produced

answer The correct answer

text The text so far revealed (randomly chosen)

category The category (could be noisy) of the question

tournaments The tournaments the question was used in

answer_type The kind of answer of the question (e.g. person, place, work)

corr Whether the guess (page) was correct or not (i.e., whether it matched “answer”)

inlinks The number of inlinks on the guessed Wikipedia page

File descriptions

qb.train.csv - the training set

qb.test.csv - the test set

qb.guess.csv - a sample submission file in the correct format

features.R - R code to generate some additional features (described in the lecture)