4.2 Baseline
which nuggets were extracted as input to the multi-
MMR is a popular approach in query-oriented sum-
document summarizers. That is, in our problem set-
marization. For example, at the TAC 2008 opin-
ting, the relevant documents are already given, al-
ion summarization track, a top performer in terms
though the given document sets also occasionally
of pyramid F score used an MMR-based method.
contain documents that were eventually never used
Our own implementation of an MMR-based base-
for nugget extraction (Mitamura et al., 2008; Mita-
line uses an existing algorithm to maximize the fol-
mura et al., 2010).
lowing summary set score function (Lin and Bilmes,
We preprocessed the Japanese documents basi-
Bạn đang xem 4. - BÁO CÁO KHOA HỌC: "QUERY SNOWBALL: A CO-OCCURRENCE-BASED APPROACH TO MULTI-DOCUMENT SUMMARIZATION FOR QUESTION ANSWERING" POT