假设我们写的是《Sample: A novel Recommendation model for sparse user-item matrix》
第一逻辑段,套话:
In this section, we depict the intuition and technical details of xxx
第二逻辑段(包含若干自然段和图)给出一个running example,所谓的running example就是在全文中通用的例子
In most recommendation scenario, sparsity of rating matrix leads to inaccurate prediction of user-item ratings. As illustrated in Fig. 1, the input user-item matrix is a very sparse matrix because every user only rates few items. Suppose unrated data are covered with shade, xxx
这段主要描述的是你在introduction里面介绍的challenge,不过要讲的更细,更直白,并且画个图把这个问题表示清楚。
有的时候这个图会被有些作者放到introduction里面去,那样的写法也是可以的。
第三逻辑段(有的时候作为一个subsection,包含若干自然段和至少一个图)给出你的主要工作细节
第一个自然段肯定是说你的主要思路
Intuitively, we can fill shaded blocks (missing data) using available information in the original matrix. If our algorithm achieves confidence in the prediction, then we give this shaded block a predicted rating value . Otherwise, we set the value of this missing
data to zero.