日积月累，以备写作之需。
摘要
1.We also experimentally demonstrate the effectiveness of ListRank-MF by comparing its performance with that of item-based collaborative recommendation and a related state-of-th-art collaborative filtering ranking approach (CoFiRank).
2.We perform extensive experiments on two large real-world CF datasets, and the results clearly show the effectiveness and efficiency of our proposed model.
3.yield better efficiency and effectiveness on two real datasets.
4.We conduct extensive experiments on four publicly accessible benchmarks, showing signifiicant improvements relative to several state-of-the-art collaborative filtering and graph neural network-based recommendation models. Further experiments quantitatively verify the effectiveness of each component of our proposed model and demonstrate that the learned embeddings capture the important relationship structure.
###引言
1.Collaborative filtering (CF) has been regarded as one of the most successful recommender techniques.
2.These models show competing performance for task such as xxx, xxx, and so on.
3.These GCN based recommender models show better performance compared to traditional models.
4.degrade/enhance the recommendation performance.
5.Extensive results demonstrate the superiority of xxx over the strongest state-of-the-art models.
6.The remaining of this article is organized as follows:
7.The rest of this paper is organized as follows. We first provide some preliminaries for ICF in Section 2. We then elaborate our proposed DeepICF methods in Section 3. Afterwards we report experimental results in 4 and review related work in Section 5. Finally we conclude the paper and highlight some future directions in Section 6.
模型
1.有研究表明：It’s reported that/As reported in xx,
2.Before we dive into the details of this survey/ Before going further into different sections
时间复杂度分析
1.we analyze the time complexity of NGCF.
2.For the l-th hidden layer, the multiplication between matrices and vectors is the main operation which can be done in O().
3.The prediction layer only involves inner product of two vectors, for which the complexity is
4.As such, the overall time complexity for evaluating a DeepICF model is
5.Therefore, the overall time cost of evaluating a prediction with DeepICF+a is
泛化
1.can be viewed as a special case of / can also be seen as a instance of
实验
1.Xxx achieves a performance improvement of ca. xx% over xx.
2.Keeping the rest parameters constant, we did a full parameter study for different values of α.
3.To test the different parameters to study the value of α, we fixed the remaining parameter values.
4.without special mention / unless stated differently / unless specified / unless otherwise stated / without additional explanation / without special declaration
5.As we can see, /We can find that/ It can be seen clearly that
We can make the following observations
6.To verify this/ In order to verify this
7.We evaluate ConvNCF of the specific setting as illustrated in Table/Figure 2.
8.The experimental results are provied in Table/Figure x.
9.超过第二好的方法多少，超过最好的baseline
Multi-GCCF consistently yields the best performance for all datasets. More precisely, Multi-GCCF improves over the strongest baselines with respect to recall@20 by 9.01%, 12.19%, 5.52%, and 3.10% for Yelp2018, Amazon-CDs, Amazon-Books and Gowalla, respectively. Multi-GCCF further outperforms the strongest baselines by 12.41%, 15.43%, 24.54% and 6.39% on recall@20 for Yelp2018, Amazon-CDs, Amazon-Books and Gowalla, respectively, when increasing the latent dimension. For the NDCG@20 metric, Multi-GCCF outperforms the next best method by 5% to 25% on three dataset.
10.The above findings provide empirical evidence for the rationality and effectiveness of optimizing the log loss for learning from implicit data.
11.show/report/provide/depict
结论
1.Extensive experiments on two real-world datasets demonstrate the superior perfor mance of our proposed model compared with the state-of-the-art methods.
2.符号表示
denote A by B，用B表示A
set A to B，将A设置为B
refer to A as B 将A称为B
replace A by B 用B替换A
to be specific / specifically
to be fair
leave … as future work
other than
take … into consideration/account
account for
provide A with B
distinguish A from B
associate A with B
relate A to B
combine A with B
replace A with B 用B代替A
be formulated as 归结为
include but are not limited to n. / doing 包括但不限于
be limited by
notably 显著地；尤其
rely on
in addition
as a consequence 因此
to this end
notably / noticeably 显著地
namely
to our best knowledge / to the best of our knowledge 据我们所知
be prone to doing
It is worth pointing out that …
prevent … from …
for the purpose of …
as such / as a result 因此
take … as an example.
enrich/enhance/augment
use/utilize/employ/adopt
combine/integrate
investigate/study
distinguish/discriminate/differentiate
并列关系
A, B, and C
