论文修改建议 (ZhouY 20211027 摘要)

1. Network traffic detection identifies information security threats by analyzing communication data.
   $\to$ 不分析通讯数据，还能分析啥？写具体些.
   $\to$ detection 本身是一个比较虚的名词，可以改为实体. 否则要改为 Network traffic detection aims at identifying 好像又不合适
   Network traffic detectors identify information security threats by analyzing data streams.
2. Existing detection methods analyze instances as independent individuals one by one, resulting in unsatisfactory detection efficiency.
   $\to$ 不要重复使用一个单词
   $\to$ 这里只有数据包的概念，而并没有什么实例
   $\to$ 太啰嗦了
   Popular detection methods are inefficient when analyzing data packets one by one.
3. In this paper, we propose a multi-granular network traffic detection framework based on multi-instance learning to address this issue.
   $\to$ 还行，暂时不改
4. Three major techniques make up the framework as follows.
   $\to$ 写得太 low 了，可以直接删除本句
5. The bag generation technique assembles the same number of data packets into a bag-granular space according to the protocol type.
   $\to$ 这里有空间吗？
   The bag generation technique assembles a fixed number of data packets of the same protocol type into a bag.
6. The bag mapping technique encodes the bag into a new vector through fast clustering and distance computation.
   $\to$ 这里才有空间吧？
   $\to$ 聚类与距离计算，这两个无法并列吧？明显后者更基础。
   $\to$ 可以把你使用的聚类技术写出来，我随便写了个 density clustering, 你可以改。
   The bag mapping technique encodes each bag into a new feature space through density clustering.
7. The multi-granular classification technique filters data at the bag granularity to narrow the detection range, and then performs fraudulent traffic detection at the instance granularity.
   $\to$
   The multi-granular classification technique efficiently filters normal data at the bag granularity, and then detects threats at the packet granularity.
8. Experiments on the KDD CUP 99 data set show that the detection time of our proposed algorithm is lower than that of the most state-of-the-art algorithms and maintains considerable accuracy.
   $\to$ considerable 是什么东东？我暂时也找不到好的词来替换.
   $\to$ 如果你的效率并没有明显的改善，投会议还是比较难中.
   Experiments on the KDD CUP 99 data set show that our algorithm is more efficient than most state-of-the-art algorithms while maintains good accuracy.

小结：总体而言，中规中矩，不够炫。