In the project you need to train in a batch of new datasets with the same format, without having to copy the new data set to the original training data directory, and simply create a link to the new data file in the original training data directory.
Or when we need to use the same file in a different directory, we do not need to put in every required directory must be the same file, we just put in a fixed directory, the file, and then in other directories with the ln command link, it can, do not have to repeatedly occupy disk space.
command Format :$ ln [参数] [源文件/目录] [目标文件/目录]
As the following code is built under/home/targetdoc (target directory), a/home/data/train_data (source file directory) is created with a soft connection, similar to a shortcut under Windows
$ ln -s /home/data/train_data /home/targetdoc
When you're done, you can $ ll
view the links in the current directory
See in more detail: http://www.runoob.com/linux/linux-comm-ln.html
Shell commands ln to link files to other directories