It is also necessary to clarify who is responsible for each part of the work. Therefore, metadata must be attached to the data obtained during each operation to indicate who is running the workflow and who is providing the data. It is best to have an abstract mechanism that can automatically extract a rough picture, which is admirable but cannot be copied.
Copyright issues also indicate the heterogeneity of scientific workflow, which requires the cooperation of multiple parties in different geographic locations.
Exploratory and frequent changes: scientists need to constantly adjust the parameters of workflow or even modify the control flow to try new results, but they are not as professional as process designers, therefore, you must provide an easy-to-use interface.
Scientists, such as astronomical researchers, need to share large-scale data if they want to collaborate in scientific research, which is equivalent to doing some distributed computing. This data flow process may become more and more complex as research institutions increase, so that they need to be managed independently. This is the general meaning of scientific workflow. Scientific workflow seems to be a new direction. I heard someone talk about it in the discussion class. Later, I read an article by a group of reviewers. Today I found another article in The December computer magazine.
In computer magazine, this article about scientific workflow is full of text, with almost no numbers. After reading it, I felt a bit empty. The difference between scientific workflow and business workflow is big and small. List the features/requirements of several scientific workflows:
Repeatability: This is the basic requirement of scientific research, but it is difficult to achieve it because the system is distributed and the data is distributed. It is hard to say which data can always exist. Maybe after several years of system upgrading, the previous programs will not be able to run. Who is calling it a daily change in the infrastructure of computer software?
Copyright problem: the data of scientists is very precious. the workflow of workflow and the program running on each endpoint cannot be disclosed at will. Otherwise, how can this problem be plagiarized. However, we need to show our research results to others, at least to the reviewers. Otherwise, how can we let others agree with your work? The Distributed Workflow means that the final result is shared by everyone. Therefore
Other features are boring buzzword, such as "more flexible", "Better scaling", and security.
Summary:Although there are no essential new things, in the eyes of people engaged in CS, SCI workflow is, after all, a new version of the big BBS for computer research. It is a good place for irrigation, you can grab a location. I have reprinted the post in the old version of workflow and published a new article. Maybe someone else will give it to you.