How to distinguish the meaning of relevance, significance, importance and saliency?

As stated in the title, how to distinguish these four concepts in machine learning, especially in text related tasks?

For example, in query based tasks, like information retrieval, query oriented automatic summarization, we often use the concept of relevance to match the performance of the algorithms. Then, dig a little bit further, among the results, under certain setting or measures proposed by the authors, if one result bears more importance, then this result is sure to be relevant, and even statistical significance. Then what are the differences between them?

Also, when design or learning features, we want to learn salient features. Here, does saliency means importance or siginificane or relevance. Since my mothertone is not English, these concepts always confuse me when I read papers. Is there a guidance out there telling under what circumstance should which word to be used?