As long as you have a tokenizer, i.e. you can split words in a sentence, language doesn't matter for some basic techniques. Two examples that easily come to mi

As long as you have a tokenizer, i.e. you can split words in a sentence, language doesn't matter for some basic techniques. Two examples that easily come to mind: TF-IDF [1] and TextRank [2,3]. These are good to get started, but if you want to know more about the state of the art, searching on Google Scholar for "keyword extraction $language" is your best bet (and maybe "keyword extraction overview", also in Scholar). [1] http://www.joyofdata.de/blog/tf-idf-statistic-keyword-extraction/ http://www.joyofdata.de/blog/tf-idf-statistic-keyword-extrac... [2] http://digital.library.unt.edu/ark%3A/67531/metadc30962/m2/1/high_res_d/Mihalcea-2004-TextRank-Bringing_Order_into_Texts.pdf http://digital.library.unt.edu/ark%3A/67531/metadc30962/m2/1... [3] https://github.com/davidadamojr/TextRank https://github.com/davidadamojr/TextRank