摘要：Novelty is a concept of great importance in the writing of a research paper, given that the author has to persuade the audience of the news value of the reported research, which makes it worth publishing. In this paper I use a corpus of computer science papers to investigate how novelty is created in this discipline. I analyse how the author uses evaluation and lexical cohesion to integrate his/her research into the existing knowledge structure of the field. Evaluation in computer science papers is closely associated with the Problem-Solution pattern which structures most of the papers: authors claim that the technology they introduce is the best solution to a problem that they have previously identified. Lexical cohesion highlights the novelty of the research by establishing a semantic relation of contrast between the fragment of text reporting previous research in the field and that reporting the authors' own research.
