Tag
This paper proposes a segment combination strategy for automatically classifying research methods in academic papers by partitioning full-text content. Experiments on an annotated corpus from Library and Information Science journals show that methodological information is unevenly distributed, with middle-to-late segments having higher discriminative power.