关键词: 科技文献, 主题摘要, 生成算法, hLDA

Abstract:

With the advent of the era of information explosion, and the rapid growth of the number of science and technology literature, the requirement of obtaining effective information for science and technology literature to the science and technology workers is becoming higher. This paper proposes a scientific literature theme abstract automatic generation algorithm. We make modeling theme to the data set of scientific literature by usinghLDA model, and automatically generate the abstract for the potential themes in science and technology literature through the choice of the candidate words and scoring strategy to the sentences of integrating multiple factors. In the experiment, we put forward the evaluation method based on topic coverage. The experimental results verify the validity of the generation algorithm for the theme abstract proposing in the paper.

Key words: Scientific and technical literature, Theme abstract, Generation algorithm, hLDA