CAJ | 학술논문

提出一种基于LDA( Latent Dirichlet Allocation)潜在语义模型、全面研究科技期刊主题演化过程的方法。该方法根据科技期刊的特点引入时间因素，使用困惑度确定最优主题数目，通过LDA主题提取结果及JS散度，实现主题在强度和内容两方面的演化研究，并对不同时间窗口的主题稳定性做出相应分析。实验结果表明该方法可以较好地分析某一特定科技期刊的主题随时间的强度演化规律以及主题内容的演化趋势。
제출일충기우LDA( Latent Dirichlet Allocation)잠재어의모형、전면연구과기기간주제연화과정적방법。해방법근거과기기간적특점인입시간인소，사용곤혹도학정최우주제수목，통과LDA주제제취결과급JS산도，실현주제재강도화내용량방면적연화연구，병대불동시간창구적주제은정성주출상응분석。실험결과표명해방법가이교호지분석모일특정과기기간적주제수시간적강도연화규률이급주제내용적연화추세。
This paper proposed a method based on LDA ( Latent Dirichlet Allocation) Latent semantic model to comprehensively study scientific journal's topic evolution, which according to the characteristics of the scientific journals, added time factor into the model, then confused degree method was used to determine the optimal number of topics. Through LDA topic extraction and JS divergence, we tracked the journal topic trends in different time intervals and conducted topic evolution analysis on aspects of both contents and strength, and the stability of topics over time were also included. The results show that this method can analyze scientific journal's topic strength evolution over time better, in addition, the evolution trend of the journal topics contents can be described as well.