By reading and analyzing Moonlight Chopping and Cast Swords meticulously, this study reveals the polysemous textual world constituted by these two works, covering the mockery of dystopian realities, the inheritance of the spirit of vengeance, as well as the profound reverence for the essence of life. Web crawler technology is employed to obtain the textual content of Moonlight Chopping and Cast Swords, and the data is pre-processed. A corpus is constructed using a lexical algorithm. The in-depth extraction of thematic linguistic features of the two works is realized by combining Gibbs sampling and EM algorithm to accurately estimate the parameters of the LDA topic model. It is found that Moonlight Chopping and Cast Swords show significant correlation on the story and ideological level, but apparent differences in story focus, expression form and aesthetic style. Specifically, the average sentence length of Moonlight Chopping is only 1.18 units longer than Cast Swords. At the same time, its STTR value is only about 4.3% higher than Cast Swords, indicating that Moonlight Chopping has successfully achieved subtle thematic differences while maintaining the vengeful narrative style of Cast Swords, demonstrating a profound transformation of traditional culture and historical continuity.