In this paper, we crawled and analyzed the dataset of contemporary literature works through Python technology, and after preprocessing the data with the use of the word splitting algorithm, we applied the CasRel entity-relationship extraction model based on the Transformer and the BERT model to efficiently extract the entity knowledge in contemporary literature works. Then, a knowledge map of modern literature was established by combining the Neo4j graph database. Several novels by Mo Yan, Jia Pingwa and Chen Yan were selected as cases to analyze the textual values and linguistic features of contemporary literary works. The Analysis reveals significant differences in the linguistic features and connotative values of works by different authors. For example, the average frequency of periods per thousand words in Mo Yan’s works is 0.2187, while the corresponding frequency in Jia Pingwa’s works is 36.06% lower than that of Mo Yan. The cumulative application frequency of monosyllabic and disyllabic words in Chen Yan’s works exceeds 80%, and the lexical density ranges between 67.5% and 71.5%. By clarifying the linguistic features of different contemporary literary works, it can help academic creators better find the resonance point with readers and provide diversified innovative paths for literary creation.