The invention relates to a method for superficially analyzing automatic document summarization on the consideration of conceptual abstract degree, pertains to the field of information retrieval and information science, and is characterized in that: first, a document is pre-treated and an abstract value is set; second, sentences in the document are represented by a conceptual vector model via polyseme disambiguation processing; third, the sentences are clustered into a plurality of clusters with similar subjects; forth, compression ratio of a summarization is determined according to the numberof key words extracted from a collected works; fifth, the abstract degree of the sentences is determined; sixth, a plurality of abstract sentences required by the compression ratio is picked up from the clusters in sequence according to IMMRA value; and seventh, sequencing is carried out to the abstract sentences extracted and a summarization document is output. The beneficial results include: the automatic document summarization on consideration of abstract degree is realized and information redundancy and lack of the automatic summarization abstract due to conception inclusion relation are reduced; and the method can adjust the length of the abstract in a self-adaptation way according to the number of the subjects and adjust the induction degree of the summarization according to the needs of abstract degree of a user, thereby being characterized by good adaptability.