The invention discloses an enterprise portrait method based on
label layered depth-
delay modeling, and the method comprises the steps: firstly carrying out the statistics and screening of fuzzy labelsof an enterprise, screening out labels which cannot completely summarize the characteristics of the enterprise, such as wholesale industry and retail industry, and carrying out the classification anddepth-
delay of the screened-out labels through employing a Bert model according to the business range of the enterprise and the labels of the enterprise; secondly, integrating enterprise names, enterprise brief introductions and operation range information, performing feature expansion based on a pre-established enterprise
lexicon, extracting keywords from comprehensive information by using TextRank, TF-IDF and LDA topic models respectively, and taking the processed keywords as deeper enterprise
delay labels; finally, applying the modeling method to an enterprise portrait
system, and optimizing the accurate summarization capability of the
label. The method is universally suitable for
label delay modeling and label extraction, the hierarchical relationship of label delay is fully considered, and the accuracy of labels and enterprise portrait systems can be effectively improved.