Recommendation method based on multi-dimensional alarm information text similarity analysis

• Main Authors: YANG PEIWEN, ZHU MUYAO, WANG QUANXIU, WU PEIXIN
• Format: Patent
• Language: Chinese; English
• Published: 15.05.2020
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention discloses a recommendation method based on multi-dimensional alarm information text similarity analysis, which comprises the following steps: an unstructured public security text is an alarm information text involved in a case, preprocessing the format of the unstructured public security text by adopting an NLP technology, and the unstructured public security text is output as a structured public security text format; acquiring a public security text by a scene-related multi-deep semantic similarity analysis network, and outputting semantic embedding vectors of different levels of the text; obtaining public security texts and semantic embedding vectors of different levels based on a text analysis network of multi-dimensional text semantic similarity, and outputting the publicsecurity texts and the semantic embedding vectors as various extractable entity features; acquiring semantic embedding vectors of different levels of public security texts and various extracted entity characteristics by comb