Skip to main navigation Skip to search Skip to main content

A Financial Advertisement Recognition Algorithm Model Based on Text

  • Harbin Institute of Technology

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Acquisition and intelligent recognition of financial advertising is the premise of monitoring and managing it. This paper studies how to obtain and intelligently recognize financial advertisements, proposes an algorithm of financial search terms extension based on graph constructed from context, and a XGBoost multi model algorithm of financial advertisement recognition combined with multi text representation such as Word2Vec and ELMo and machine learning models such as CNN and RNN, constructs a model of financial advertisement recognition algorithm based on text, and realizes the specific recognition of financial advertisement. The intelligent acquisition and accurate identification of advertising provides data support for further multi-dimensional analysis, and lays a solid foundation for the subsequent governance of illegal financial advertising.

Original languageEnglish
Title of host publicationProceedings - 2021 3rd International Conference on Natural Language Processing, ICNLP 2021
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages26-32
Number of pages7
ISBN (Electronic)9781665414111
DOIs
StatePublished - Mar 2021
Event3rd International Conference on Natural Language Processing, ICNLP 2021 - Virtual, Online, China
Duration: 26 Mar 202128 Mar 2021

Publication series

NameProceedings - 2021 3rd International Conference on Natural Language Processing, ICNLP 2021

Conference

Conference3rd International Conference on Natural Language Processing, ICNLP 2021
Country/TerritoryChina
CityVirtual, Online
Period26/03/2128/03/21

Keywords

  • CNN
  • ELMo
  • Louvain
  • RNN
  • Word2Vec
  • XGBoost
  • multi model

Fingerprint

Dive into the research topics of 'A Financial Advertisement Recognition Algorithm Model Based on Text'. Together they form a unique fingerprint.

Cite this