Recruitment Fraud Detection Method Based on Crowdsourcing and Multi-feature Fusion
Junling Wang, Бо Лю · 2022
Regarding the problem that false recruitment information is difficult to identify, an online recruitment fraud detection method based on crowdsourcing and multi-feature fusion is proposed. Firstly, the keywords in the recruitment text are extracted through a data crowdsourcing platform; Secondly, a text classification model merged Bidirectional Encoder Representation from Transformers (BERT), Bi-directional Long Short-Term Memory (BiLSTM), Convolutional Neural Network (CNN) and Attention mechanism is constructed. The model can fuse the contextual features and local features of recruitment text, at the same time assign different weights to the words in the text, and focus on important words; Finally, the keywords extracted by the crowdsourcing platform are integrated into the model to guide its training and highlight the role of keywords in classification. Compared with other network models, the experimental results show that the classification accuracy of the proposed model is increased by 5.3% on average, and can effectively detect online recruitment fraud.