Research on sensitive content detection in social networks

Meng, Xuyang; Xu, Yabin

doi:10.1007/s42045-019-00021-x

Cited by 3 publications

(1 citation statement)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Xiao Guanteng [8] combines bad text variant features using word2vec for word vector representation to mine the relationship between variant words and target words, and then uses Stacking model to integrate convolutional neural network, recurrent neural network and multilayer perceptual machine to recognize bad text containing variants. Meng Xuyang [9] used the sensitive word deformed word fingerprint aggregation method to correlate the sensitive deformed word with the original word and then used semantic fingerprinting technique, and then used multi-task learning convolutional neural network to recognize the bad text containing variants by integrating sensitivity and emotional tendency. Fu Cong [10] designed different algorithms by analyzing the features of different morphemes to identify bad text containing variants by matching the target word with the variants in the morpheme database.…”

Section: Introductionmentioning

confidence: 99%

Variant bad text recognition based on ChineseBERT-BiGRU

Huawei,

Wei,

Fangjie

et al. 2024

Fourth International Conference on Machine Learning and Computer Application (ICMLCA 2023)

View full text Add to dashboard Cite

The bad text containing variant words seriously harms the health of the network environment. The existing methods for the recognition of bad variant text do not take into account the importance of the phonetic and positional information for the recognition of bad variant text. In this paper, a ChineseBERT-BiGRU variant bad text recognition model is proposed. This model first learns and trains the word vector of the text by inputting the pinyin information, font information and character information of the text, and then combines the position information of the text. Then the word vector of the text is input into BiGRU to learn richer semantic information of the text. Finally, the bad text containing variant words is identified through Softmax classification. The accuracy, accuracy and F1 values of ChineseBERT-BiGRU in the data set in this paper are compared with those of other models.

show abstract

Section: Introductionmentioning

confidence: 99%