First Place Solution for NLPCC 2018 Shared Task User Profiling and Recommendation

Deep parallel contextual analysis framework based emotion prediction in community wellness communications on social media

“…In this section, we explore the comparative effectiveness of emotion prediction across different models. Specifically, we conduct a comparative analysis of five benchmark models: SVM [ 30 ], ConvNet [ 8 ], RNN [ 10 ], LSTM [ 11 ], and Bi-LSTM [ 12 ], as well as three advanced models, namely ConvNet&Bi-LSTM [ 36 ], Hybrid attention network [ 38 ], and Self-adapting BERT [ 50 ] (their introductions can be found in the related work section), in comparison with the proposed model. The experimental results are presented in Table 6 .…”

“…Convolutional Neural Networks (ConvNets) are commonly used in emotion prediction. Document [ 8 ] used this model for short text emotion prediction, achieving an accuracy of 74.50 %. Zeng et al proposed an improved ConvNet model, which performed well in aspect-level emotion prediction.…”