Seq-CNN: A Lightweight and Efficient Neural Network for Sentimental Analysis of Text

Today, numerous smart devices have brought convenience to people's work and life, especially chatbots, which can simulate real-world conversations, adding much fun to people's lives. However, there are few intelligent chatbots suitable for children, especially those with Autism Spectrum Disorder (ASD). In order to develop chatbots suitable for children with autism, aiming to enhance their social and communication skills, this research serves as a foundational study for an intelligent emotion-perception chatbot, conducting preliminary research on text sentiment analysis. This study optimizes an open-source dataset to make it more applicable to text sentiment analysis tasks for children with autism. A new model, Seq-CNN, combines LSTM and Convolutional Neural Networks cleverly and applies attention mechanisms, offering a concise structure and superior performance compared to existing models like CNN, TextCNN, Bi-LSTM, and so on. This provides a possibility to better realize an emotion-automatic-perception chatbot, aiming to improve the language and social abilities of children with ASD through early intervention. Furthermore, due to the lightweight and high efficiency of this model, it can also be applied to general text sentiment analysis tasks.