9 June 2021 in IEEE/ACM Transactions on Audio, Speech, and Language Processing
The ability of dialogue systems to express pre-specified style during conversations has a direct, positive impact on their usability and user satisfaction. While it has attracted much research interest, existing methods often generate stylistic responses at the cost of content quality. In this work, we introduce a prototype-to-style (PS) framework to tackle the challenge of stylistic dialogue generation. The proposed framework first exploits an Information Retrieval (IR) system and extracts a response prototype from the retrieved response. A stylistic response generator then takes the response prototype and the desired style as input to produce a high-quality and stylistic response. To effectively train the proposed model and imitate the real testing environment, we introduce a new style-aware learning objective and a denoising learning strategy. Results on three benchmark datasets (gender, emotion, and sentiment) from two languages demonstrate that the proposed approach significantly outperforms existing baselines both in terms of in-domain and cross-domain evaluations.
14 March 2022 in International Journal of Emerging Technologies in Learning (iJET)
The existing research on teaching development of teachers fails to effectively quantify the teaching development trend. This paper deeply mines the evaluation data on the teaching quality of college teachers, before analyzing and predicting the trend features for teaching development of college teachers based on knowledge discovery. Firstly, the knowledge features of the teaching development trend of college teachers were examined. Next, the fluctuation features of the time series on the teaching quality development of college teachers were described based on chaotic time series. In addition, a prediction model for teaching development of college teachers was established for weighted first-order chaotic time series, and used to simulate the nonlinear features of the time series on the teaching quality development of college teachers. The prediction model was proved effective through experiments.