Research on Modeling and Analysis of Generative Conversational System Based on Optimal Joint Structural and Linguistic Model.

Generative conversational systems consisting of a neural network-based structural model and a linguistic model have always been considered to be an attractive area. However, conversational systems tend to generate single-turn responses with a lack of diversity and informativeness. For this reason, the conversational system method is further developed by modeling and analyzing the joint structural and linguistic model, as presented in the paper. Firstly, we establish a novel dual-encoder structural model based on the new Convolutional Neural Network architecture and strengthened attention with intention. It is able to effectively extract the features of variable-length sequences and then mine their deep semantic information. Secondly, a linguistic model combining the maximum mutual information with the foolish punishment mechanism is proposed. Thirdly, the conversational system for the joint structural and linguistic model is observed and discussed. Then, to validate the effectiveness of the proposed method, some different models are tested, evaluated and compared with respect to Response Coherence, Response Diversity, Length of Conversation and Human Evaluation. As these comparative results show, the proposed method is able to effectively improve the response quality of the generative conversational system.

Three-dimensional numerical simulation of water quality and sediment-associated processes with application to a Mississippi Delta lake.

A three-dimensional water quality model was developed for simulating temporal and spatial variations of phytoplankton, nutrients, and dissolved oxygen in freshwater bodies. Effects of suspended and bed sediment on the water quality processes were simulated. A formula was generated from field measurements to calculate the light attenuation coefficient by considering the effects of suspended sediment and chlorophyll. The processes of adsorption-desorption of nutrients by sediment were described using the Langmuir Equation. The release rates of nutrients from the bed were calculated based on the concentration gradient across the water-sediment interface and other variables including pH, temperature and dissolved oxygen concentration. The model was calibrated and validated by applying it to simulate the concentrations of chlorophyll and nutrients in a natural oxbow lake in Mississippi Delta. The simulated time series of phytoplankton (as chlorophyll) and nutrient concentrations were generally in agreement with field observations. Sensitivity analyses were conducted to demonstrate the impacts of varying suspended sediment concentration on lake chlorophyll levels.