This study aims to probe into how the semantic association and causal relation of the predictive event contained in the predictive context contribute to the coding of predictive inference(PI).

Moving window display technique was used to investigate the predictive inference elicited by goal-focus information during text reading.

This study mainly deals with the influence of working memory capacity on the generation of predictive inference during text reading.

Based on BP neural network algorithm,utilizing the experimental data of concrete under sulfate corrosion as the training samples,the models are proposed for inside and outside predicting the corrosion depth of concrete,which consider the influence factors of the concentration of sulfate radical in solution,the tensile and compressive stress levels of concrete and the corrosion time.

According to the example,the paper infers and forecasts systemically,and calculates the number for in formation entropy,then compare with maximum of Shannon to make sure the main reason which lead to "bad phenomenon" in engineering management in formation entropy.