Embedding's from Language Model with Chi Square Test based Smart City Construction Evaluation

In recent days, cities are attempting to categorize as smart worldwide but classifying truly is becoming a great challenge. The existing methods including Adversarial Neural Network (ANN) for the evaluation of smart city construction mainly suffered with overfitting. In this paper, Embedding's from Language Model with Chi square test (ELMo-Chi Square test) method is proposed. The ELMo feature extraction model is utilized to reduce the overfitting issue of the training model. Initially, the input data is taken from smart portal and then preprocessed with the help of various techniques such as reducing stop words, lemmatization, stemming and data cleaning. After that, the features are extracted with ELMo and then Particle Swarm Optimization (PSO) selects the optimal features such as buildings, roads, streetlights, water management. Finally, the Chi-Square Test is exploited to evaluate the values of extracted data in a sequence. Therefore, the evaluation of smart city construction is performed effectively through proposed ELMo-Chi-Square test model. From the results, the proposed ELMo-Chi Square test established better results in terms of accuracy (0.98), recall (0.96), precision (0.97), and Fl-score (0.97) respectively when compared with Artificial Neural Network (ANN).