Big Data and Deep Learning‐based Tourism Industry Sentiment Analysis Using Deep Spectral Recurrent Neural Network

Big data has fascinated the consideration due to its great potential and ability to solve problems associated with large amounts of data. The tourism industry is also one of the industries that seek to use the concept of big data to improve business processes. Large-scale tourism, with its convenient, fast and low gateway, makes it very convenient for tourists to make sentiment calculations, and it has become one of the main sources of tourism big data. However, tourism review texts are often short texts, and the erratic distribution of emotions makes it difficult to obtain accurate emotional analysis results and during training can be very difficult when there is a lot of irrelevant and unwritten information or when there is noisy and unreliable data. To overcome the issues, in this work the proposed, Deep Spectral Recurrent Neural Network (DSRNN) for classify the tourism sentiment analysis (SA). Initially start with, input of the tourism review data for preprocessing stage in this step using Individual Value Decomposition Analysis (IVDA) to remove unwanted data from the dataset. Then extracting the data from the previous step for feature extraction using Spider Optimization to select the Effective Features Weight (SO-EFW) of evaluating the feature weights to classify the data. And the SA for tourism industry to get a positive and negative reviews to reduce these steps. In the training features estimating in the Softmax activation function. Finally, classification using Deep Spectral Recurrent Neural Network (DSRNN) based on the testing data features to get better classification results. Simulation results show the better accuracy comparing with previous methods.