Sentiment analysis in conversations with deep learning using quantum computing

Quantum theory was originally proposed as a physical theory to describe the behaviour of microscopic particles, and was later applied to various non-physical fields that exhibit quantum-like properties, including human cognition and decision-making. One of these fields is sentiment analysis. Sentiment Analysis is one of the most popular applications of Natural Language Processing, which allows the study of people's emotions or attitudes towards a situation, event or speech. Many studies in the literature work on film reviews and twitter data. There are almost no studies that perform sentiment analysis on health data. In this thesis, new deep learning based methods using classical and quantum computing are proposed to perform sentiment analysis on health data. The first of the proposed methods uses the LSTM deep learning model to perform a three-pole classification as positive, neutral and negative with sentiment analysis from mutual interviews in the field of health. In this method, firstly, the data for model training was prepared by text processing methods and the model was trained with the data set prepared with the LSTM deep learning algorithm. As a result of the training, the success rate was calculated as 94%. When the success rate obtained is compared with the studies examined in the literature, the success rate of the proposed method is quite high. Another one of the proposed methods is the sentiment analysis of patients' reviews about hospitals. Medical reviews of patients are very important for medical management departments. Four neural models were developed to classify patient reviews as positive or negative. Firstly, the online data were preprocessed. Then Skipgram word embedding was used to perform neural training and finally training was performed. After the training stages, LSTM-CNN and LSTM architectures were the two models with the best success score with over 85% performance. The last of the proposed methods is the LSTM recurrent neural network model used in quantum computing and deep learning to classify the emotions contained in texts consisting of doctor-patient dialogues as positive-neutral-negative. The success rate of the proposed method is 94.52%. Quantum computing method plays a critical role in the execution of programmes in terms of time. Therefore, it is thought that the proposed method will pioneer the classical-quantum hybrid method for studies in other fields