Objective: Hardly are traders inclined to indicate sentiment in the classical finance, however the behavioral finance paradigm shows that in some cases stock price changes have no fundamental triggers, and just the emotional tendency of the investors play the stimulant role in determining prices. Method: Investor sentiment is defined as the tendency of market participants for speculation and this tendency can be related to the psychological state of the minds of investors. Given the fact that distressed traders are influenced by their emotions and the total emotion of the market, emotional tendency indicators are used to explain the behavior of these types of traders. In this research, the emotional tendency indicators are extracted in two ways: direct approach (survey method) and indirect approach (through the analysis of statistics and market data). Results: Furthermore, this study, based on literature review and stock market conditions in Iran, applied the principal component analysis method (PCA) with different sentiment variables and indicators for extracting a composite sentiment index for extracting noise traders in Iranian stock market. Regarding the special value of the first component and the factor load (coefficients) of the variables, three variables are used in the final index. These three variables are: "Monthly volume of retail transactions by the volume of total stock trades", "Monthly volume of online transactions by the volume of total stock trades", and "Monthly volumes of stock trades by wholesalers and wholesalers by the volume of total market transactions".

Stock market prediction has always been a focus of researchers. Advances in artificial intelligence and machine learning algorithms have enabled the use of textual data alongside numerical data for better stock market forecasting and performance. In this research, to predict the trend of the NewYork Stock Exchange (NYSE) index, numerical data, textual data, and a machine learning model were employed. The model's input includes numerical data as well as the results of sentiment analysis from texts extracted from X (formerly Twitter). Sentiment analysis is performed using a specific machine learning algorithm, Fin-BERT. Additionally, to improve prediction results, prior knowledge of data distribution is incorporated into the cost function of the proposed classifier (SVM). This knowledge is obtained through the calculation of sentiment entropy. Experimental results show that incorporating sentiment entropy into the model's cost function improves prediction performance.

vestor sentiment index and macroeconomic indicators such as inflation, exchange rate, employment growth and real GDP of representative variables were used to market timing to predict the direction and return of the total index of Tehran Stock Exchange. In this regard, four models of logistic regression, Lasso, Ridge and Elastic Net were used using monthly data in the period 1395 to 1399. In order to develop the sentiment index, using the exploratory factor analysis model, six different emotional variables were used, and finally, three variables of stock ratio in the portfolio of investment funds, Tehran price index and top 50 index were selected. The output of the logistic regression model for forecasting based on a single index was compared with the value of forecasting based on other indicators, which showed that logistics forecasting based on all variables was superior to logistic forecasting based on a single index. Comparison of Lasso, Ridge and Elasticnet models for prediction showed that the strength and accuracy of Ridge regression model was more than the other two models, in addition, Lasso and Elasticnet models were almost equally accurate. The results of this research can be useful for investment companies and portfolio managers, analysts and investors.

Sentiment analysis is the process of recognizing or classifying people's opinions and sentiments about a topic. Although earlier sentiment analysis research primarily relied only on text data, recent studies have shown that incorporation of multiple types of data in multimodal models improves performance. In this research, we address multimodal sentiment analysis in the Persian language, proposing a method based on transformer-based models for the first time in this context. For text feature extraction, ParsBERT model is used and DinoV2, a Vision Transformer-based model, is employed for extracting visual features. For sentiment recognition in each modality, sentiment detection capsules are utilized. Finally, to predict sentiment in the multimodal setup, we applied a late fusion technique at the final layer. Furthermore, a model-agnostic explainable AI technique, LIME, is used to gain insights into the predictions made by unimodal branches of the proposed multimodal model. Our experiments showed that our proposed model achieved 96.5% accuracy and 96.48% F1-score on the Aks-Nazar dataset.

Awareness of others' opinions plays a crucial role in the decision making process performed by simple customers to top-level executives of manufacturing companies and various organizations. Today, with the advent of Web 2. 0 and the expansion of social networks, a vast number of texts related to people's opinions have been created. However, exploring the enormous amount of documents, various opinion sources and opposing opinions about an entity have made the process of extracting and analyzing opinions very difficult. Hence, there is a need for methods to explore and summarize the existing opinions. Accordingly, there has recently been a new trend in natural language processing science called "opinion mining". The main purpose of opinion mining is to extract and detect people’ s positive or negative sentiments (sense of satisfaction) from text reviews. The absence of a comprehensive Persian sentiment lexicon is one of the main challenges of opinion mining in Persian. In this paper, a new methodology for developing Persian Sentiment WordNet (HesNegar) is presented using various Persian and English resources. A corpus of Persian reviews developed for opinion mining studies are introduced. To develop HesNegar, a comprehensive Persian WordNet (FerdowsNet), with high recall and proper precision (based on Princeton WordNet), was first created. Then, the polarity of each synset in English SentiWordNet is mapped to the corresponding words in HesNegar. In the conducted tests, it was found that HesNegar has a precision score of 0. 86 a recall score of 0. 75 and it can be used as a comprehensive Persian SentiWordNet. The findings and developments made in this study could prove useful in the advancement of opinion mining research in Persian and other similar languages, such as Urdu and Arabic.