In an e-commerce environment where millions of transactions take place between suppliers and users, the need arises to establish the validity of the service provided. Market players have provided a customer feedback system to meet this need. But the feedback generated cannot always be relied on. Feedback can positively or negatively influence sales, instead of showing the actual authenticity of the product or service, from the customer's point of view. Our work proposes an improvement on the traditional feedback system by introducing a Trust Reputation System (TRS) that helps filter good customers using a set of algorithms, thus creating a degree of trust for the user. Say no to plagiarism. Get a tailor-made essay on "Why Violent Video Games Shouldn't Be Banned"? Get an Original Essay Consumers in the online market have the problem of filtering the best products from a list of a variety of options. There are various market players who provide feedback system to help the customer to identify quality products by examining the customer's opinion and accordingly choose the product. Most consumers purchase products based on product reviews. This negatively or positively affects the sale of products. Furthermore, this paves the way for spammers to undercut the sale of the product. To eliminate this, the paper focuses on improving the feedback system by introducing the concept of reliability. This can be done through the Trust Reputation System. TRS are programs that allow users to rate each other. Using these methods can help reduce the number of spammers, potentially increasing the number of genuine reviews. The advantage of such reviews is that they help determine the authenticity of the product. Sentiment analysis has been studied in a wide area of ​​the domain such as movie review, teaching review, product review, e-learning, hotel review and many more. Most scholars have focused on quantitative data analysis. However, some studies have been conducted on qualitative data using sentiment analysis, we found six works that mentioned the idea of ​​using opinion mining and sentiment analysis in education. Algorithms such as Naive Bayes, k-means and Support Vector Machines are used in opinion classification. . The paper also focuses on the actual reputation system. There are different reputation system architectures with different algorithms for calculating the reputation score for your product. Many authors have proposed in their work different TRS architectures with different algorithms to calculate the reputation score related to the product. Furthermore, some academic works on the Truthreputation system have been dedicated to the inclusion of semantic analysis of feedback in the calculation of the product trust score and especially the degree of user trust. Even in studies that attempt to provide more complex reputational methods, some issues are still not taken into account, such as the credibility of referees, the updating of the user's degree of trust with each posting, the age of rating and feedback or the agreement between the rating given that it is a scalar value and the textual feedback associated with it. In contrast to the aforementioned TRS, our proposed design overcomes these problems and makes use of an algorithm that includes the analysis of textual feedback in order to calculate the degree of trust of the user providing the feedback and a reliable reputation score for the product. Consumers in the online market have the problem of filtering the best reviews or feedback for purchasing products.We try to eliminate the problem by listing the best reviews so that it becomes easy for customers to decide on a product by analyzing the experiences of other consumers, allowing them to post their own reviews. Consumers dealing with the online market may sometimes purchase substandard products. Although the e-commerce company provides services like returning and exchanging products, the process sometimes becomes a tedious task. The project aims to provide customers with the opportunity to select the desired products based on the rating of the item they want or intend to purchase, which has been rated based on the rating and reviews provided by consumers with the help of TruthReputation System ( TRS). The Opinion Mining of our project will be based on Sentiment Analysis algorithms and methods and also on the Truth Reputation System algorithm. Trust Reputation Systems (TRS) will provide the information needed to support reliance on parties to make the right decision in an electronic transaction. Indeed, as providers of security in electronic services, TRS must faithfully calculate the most reliable score for a targeted product or service. TRS must therefore rely on a robust architecture and adequate algorithms capable of selecting, storing, generating and classifying scores and feedback. In the proposed architecture, for each user who wishes to leave a rating (appreciation) and feedback (semantic review), we analyze the customer's attitude towards a set of short, selected feedback and the by-products stored in the knowledge base. This user's review will be reached by any other user. So, let's say we have a route that connects all users (nodes). Therefore, we need to know the degree of trust of the user and determine the degree of trust of the feedback. Trust Reputation System DesignAlgorithm DescriptionThe customer begins by providing a rating and textual feedback on a specific product. When they click Submit, to validate the information provided, we will redirect the user to another interface showing this message, for example: "please give us your opinion on the following feedback before validating the information provided below:" In this interface we will find feedback chosen from the database of different types. Such feedback can be fabricated to summarize numerous user feedback stored in the database. The generated feedback can be stored in another knowledge base. So, as much as we add feedback into the ordinary database, we will fill the knowledge database with pre-fabricated feedback using algorithms and text mining tools. However, some users can provide pre-summarized feedback that can be included directly in the knowledge database. In fact, are there many text mining and data mining algorithms and tools capable of searching for the most appropriate feedback that is first and foremost linked to the product and that can summarize and summarize the majority of each type of user? feedback. In fact, before sending customer feedback and appreciations on the product to the reputation trust system, we must verify the concordance between them in order to avoid and eliminate contradictions or malicious programs that attack our system. In the redirected interface, we will show several feedbacks of different types. However, the user can specify the number of feedback to like or dislike. Of course we can also specify the minimum and maximum number of feedback the user should see. Indeed, we are trying through this redirect to detect and analyze the user's intention behind his intervention on the e-commerce application. Sowe examine and evaluate his intention using other prefabricated feedback of different types. Of course we already have the reliability of any feedback. Consequently, we use our reputation algorithm studied in the section to generate the user's degree of trust which plays the role of coefficient and then rectify his rating according to his degree of trust and generate the feedback score. In fact, the reliability of each feedback falls within a threshold. The closer the reliability is to 5, the more reliable the feedback. The closest reliability is to -5, the most unreliable feedback is. If the feedback is reliable, its score would be included otherwise it would be included in the B algorithm. TRS The reputation algorithm used in this TRS uses semantic feedback analysis to generate a reliable reputation score for the product. In reality, we have 3 types of feedback: Positive feedback: represents opinions that express a positive point of view on the product. These improving opinions contain positive content regarding the product. Thus, the adjective positive refers to the nature of the content of the feedback, not its reliability. However, any feedback, whatever its type, can have a positive or negative reliability. Both positive and negative reliability, it is gradual: it has degrees as fluctuating in a threshold of. Negative feedback: represents opinions that speak negatively about the product. Logically, users who express such opinions are not satisfied with the commented product. This feedback may or may not be true or may be far from the truth. This is why each feedback has its reliability represented by a float number between -5 and 5. Mitigated feedback: represents feedback that speaks positively about some aspects of the product and negatively about other aspects. They are also characterized by the reliability included in the feedback. Contradictions: represent feedback with contradictory content, for example a feedback where the user is not talking about the specified product but about another one or is saying that the camera of a mobile phone is great and later according to the same opinion says that the camera it's terrible. In fact, we need to start by noting contradictory feedback. So we need an algorithm and a semantic analysis tool that can detect contradiction in specific content related to a product. We can customize the analysis based on the product. For example, if the user says that “the swimming pool of the hotel that does not have one is not clean”, the algorithm must be able to detect this major contradiction. We can give the algorithm for each product as input the property of the algorithm; if there is no similarity we can consider it a contradiction. But the agreement obviously understands its meaning. Because if the customer writes that the negative thing about this hotel is that there is no swimming pool. He is telling the truth so obviously the presence of an absent property in feedback does not mean there is a contradiction. In fact, before sending customer feedback and appreciations on the product to the reputation trust system, we must verify the concordance and alliance between them so as not to have contradictions. After verifying the concordance between the ratings and the textual feedback we will redirect the user to the selection of pre-fabricated feedback. Then the user will click Like or Dislike based on each feedback. The click event will be handled in order to obtain some information necessary in calculating the user's degree of trust. The function uses the feedback id as a parameter to obtain its reliability from the knowledge base. We also need to obtain the user's previous level of trust if it has already been.