چكيده به لاتين
Nowadays, with the spread of social networks, people share their brief contents, opinions, and data through such networks. The shared content includes the authors' stance, sentiments, and emotions; and impacts peoples' opinions and thoughts. To publish efficient information, users should investigate and analyze a wide range of published data. This activity helps them to maintain contacts and influence their followers' thoughts.
Accessing the available information in a social network by user, reviewing them, and creating a text according to the user's request, requires plenty of time. It is advantageous to construct a summary of the data on social media. If the summary contains the user's requested emotions, it will be more effective. With the help of the generated summary, the user can discover his desired content; and express his opinion according to the information he has obtained. This process helps users' high participation in publishing content and maintaining their position in the network.
In this work, we will produce online summaries of several documents that conform to the user's needs. The summary generates a specific topic, according to users' requests, in types of emotion, sentiment, and stance. Also, the" text emotion changing unit" is defined to achieve the selected kind of emotion. This unit aims to create and elevate the preferred type of emotion. In produced summary, we also aim to display the user's thoughts. To achieve this goal, we use a 'navigator sentence" and a set of keywords, which the user provides. The "navigator sentence" includes the users' thoughts to obtain information related to it. As a result, with the help of similarity measurement methods, a summary containing the user's desired content is produced.
The method presented in this research is evaluated on the Twitter dataset. The evaluation of the proposed method shows that 82.7% of the generated summaries have the expected changes. Users can recognize 75.18 % of the target emotion in the created summary. Also, the probability of understanding the concept and opinion that the user wants to become manifest is 61.48%.
