Adaptive Summaries: A Personalized Concept-based Summarization Approach by Learning from Users' Feedback

Exploring the tremendous amount of data efficiently to make a decision,\nsimilar to answering a complicated question, is challenging with many\nreal-world application scenarios. In this context, automatic summarization has\nsubstantial importance as it will provide the foundation for big data analytic.\nTraditional summarization approaches optimize the system to produce a short\nstatic summary that fits all users that do not consider the subjectivity aspect\nof summarization, i.e., what is deemed valuable for different users, making\nthese approaches impractical in real-world use cases. This paper proposes an\ninteractive concept-based summarization model, called Adaptive Summaries, that\nhelps users make their desired summary instead of producing a single inflexible\nsummary. The system learns from users' provided information gradually while\ninteracting with the system by giving feedback in an iterative loop. Users can\nchoose either reject or accept action for selecting a concept being included in\nthe summary with the importance of that concept from users' perspectives and\nconfidence level of their feedback. The proposed approach can guarantee\ninteractive speed to keep the user engaged in the process. Furthermore, it\neliminates the need for reference summaries, which is a challenging issue for\nsummarization tasks. Evaluations show that Adaptive Summaries helps users make\nhigh-quality summaries based on their preferences by maximizing the\nuser-desired content in the generated summaries.\n