LitVis: a visual analytics approach for managing and exploring literature

Reading literature is essential to research. However, the explosive growth, the multidimensional attributes, and the complex relationships pose a tremendous challenge for researchers to understand and analyze literature efficiently. We propose LitVis, a visual analysis approach to help users manage and explore literature based on its metadata. LitVis allows users to select literature collection of interest and analyze them from their attributes, text, and citation networks. From the perspective of attribute values, LitVis supports users in understanding the distribution of literature and filtering individuals of interest. From the perspective of the text, LitVis uses the Latent Dirichlet Allocation model to extract topics from the literature and allows users to adjust the topic extraction results interactively. From the citation network perspective, LitVis enables users to analyze citation relationships within and between topics to help them understand research development. One use case and carefully designed interviews with domain experts validate the effectiveness of LitVis in the management and analysis of the literature. The results show that LitVis help users comprehensively identify the literature collection of interest and efficiently analyze the evolution of research topics.