Lingvis.io - A Linguistic Visual Analytics Framework.pdf

The lingvis.io framework is a modular platform for rapid-prototyping of linguistic, web-based, visual analytics applications. It provides a rich set of machine learning and natural language processing steps, which are encapsulated into micro-services and combined into a computational pipeline. The pipeline is auto-configured based on the requirements of the visualization front-end, allowing for detached and independent development of linguistic processing and visualization design. The framework supports various human-in-the-loop, linguistic visual analytics research techniques and applications, including content analysis, discourse analysis, language change, readability analysis, language modeling, and argumentation analysis. It enables developers to request natural language processing steps for their visualizations, which are auto-configured based on user-defined parameters, and provides a flexible pipeline with automatic dependency resolution. The framework is compared to other notable frameworks, including Stanford CoreNLP, GATE, and Weblicht, and is designed to accommodate the needs of rapid-prototyping and to disguise the involved engineering complexity for application developers.