WINS: Web Interface for Network Science via Natural Language Distributed Representations Dario Borrelli (B ) , Razieh Saremi, Sri Vallabhaneni, Antonio Pugliese, Rohit Shankar, Denisse Martinez-Mejorado, Luca Iandoli, Jose Emmanuel Ramirez-Marquez, and Carlo Lipizzi School of Systems and Enterprises, Stevens Institute of Technology, 525 River Street, Hoboken, NJ 07030, USA {dborrell,clipizzi}@stevens.edu Abstract. This work proposes a novel approach to visually interact with semantic networks constructed via natural language processing tech- niques. The proposed web interface, WINS, allows the user to select a textual document to be analyzed, choose the algorithm to construct the semantic network, and visualize the network with its metrics. Unlike pre- vious works, which are typically based on co-occurrence matrix for con- structing the text network, the proposed interface embeds an additional approach based on the combination of network science with distributed representations of words and phrases. Keywords: Distributional hypothesis · Networks · Natural language 1 Introduction The large-scale amount of digital information produced nowadays, especially tex- tual information, can be an important resource for studying how words, idioms and their semantic meanings vary depending on different variables. These vari- ables may be time, domain knowledge, socio-cultural bias, political bias, the author of the text, the audience, just to name few examples. Therefore, mean- ing is a relative concept that may assume a different connotation in different contexts. Books, newspapers articles, scientific articles, patents, unstructured text from social media, web search engines, medical reports, contracts, government forms, all are examples of textual information that is available in digital format or can be converted into digital format. Having such resource potentially available nat- urally fosters research on methodological, computational, and visual ways to find relationships among meaning and concepts in these documents. Detecting these relationships would enable an analytic approach that could (i) reduce the effort that results from a manual analysis, (ii) will provide a scalable, computational way to compare different documents that could make latent knowledge emerge. c  Springer Nature Switzerland AG 2020 C. Stephanidis and M. Antona (Eds.): HCII 2020, CCIS 1224, pp. 614–621, 2020. https://doi.org/10.1007/978-3-030-50726-8_80