Vol.:(0123456789) 1 3 Evolving Systems https://doi.org/10.1007/s12530-020-09348-z ORIGINAL PAPER Discovering sentiment potential in Twitter conversations with Hilbert–Huang spectrum Georgios Drakopoulos 1  · Andreas Kanavos 2  · Phivos Mylonas 1  · Spyros Sioutas 1,2 Received: 15 November 2019 / Accepted: 18 June 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020 Abstract Does a tweet with specifc emotional content posted by an infuential account have the capability to shape or even completely alter the opinions of its readers? Moreover, can other infuential accounts further enhance its original emotional potential by retweeting it and, thus, letting their followers read it? Real Twitter conversations seem to imply an afrmative answer to both questions. If this is indeed the case, then what is the key for not only successfully reaching to a large number of accounts but also for convincingly ofering an alternative perspective via afective means, therefore triggering a large scale opinion change in an ongoing Twitter conversation? This work primarily focuses on determining which tweets cause multiple sentiment polarity alternations to occur based on a window segmentation approach. Moreover, an ofine framework for discovering afective pivot points in a conversation based on its Hilbert–Huang spectrum, which has close ties to the Fourier transform, is introduced. Finally, given that it is highly desirable to track the sentiment shifts of a Twitter conversation while it unfolds, an adaptive scheme is presented for approximating the window sizes obtained by the ofine methodology. As a concrete example, the abovementioned methodologies are applied to three recent long Twitter discussions and the results are analyzed. Keywords Opinion polarity · Functional analytics · Emotional infuence · Social media analytics · Topic sampling · Signal processing for social media · Fourier spectrum · Hilbert–Huang transform Mathematics Subject Classifcation 42A16 · 46E20 · 62H30 · 62P25 · 91C20 · 91C99 · 91D10 · 91D30 · 91D99 1 Introduction Microblogging platforms like Twitter are today the con- stantly changing melting pot of opinions typically shaped from and expressed in conversations about a broad array of subjects. Sharing thoughts and sentiments may well lead to a viral tweet, especially by accounts highly regarded by their respective communities, which may be able to change the collective sentiment of a conversation despite the restric- tion placed on the length of tweets. In fact, the latter may well be a major driver behind highly sentimental tweets, as there is barely sufcient space available for long, articulate arguments. Instead, terse and laconic tweets, conveying a substantial amount of information nonetheless, frequently appear as stated in Bollen et al. (2011). Therefore, harness- ing the emotional content in this enormously continuous and volatile Twitter stream is bound to reveal trending opinions about and reactions, which ultimately shape public attitude, to a wide array of phenomena ranging from online market- ing campaigns to political events as shown among other in Suttles and Ide (2013). A key factor towards discovering the dynamics of online public sentiment lies in identifying the evolving set of emotionally infuential accounts. This set may be evolving over time and depends heavily on the conversation topic as observed in Roberts et al. (2012). Typically, candidate infuential accounts include corporate accounts, verifed * Andreas Kanavos kanavos@ceid.upatras.gr Georgios Drakopoulos c16drak@ionio.gr Phivos Mylonas fmylonas@ionio.gr Spyros Sioutas sioutas@ceid.upatras.gr 1 Department of Informatics, Ionian University, Corfu, Greece 2 Computer Engineering and Informatics Department, University of Patras, Patras, Greece