• Digital Humanities and Cultural Heritage
• 2021-2021close
• Open Access close
• Publications close
• SoBigData close

The Covid-19 pandemic has had a deep impact on the lives of the entire world population, inducing a participated societal debate. As in other contexts, the debate has been the subject of several d/misinformation campaigns; in a quite unprecedented fashion, however, the presence of false information has seriously put at risk the public health. In this sense, detecting the presence of malicious narratives and identifying the kinds of users that are more prone to spread them represent the first step to limit the persistence of the former ones. In the present paper we analyse the semantic network observed on Twitter during the first Italian lockdown (induced by the hashtags contained in approximately 1.5 millions tweets published between the 23rd of March 2020 and the 23rd of April 2020) and study the extent to which various discursive communities are exposed to d/misinformation arguments. As observed in other studies, the recovered discursive communities largely overlap with traditional political parties, even if the debated topics concern different facets of the management of the pandemic. Although the themes directly related to d/misinformation are a minority of those discussed within our semantic networks, their popularity is unevenly distributed among the various discursive communities. Comment: 29 pages, 11 figures

Obtaining high-quality labelled data for training a classifier in a new application domain is often costly. Transfer Learning (a.k.a. “Inductive Transfer”) tries to alleviate these costs by transferring, to the “target” domain of interest, knowledge available from a different “source” domain. In transfer learning the lack of labelled information from the target domain is compensated by the availability at training time of a set of unlabelled examples from the target distribution. Transductive Transfer Learning denotes the transfer learning setting in which the only set of target documents that we are interested in classifying is known and available at training time. Although this definition is indeed in line with Vapnik’s original definition of “transduction”, current terminology in the field is confused. In this article, we discuss how the term “transduction” has been misused in the transfer learning literature, and propose a clarification consistent with the original characterization of this term given by Vapnik. We go on to observe that the above terminology misuse has brought about misleading experimental comparisons, with inductive transfer learning methods that have been incorrectly compared with transductive transfer learning methods. We then, give empirical evidence that the difference in performance between the inductive version and the transductive version of a transfer learning method can indeed be statistically significant (i.e., that knowing at training time the only data one needs to classify indeed gives an advantage). Our clarification allows a reassessment of the field, and of the relative merits of the major, state-of-the-art algorithms for transfer learning in text classification.

My First article revealed about Fifth Force or New Force discovered by Fermi-Lab or LHC etc. other than Strong, Weak, Electromagnetic & Gravitational force published in the Journal of Calcutta Mathematical Society, Salt Lake, Kolkata, West Bengal, India who received this article on 2007 but published on 2011.

We study an important fact called Mind & Sense Datum (matter based). Considering these emerges by the photoelectric effects conducted by the photon-like energy packets “narayan” including photons or quanta of ordinary light through electromagnetic interactions of the matter nuclei weak energy forces of SU(2) in the respective framework of SU(2) ´ U(1), also interacted with bio-molecular particles constructed by the various combinations of lepton-like but quark-type [assuming six set of quark-types each set having five different quark-type] of exotic matter fluids in wave status are tightly binding by the bosons of SU(6) combining with quarks of usual matter energy in brain neuron microtubules for the case of human or any other suitable places of brain-likes of lives. We then found several new unknown Particles-Likes “bhadras” with a new kind of strong forces those are combining with quarks of ordinary matter particles formed Bio-Molecules through chemical bonding of matter atoms or polymers etc. are internally linked through photoelectric-like current for exchanging the necessary information between all parts of the body-system. Hence created a living-body-system of the whole system link through laser-like beam of new energies etc. inclinable to the photo-electrodes-like tube in brain cell for the case of Human-likes and then created holography mind with gradually unfolding sense datum. Thus human brain conceived as an interfacing organ that not only produces mind through consciousness but also received instructional information. Considering there possible variable wave frequencies packets of light constructed by photon-like [made up from neutrinos-like of the new energies SU(6), SU(12), SU(24) etc.] quanta in wave status "narayan" within our Physical Universe produce consciousness of us then by photoelectric effects created mind with sense datum.

The recent advances in language modeling significantly improved the generative capabilities of deep neural models: in 2019 OpenAI released GPT-2, a pre-trained language model that can autonomously generate coherent, non-trivial and human-like text samples. Since then, ever more powerful text generative models have been developed. Adversaries can exploit these tremendous generative capabilities to enhance social bots that will have the ability to write plausibledeepfakemessages, hoping to contaminate public debate. To prevent this, it is crucial to develop deepfake social media messages detection systems. However, to the best of our knowledge no one has ever addressed the detection of machine-generated texts on social networks like Twitter or Facebook. With the aim of helping the research in this detection field, we collected the first dataset ofrealdeepfake tweets,TweepFake. It isrealin the sense that each deepfake tweet was actually posted on Twitter. We collected tweets from a total of 23 bots, imitating 17 human accounts. The bots are based on various generation techniques, i.e., Markov Chains, RNN, RNN+Markov, LSTM, GPT-2. We also randomly selected tweets from the humans imitated by the bots to have an overall balanced dataset of 25,572 tweets (half human and half bots generated). The dataset is publicly available on Kaggle. Lastly, we evaluated 13 deepfake text detection methods (based on various state-of-the-art approaches) to both demonstrate the challenges that Tweepfake poses and create a solid baseline of detection techniques. We hope that TweepFake can offer the opportunity to tackle the deepfake detection on social media messages as well.

Twitter is among the most used online platforms for the political communications, due to the concision of its messages (which is particularly suitable for political slogans) and the quick diffusion of messages. Especially when the argument stimulate the emotionality of users, the content on Twitter is shared with extreme speed and thus studying the tweet sentiment if of utmost importance to predict the evolution of the discussions and the register of the relative narratives. In this article, we present a model able to reproduce the dynamics of the sentiments of tweets related to specific topics and periods and to provide a prediction of the sentiment of the future posts based on the observed past. The model is a recent variant of the P\'olya urn, introduced and studied in arXiv:1906.10951 and arXiv:2010.06373, which is characterized by a "local" reinforcement, i.e. a reinforcement mechanism mainly based on the most recent observations, and by a random persistent fluctuation of the predictive mean. In particular, this latter feature is capable of capturing the trend fluctuations in the sentiment curve. While the proposed model is extremely general and may be also employed in other contexts, it has been tested on several Twitter data sets and demonstrated greater performances compared to the standard P\'olya urn model. Moreover, the different performances on different data sets highlight different emotional sensitivities respect to a public event. Comment: 19 pages, 12 figures