1,721,025 research outputs found
Identifying the features of ProVax and NoVax groups from social media conversations
No full textDifferent studies state that doubts and fears posted on social media are making people hesitant towards vaccines and this hesitancy causes people to post doubts and fears. To break this vicious circle, we propose a method based on psycho-linguistics and time-domain analyses of social media conversations that talk about vaccines and vaccinations. The idea is to identify psycho-linguistics signals of distrust towards vaccines in order to help health authorities to restore the trust toward vaccines. The method allows the implementation of a well-known strategy used in marketing to restore the trust towards a brand when negative comments are posted on social media. The proposal has been evaluated using a two-year Facebook dataset composed of more than 170 thousand posts written by ProVax and NoVax people. The obtained results show that our method might be helpful to define a crisis communication plan to get people to trust vaccines, vaccinations and healthcare institutions again
Recommendation Systems: Issues, Challenges and Regulations
No full textThe European Parliament's ratification of the Artificial Intelligence Act in March 2024 represents a crucial step in regulating AI technologies. Aimed at mitigating risks, enhancing opportunities, and promoting transparency and democratic values, the Act targets AI practices with potential societal implications. Notably, it imposes strict risk assessment processes, particularly on high-risk systems like recommendation systems. These systems permeate various aspects of modern life, influencing everything from entertainment to politics. Operating by collecting personal data to generate personalized suggestions, they raise concerns about privacy, discrimination, and manipulation. This paper examines the literature to explore the risks posed by recommendation systems, explores challenges, and highlights regulations initiatives
On Using Twitter to Understand the Stablecoin Terra Collapse
No full textStablecoins have emerged as a solution to the legal status and volatile exchange rate issues surrounding cryptocurrencies like Bitcoin. Terra is a blockchain protocol that enables the creation of stablecoins pegged to fiat currencies, including UST and KRT, which have gained popularity for their use in DeFi applications. However, in May 2022, TerraUSD (UST) lost its peg to the USD, collapsing to a value of 0.091 USD due to a massive withdrawal of UST from the Anchor Protocol, leading to a sharp drop in price. In this paper we analyze people's opinions and thoughts about the topic using Twitter data and we propose a method to identify the accounts that acted as opinion leaders. We observed an increase in negativity in the month leading up to the collapse and identified the United States as the most prolific country in terms of tweets; we identified opinion leaders accounts and we showed that the number of followers is not an important metric to identify opinion leaders. The failure of the Terra project highlights the intrinsic fragility of algorithmic stablecoins, which can fail due to sudden fluctuations in demand and supply, and their potential for success and stability remains uncertain
Automatic Music Playlist Generation Based on Music-Programming of FM Radios
No full textStreaming music services are flooding their platforms with thousands of different playlists and this huge catalog is backfiring users who struggle to find the playlist that best suits their needs. In this paper, we aim to facilitate the music listening by designing a novel approach to automatically generate a playlist suited to the user's musical taste. Our approach is based on the FM radio music programming: starting from the user's favorite radio, we analyze seven days of music programming, we transform songs into vectors of audio features, we generate a music programming for a virtual day, and we transform this virtual day music programming into a real playlist when the user begins the music playout
Automatic and smart content transformation of video lectures
No full textThe lockdown caused by the Covid-19 pandemic has forced many educational institutions around the world to produce video lectures in order to support their students. A popular approach was to produce video lectures with a generic layout without considering neither the type of student nor the type of device used by the student. This approach complicated the learning process (e.g., students equipped with mobile devices and limited bandwidth connections have been forced to watch video produced for large screens and infinite bandwidth availability). In this paper, we investigate if it is possible to automatically transform the original video lecture to produce smartphone suitable videos and we involve students to understand the viewing experience. We consider the video lectures available within the ONELab University video lecture catalog and we design five different heuristics based on the semantic analysis of the video lectures. The experimental evaluation considers both the quantitative and qualitative aspects and the obtained results show that it is possible to save more than 90% of the bandwidth while maintaining a viewing experience equals to the one of the original video lectures
Social music discovery: an ethical recommendation system based on friend’s preferred songs
Full text linkUnderstanding users music listening habits for time and activity sensitive customized playlists
No full textMusic playlists have rapidly become one of the top services of streaming platforms: users do not need to spend time into deciding what to listen to, they just have to select the proper pre-compiled playlists composed by some music they know and some new songs close to their taste. However, users will continue to listen to playlists only if these conform to their musical tastes. Indeed, users listening habits might differ according to the activity they are performing while listening to music, and suggested playlists should reflect that. As there still are not in literature standard methods to produce customized playlists automatically, in this work in progress paper we focus on how to exploit users past music hearings to define genre listening habits in specific days of the week and hours of the day. We present our first investigation toward this direction by proposing a simple and computationally inexpensive method and by testing it using the Spotify listening history of volunteer users. We show that the method is promising and discuss several future working directions to improve the method and make it more effective
1st International workshop on social media sensing (SMS'19): message from the workshop chairs
No full textAutomatic and Personalized Sequencing of Music Playlists
No full textMusic playlists are appreciated by users, music artists and service providers for various reasons (i.e., no need to waste time choosing what to listen to, showcase to increase popularity, engage users to the provided services). However, despite their ever-increasing centrality, in literature there is no precise definition on how to produce them. Often, playlists are produced by music recommendation algorithms that focus on the songs selection process and don't give enough importance to songs sequencing. Indeed, until a few years ago the listening order was not considered important. In this paper, we address the songs sequencing problem in a novel way. Through dynamic programming, we transform a set of non-ordered songs into a user-tailored sequence of songs that meets the user's musical preferences. To the best of our knowledge, this approach has never been used in the literature
- …
