| J. S. Hammer, K. Theodoropoulos, G. Drakopoulos, Ph. Mylonas, G. Bardis |
| Twitter recommendations across dialects: Seeing the big board through Faiss |
| 20th International Workshop on Semantic and Social Media Adaptation & Personalization (SMAP 2025), 27-28 November 2025, Mystras, Greece |
|
ABSTRACT
|
| Web and social media personalization depends on a broad spectrum of factors including language, which influences all spheres of social life. Most notably, virtually every action of informal human communication online is almost exclusively actualized by language. As it largely remains the main premise upon which such communicative acts unfold freely, language is subject to a myriad of minor changes eventually accumulating to discernible patterns. Since the latter are discrete, they can be ultimately expressed as graphs indicating lexical, syntactic, or semantic changes in a particular cultural context. In turn, the set of all such graphs over a time frame of reference for each language under examination can be considered as a basic pattern collection. Thus, it is statutory to ask whether a pattern subset taken from a given account can serve in building a comprehensive personalized profile and, if so, whether the aforementioned profile is lightweight enough to be applied at the Web scale. A large number of linguistic patterns has been collected after an extensive bibliography review and probabilistic graph operations are subsequently shown how they can build language and account profiles.
|
| 27 November , 2025 |
| J. S. Hammer, K. Theodoropoulos, G. Drakopoulos, Ph. Mylonas, G. Bardis, "Twitter recommendations across dialects: Seeing the big board through Faiss", 20th International Workshop on Semantic and Social Media Adaptation & Personalization (SMAP 2025), 27-28 November 2025, Mystras, Greece |
[  PDF] [
BibTex] [
Print] [
 Back] |
