V ponedeljek, 22. junija 2026, bodo ob 16:00 uri izvedeni dve 
predavanji v okviru PONEDELJKOVEGA SEMINARJA RAČUNALNIŠTVA IN INFORMATIKE
Oddelkov za Informacijske znanosti in tehnologije UP FAMNIT in UP IAM.

ČAS/PROSTOR: 22. junij 2026 ob 16.00 prek Zoom-a.

1. predavanje:
============

-----------------------------------------
PREDAVATELJ: Jovan VUKOVIĆ
-----------------------------------------

Jovan Vuković is a Master's student in Data Science at the University of Primorska (UP FAMNIT). He previously completed a Bachelor's degree in Computing and Information Technology at the University of Montenegro. His professional background is primarily focused on software engineering, microservice architectures, distributed systems, and large-scale backend development. His current research interests include knowledge graphs, machine learning, data integration, and scalable graph processing.

--------------------------------------------------------------------------------------------------------------------------------------
NASLOV: Embedding-Based Entity Alignment between Large-Scale Knowledge Graphs: A Case Study of YAGO and OpenAlex
--------------------------------------------------------------------------------------------------------------------------------------

POVZETEK:

Knowledge graphs have become an important approach for representing and integrating structured information in domains such as web search, digital libraries, recommendation systems, and scientific knowledge management. By representing entities and their relationships as graphs, they enable the discovery of connections that are difficult to capture using traditional relational data models. As the number and size of available knowledge graphs continue to grow, combining information from multiple sources has become an increasingly important challenge.
One of the key problems in this context is entity alignment, the task of identifying entities in different knowledge graphs that refer to the same real-world object. Solving this problem is essential for building integrated knowledge bases, but it becomes particularly challenging when working with large-scale graphs containing hundreds of millions of entities and relationships.
This presentation introduces the entity alignment problem through a case study involving the YAGO and OpenAlex knowledge graphs. It focuses on the first stages of a scalable alignment pipeline, including large-scale RDF preprocessing, extraction of structural and textual information, graph transformation and integer encoding, and the generation of knowledge graph embeddings using PyTorch-BigGraph.
Several embedding models, including TransE, DistMult, and ComplEx, are evaluated through link prediction experiments on large-scale datasets. The obtained results are used to assess how well the learned embeddings capture the underlying graph structure and to identify the most suitable embedding model for the subsequent entity alignment task.

Seminar bo potekal v angleškem jeziku.

===============================================================================

2. predavanje:
============

--------------------------------------
PREDAVATELJ: Uroš SERGAŠ
--------------------------------------

Uroš Sergaš is a teaching assistant of Computer Science at UP FAMNIT and a member of the Centre for Responsible AI UP. As a researcher specializing in recommender systems and computational social science, his work focuses on applying machine learning methods to address societal challenges. Recently, his work also addresses the issue of AI alignment. He is currently pursuing his PhD in recommender systems.

--------------------------------------------------------------------------------------------------------------------------------------
NASLOV: Can News Ranking Reduce Political Polarization? Evidence from an Online Experiment with LLM-Generated News
--------------------------------------------------------------------------------------------------------------------------------------

POVZETEK:

Algorithmic news ranking and large language models (LLMs) increasingly mediate how citizens receive political information. While personalization is often criticized for reinforcing echo chambers, ranking has also been proposed as a lever for depolarization by shaping exposure to cross-cutting viewpoints. We tested this claim in an online experiment (N=100, Prolific) using five LLM-generated news articles on gun legislation spanning pro–gun freedom to pro–gun control. All participants read the same articles; the only manipulation was article order, instantiated as a random baseline and six stance-aware depolarization-oriented strategies (counter-narrative sandwich, balanced alternation, and directional gradients). Pre–post questionnaires measured ideological selfplacement (feeling thermometer) and affective evaluations of gun-control and gun-freedom advocates. Across six research questions, we tested to see whether such recommender strategies had any reduction in ideological or affective polarization.

Seminar bo potekal v angleškem jeziku.

===============================================================================

Seminarja bosta potekala online prek aplikacije Zoom s pričetkom ob 16:00 uri na sledeči povezavi:

https://upr-si.zoom.us/j/297328207?pwd=S3Zpdk1VR3pjckNtWkQwKzlvcDR5UT09

Meeting ID: 297 328 207
Passcode: 123456789

Vabljeni!