20.500.12556/RUL-176823 Pojasnjevalni priporočilni sistem za aktivnosti starejših uporabnikov An explainable recommendation system for older users' activities V magistrskem delu obravnavamo razvoj razložljivega priporočilnega sistema. Izhodišče je demografsko staranje prebivalstva in potreba po tehnoloških rešitvah, ki podpirajo zdravo in aktivno staranje ter so hkrati prilagojene posebnostim starejših uporabnikov. Obstoječi priporočilni sistemi pogosto delujejo kot "črne skrinje", kar v občutljivih domenah, kot je skrb za starejše, predstavlja oviro zaradi pomanjkanja transparentnosti in zaupanja. Namesto klasičnih črnih skrinj, problem priporočanja formuliramo kot nalogo napovedovanja ocen na tabelaričnih podatkih. Raziskujemo, kako različni sklopi kontekstualnih, demografskih in vedenjskih spremenljivk vplivajo na natančnost napovedi in berljivost pojasnil modela. Za reševanje problema smo uporabili podatke nacionalne raziskave Inštituta Antona Trstenjaka na vzorcu 1.047 oseb, starejših od 50 let ter vključuje validirane vprašalnike o zdravju,aktivnostih, odnosih in stališčih. Na osnovi konstruktov konstruiramo ocenjevalno matriko uporabnik–aktivnost in šest konfiguracij značilk: osnovni model (samo userID, itemID), model s trenutnim kontekstom aktivnosti, model z razredi konteksta, demografski model, model z vedenjsko-nagnitvenimi dejavniki ter model z uporabniškimi skupinami. Za vse konfiguracije treniramo regresijski model XGBoost z uporabniško razdelitvijo na učni in testni del, uspešnost pa merimo z MAE in RMSE. Za razlago napovedi uporabimo ogrodje SHAP, ki omogoča globalne in lokalne razlage prispevkov posameznih značilk. Na podlagi SHAP vrednosti uvedemo tudi mero entropije pojasnil, ki kvantitativno oceni, kako koncentrirana ali razpršena je razlaga. Rezultati eksperimentov so pokazali, da vključitev kontekstualnih informacij bistveno izboljša napovedno točnost modela v primerjavi z osnovnim modelom. Najboljši rezultat je dosegel model z vključenim trenutnim kontekstom aktivnosti, ki je zmanjšal srednjo absolutno napako (MAE) z 1,1393 na 0,6031, kar potrjuje, da so preference starejših uporabnikov močno odvisne od situacije. Analiza entropije razlage je razkrila, da vključitev vedenjskih in stališčnih dejavnikov vodi do najbolj osredotočenih in s tem človeku najbolj razumljivih razlag Demografske značilnosti in grobe uporabniške skupine prispevajo k stabilnosti razlag, čeprav je bil njihov vpliv na točnost manjši. Zaključujemo, da je mogoče s kombinacijo gradientnega povečevanja dreves (XGBoost), SHAP-pojasnil in entropije pojasnil zgraditi učinkovit sistem priporočanja, ki hkrati dosega dobro napovedno natančnost in generira kratka, razumljiva pojasnila. Takšen sistem je primeren za podporo zdravemu staranju, saj omogoča, da strokovnjaki, skrbniki in sami starejši uporabniki razumejo, zakaj je bila predlagana določena aktivnost, ter lažje presodijo primernost priporočila. This master's thesis addresses the development of an explainable recommendation system. The starting point is the demographic aging of the population and the need for technological solutions that support healthy and active aging while being tailored to the specific needs of older users. Existing recommender systems often operate as "black boxes," which in sensitive domains, such as elderly care, poses an obstacle due to a lack of transparency and trust. Instead of classic black boxes, we formulate the recommendation problem as a rating prediction task on tabular data. We investigate how various sets of contextual, demographic, and behavioral variables influence the prediction accuracy and the readability of the model’s explanations. To solve the problem, we used data from the national survey of the Anton Trstenjak Institute, based on a sample of 1,047 individuals over 50 years of age, which includes validated questionnaires on health, activities, relationships, and attitudes. Based on these constructs, we build a user–activity rating matrix and six feature configurations: the baseline model (only userID, itemID), a model with current activity context, a model with context classes, a demographic model, a model with behavioral and attitudinal factors, and a model with user groups. For all configurations, we train an XGBoost regression model with a user-based split into training and test sets, measuring performance with MAE and RMSE. For prediction explanation, we use the SHAP framework, which allows for global and local explanations of individual feature contributions. Based on SHAP values, we also introduce the measure of explanation entropy to quantitatively assess how concentrated or dispersed the explanation is. Experimental results showed that the inclusion of contextual information significantly improves the model’s prediction accuracy compared to the baseline model. The best result was achieved by the model including the current activity context, which reduced the Mean Absolute Error (MAE) from 1.1393 to 0.6031, confirming that the preferences of older users are highly situation-dependent. The analysis of explanation entropy revealed that the inclusion of behavioral and attitudinal factors leads to the most focused and thus the most humanly understandable explanations. Demographic features and broad user groups contribute to the stability of the explanations, although their impact on accuracy was smaller. We conclude that by combining Gradient Boosting Trees (XGBoost), SHAP explanations, and explanation entropy, it is possible to build an effective recommendation system that simultaneously achieves good predictive accuracy and generates brief, understandable explanations. Such a system is suitable for supporting healthy aging, as it allows professionals, caregivers, and older users themselves to understand why a certain activity was suggested and to more easily assess the appropriateness of the recommendation. priporočilni sistemi razložljiva umetna inteligenca (XAI) SHAP zdravo staranje starejši odrasli entropija pojasnil kontekstno odvisni sistemi aktivno staranje recommender systems explainable artificial intelligence (XAI) SHAP healthy ageing older adults explanation entropy context-aware systems active ageing true false false Slovenski jezik Angleški jezik Magistrsko delo/naloga 2025-12-11 11:45:02 2025-12-11 11:45:20 2026-01-20 10:27:09 0000-00-00 00:00:00 2025 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 63098 265356803 Trebovac_Anja_-_Pojasnjevalni_priporocilni_sistem_za_aktivnosti_starejsih_uporabnikov.pdf Trebovac_Anja_-_Pojasnjevalni_priporocilni_sistem_za_aktivnosti_starejsih_uporabnikov.pdf 1 DA118D919E0F16401735BD52849869EC fff57fb5a557032268872cb05a53efb7737e3bb3c9a66e30b4249947bd12c4a4 131c2cf1-d67e-11f0-9328-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=223418 Fakulteta za elektrotehniko 0 0 0