20.500.12556/RUL-184372 Razvoj sistema za avtomatizirano generiranje tehničnih poročil iz nadzornih sistemov Development of a System for Automated Generation of Technical Reports from Monitoring Systems Aplikacija AviChron naslavlja izziv ročnega, časovno potratnega in pogosto nekonsistentnega tehničnega poročanja v sodobnih informacijskih okoljih. Predlagana rešitev združuje infrastrukturne metrike iz nadzornega sistema Zabbix ter podatke o incidentih iz platforme ServiceNow in jih dodatno obogati z lokalno nameščenim velikim jezikovnim modelom (LLM) GPT-OSS s 120 milijardami parametrov. Sistem avtomatizira zajem, korelacijo in interpretacijo podatkov ter na tej osnovi generira standardizirana poročila v formatih PDF, DOCX ali XLSX. Sistem smo empirično ovrednotili z vključitvijo 30 sodelujočih v evalvacijo. Rezultati kažejo, da se je ocenjen čas priprave primerljivega poročila z več ur skrajšal na približno 35 minut. Rešitev je posebej primerna za okolja z visokimi varnostnimi zahtevami, kjer lokalna obdelava podatkov in lokalno nameščen LLM zagotavljata popoln nadzor nad podatkovnimi tokovi. The AviChron application addresses the challenge of manual, time-consuming, and often inconsistent technical reporting in modern information environments. The proposed solution consolidates infrastructure metrics from the Zabbix monitoring system and incident records from the ServiceNow ITSM platform, enriching them using a locally deployed open-weight language model gpt-oss with 120 billion parameters. The system automates data acquisition, correlation, and interpretation, generating standardized reports in PDF, DOCX, or XLSX formats. This approach reduces the administrative workload of operations teams, improves report comparability over time, and strengthens compliance by ensuring that operational data is not sent to external LLM services and that model inference remains within the organization’s infrastructure. The system was evaluated with 30 participants. Results indicate that the estimated time for preparing a comparable report decreased from several hours to approximately 35 minutes. The solution is particularly suitable for security-sensitive environments, as locally deployed data processing and a locally hosted language model ensure complete control over data flows. tehnično poročanje Zabbix ServiceNow umetna inteligenca lokalni LLM technical reporting Zabbix ServiceNow artificial intelligence local LLM true false false Slovenski jezik Angleški jezik Diplomsko delo/naloga 2026-07-06 09:45:07 2026-07-06 09:45:13 2026-07-24 06:26:13 0000-00-00 00:00:00 2026 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 38726 285902595 Haziri_Egzon_-_Razvoj_sistema_za_avtomatizirano_generiranje_tehnicnih_porocil_iz_nadzornih_siste.pdf Haziri_Egzon_-_Razvoj_sistema_za_avtomatizirano_generiranje_tehnicnih_porocil_iz_nadzornih_siste.pdf 1 343A99793C97FF951FC3CB4FC297EFD2 c8ee620e453ff3896a48280775bab857122a407badf05ac1f11b56115e439866 817d1d3d-790e-11f1-9b0d-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=237055 Fakulteta za računalništvo in informatiko 0 0 0