Gracia Urzelai, SaraOronoz Antxordoki, MaitePerez Ramirez, Alicia2024-11-272024-11-27production.46773https://dx.doi.org/10.26876/ikergazte.v.03.16https://gordailua.ueu.eus/handle/123456789/2673Suizidioa gizartearen kezka nagusietako bat bilakatu da azken urteetan. Gainera, sare sozialak gure eguneroko-tasunaren parte bilakatu dira, eta emozioak adierazteko erabiltzen dira askotan. Lan honetan sailkapen bitarraburutu da Reddit sare sozialeko mezu baten edukia suizidioarekin erlazionatua dagoen ala ez erabakitzeko.Alde batetik, artearen egoerako sistema gainbegiratuei dagokienean, ELECTRA transformerrarekin lortu daasmatze-tasarik altuena, %97,9koa. Bestalde, ondorioztatu da LDA topiko-ereduak sortutako errepresentazioakbaliagarriak izan daitezkeela ataza honetan, eta hau frogatzeko oinarri-lerroa den sailkatzailea proposatu da,%83,3ko asmatze-portzentaia izan duena 5 topiko erabiliz.Suicide has become one of society’s main concerns in recent years. In addition, social media has become partof our everyday life and is often used to express emotions. In this work, a binary classification has been carriedout to determine whether or not the content of a message on the Reddit social network is related to suicide. Onthe one hand, with regard to supervised systems in the state of the art, the best performance has been achievedwith the ELECTRA transformer, with an accuracy-rate of 97.9%. On the other hand, it has been concluded thatthe representations produced by the LDA topic-model can be useful for this task, and to prove this, a baselineclassifier has been proposed, which has reached an accuracy of 83.3%.hizkuntzaren prozesamenduasuizidio ideiagintzaren detekzioatopiko-ereduakikasketa sakonanatural language processingsuicidal ideation detectiontopic modelsdeep learningHizkuntzalaritzaInformatikaIngeniaritzaPsikologiaintroduction