Cognitive Emotional Embedded Representations of Text to Predict Suicidal Ideation and Psychiatric Symptoms

摘要: Mathematical modeling of language in Artificial Intelligence is the utmost importance for many research areas and technological applications. Over last decade, on text representation has been directed towards investigation dense vectors popularly known as word embeddings. In this paper, we propose a cognitive-emotional scoring framework based This aims to mathematically model emotional content words short free-form messages, produced by adults follow-up due any mental health condition outpatient facilities within Psychiatry Department Hospital Fundacion Jimenez Diaz Madrid, Spain. Our contribution geometrical-topological Sentiment Analysis, that includes hybrid method uses cognitively-based lexicon together with embeddings generate graded sentiment scores words, new topological clustering vector representations high-dimensional spaces, where points are very sparsely distributed. useful detecting association topics, patterns, embedded vectors’ geometrical behavior, which might be understanding use kind texts. proposed system helpful studying relations between behavior their have predictive potential prevent suicide.