From Narratives to Portraits: Automating Elderly Portraits Construction through Named Entity Recognition with GPT-3.5

Anonymous

Abstract: Elderly portraits, descriptive summaries of older individuals, aid caregivers in offering personalized care. However, the manual construction of these portraits is time-consuming and challenging. This paper introduces an automated framework for constructing elderly portraits with event elements. The primary objective is to efficiently extract relevant features from elderly narratives. Traditional named entity recognition (NER) methods often falter due to data limitations and the inherent complexity of the stories. To address this, we present EPNER (Elderly Portrait Named Entity Recognition), a NER approach leveraging in-context learning with large language models. Our experimental results confirm that EPNER surpasses existing techniques.

Paper Type: long

Research Area: NLP Applications

Languages Studied: Chinese