BERT-based Ensembles for Modeling Disclosure and Support in Conversational Social Media Text
BERT-based Ensembles for Modeling Disclosure and Support in Conversational Social Media Text
复制标题
基于 BERT 的集成，用于对话式社交媒体文本中的建模披露和支持
DOI：
--
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
R. Mamidi
中科院分区：
文献类型：
--
作者：
Kartikey Pant;Tanvi Dadu;R. Mamidi
人们越来越有兴趣了解人类如何发起和进行对话。会话的情感理解关注的是说话者如何利用情感对情景和彼此做出反应的问题。在CL-Aff共享任务中，组织者发布了Get it #OffMyChest数据集，其中包含来自休闲和忏悔对话的Reddit评论，标记为披露和保密特征。在本文中，我们介绍了一个预测集成模型，利用微调上下文的词嵌入，RoberTa和ALBERT。我们表明，我们的模型在所有考虑的指标优于基础模型，实现了3\%$的F1得分的改善。我们进一步进行统计分析，并概述对给定数据集的更深入见解，同时为数据集提供新的影响特征。
There is a growing interest in understanding how humans initiate and hold conversations. The affective understanding of conversations focuses on the problem of how speakers use emotions to react to a situation and to each other. In the CL-Aff Shared Task, the organizers released Get it #OffMyChest dataset, which contains Reddit comments from casual and confessional conversations, labeled for their disclosure and supportiveness characteristics. In this paper, we introduce a predictive ensemble model exploiting the finetuned contextualized word embeddings, RoBERTa and ALBERT. We show that our model outperforms the base models in all considered metrics, achieving an improvement of $3\%$ in the F1 score. We further conduct statistical analysis and outline deeper insights into the given dataset while providing a new characterization of impact for the dataset.