DICE @ TREC-IS 2018: Combining Knowledge Graphs and Deep Learning to Identify Crisis-Relevant Tweets

DICE @ TREC-IS 2018: Combining Knowledge Graphs and Deep Learning to Identify Crisis-Relevant Tweets
复制标题

DICE @ TREC-IS 2018:结合知识图和深度学习来识别与危机相关的推文

DOI:
--
复制
发表时间:
2018
期刊:
Text Retrieval Conference
影响因子:
--
通讯作者:
Ricardo Usbeck
Ricardo Usbeck
中科院分区:
--
文献类型:
--
作者:
Hamada M. Zahera;Rricha Jalota;Ricardo Usbeck

文献摘要

被引文献

相似文献

.在本文中,我们描述了我们提交的TREC事件流(TREC-IS)挑战2018。我们研究了不同的机器学习方法,将危机相关的推文分类为不同的信息类型。除了词袋、词嵌入、时间数据和事件类型之外,我们还将知识图作为特征纳入到这种社交媒体分析中。此外,我们评估了31个生成的特征集的最新分类模型。我们的TREC-IS结果表明,基于组合知识图的模型(即,Babelfy),词嵌入和文本特征优于经典的机器学习模型。
. In this paper, we describe our submissions to the TREC Incident Stream (TREC-IS) challenge 2018. We investigated di ff erent machine learning approaches to classify crisis-related tweets into di ff erent information types. We incorporated knowledge graphs as features into this social media analysis, in addition to bag of words, word embeddings, time data, and event-types. Further, we evaluate state-of-the-art classification models on 31 generated features sets. Our TREC-IS re-sults indicate that a model based on combining knowledge graphs (i.e., Babelfy), word embeddings and textual features outperformes classical machine learning models.