A survey on extraction of causal relations from natural language text

Abstract

As an essential component of human cognition, cause–effect relations appear frequently in text, and curating cause–effect relations from text helps in building causal networks for predictive tasks. Existing causality extraction techniques include knowledge-based, statistical machine learning (ML)-based, and deep learning-based approaches. Each method has its advantages and weaknesses. For example, knowledge-based methods are understandable but require extensive manual domain knowledge and have poor cross-domain applicability. Statistical machine learning methods are more automated because of natural language processing (NLP) toolkits. However, feature engineering is labor-intensive, and toolkits may lead to error propagation. In the past few years, deep learning techniques attract substantial attention from NLP researchers because of its powerful representation learning ability and the rapid increase in computational resources. Their limitations include high computational costs and a lack of adequate annotated training data. In this paper, we conduct a comprehensive survey of causality extraction. We initially introduce primary forms existing in the causality extraction: explicit intra-sentential causality, implicit causality, and inter-sentential causality. Next, we list benchmark datasets and modeling assessment methods for causal relation extraction. Then, we present a structured overview of the three techniques with their representative systems. Lastly, we highlight existing open challenges with their potential directions.

Extraction of Explicit and Implicit Cause-Effect Relationships in Patient-Reported Diabetes-Related Tweets From 2017 to 2021: Deep Learning Approach

“…To intervene and potentially prevent diabetes distress, it is necessary to understand the causes of diabetes distress from a patient’s perspective to understand how patients see their disease. Causal relation extraction in natural language text has gained popularity in clinical decision-making, biomedical knowledge discovery, or emergency management [ 11 ]. In particular, causal relations on Twitter have been examined for diverse factors causing stress and relaxation [ 12 ], adverse drug reactions [ 13 ], or causal associations related to insomnia or headache [ 14 ].…”

“…In particular, causal relations on Twitter have been examined for diverse factors causing stress and relaxation [ 12 ], adverse drug reactions [ 13 ], or causal associations related to insomnia or headache [ 14 ]. Most approaches examine explicit causality in text [ 14 - 16 ], when cause and effect are explicitly stated, for instance, by connective words (eg, so, hence, because, lead to, since, if-then) [ 11 , 17 ]. An example for an explicit cause-effect pair is “diabetes causes hypoglycemia.” However, implicit causality is more complicated to detect such as in “I reversed diabetes with lifestyle changes” with cause “lifestyle changes” and effect “reversed diabetes.”…”

“…In combination with machine learning and deep learning models, which are artificial intelligence algorithms designed to learn from experience, they have also been applied to extract causal relations [ 18 , 19 ]. Machine learning methods are able to explore implicit relations and provide better generalization contrary to rule-based approaches [ 11 , 20 - 22 ]. An interesting approach leveraging the transfer learning paradigm and addressing both explicit and implicit cause-effect extraction is provided by Khetan et al [ 23 ].…”

The Causal News Corpus: Annotating Causal Relations in Event Sentences from News

“…Causal Question Answering and Generation applications (Dalal et al, 2021;Hassanzadeh et al, 2019;Stasaski et al, 2021) are some immediate downstream Natural Language Understanding (NLU) applications popular in NLP today. Despite the importance of identifying causality in text, datasets are limited (Asghar, 2016;Xu et al, 2020;Tan et al, 2021;Yang et al, 2021) and oftentimes, different researchers craft their datasets with different rules, leaving users with no proper way to compare models across datasets. Our work is directed at annotating parts of the multilingual protest news detection dataset (Hürriyetoglu et al, 2021a;Hürriyetoglu et al, 2021b;Hürriyetoglu, 2021) for event causality.…”

Fine-Grained Causality Extraction From Natural Language Requirements Using Recursive Neural Tensor Networks

“…More specifically, the current state of research has the following gaps: P1: Rule-based systems [6], [7], [8] are highly dependent on hand-crafted patterns. For this reason, more recent approaches [9], [10] apply Deep Learning (DL) in order to automatically extract useful features from raw text. However, the existing approaches [11], [12], [13] have been trained on corpora not originating from software engineering (e.g., BBC news article set [14]) and are therefore difficult to utilize for RE purposes.…”