Detecting and Explaining Causes From Text For a Time Series Event

  title={Detecting and Explaining Causes From Text For a Time Series Event},
  author={Dongyeop Kang and Varun Gangal and Ang Lu and Zheng Chen and Eduard H. Hovy},
Explaining underlying causes or effects about events is a challenging but valuable task. We define a novel problem of generating explanations of a time series event by (1) searching cause and effect relationships of the time series with textual data and (2) constructing a connecting chain between them to generate an explanation. To detect causal features from text, we propose a novel method based on the Granger causality of time series between features extracted from text such as N-grams… 

Figures and Tables from this paper

Identifying Predictive Causal Factors from News Streams

The Predictive Causal Graph (PCG) is presented, which allows to detect latent relationships between events mentioned in news streams and can be used to extract latent features from news streams, outperforming other graph-based methods in prediction error.

A survey of the extraction and applications of causal relations

A comprehensive survey of the work on the extraction of causal relations and their applications, while also discussing the nature of causation and its representation in text.

An Overview of Ways of Discovering Cause-Effect Relations in Text by Using Natural Language Processing

A survey on trends in this field related to understanding how linguistically causal dependencies can be expressed in the text, what patterns and models exist, which of them are more and less successful and why is presented.

Unsupervised Word Influencer Networks from News Streams

The efficacy of Word Influencer Networks, a graph framework to extract longitudinal temporal relationships between any pair of informative words from news streams is demonstrated by using it for unsupervised extraction of latent features for stock price prediction and obtain 2 orders lower prediction error compared to a similar causal graph based method.

Identification of Causal Dependencies by using Natural Language Processing: A Survey

The current research illustrates results of a survey of research papers on identification and extracting cause-effect relations from text using NLP and other techniques, showing that expression of cause- effect relations in text can be very different.

A Review of Dataset and Labeling Methods for Causality Extraction

The limitations and shortcomings of existing causality research field are surveyed from the aspects of basic concepts, extraction methods, experimental data, and la-bel methods, so as to provide reference for future research on causality extraction.

Homicidal Event Forecasting and Interpretable Analysis Using Hierarchical Attention Model

Through quantitative and qualitative evaluation, it is shown that the proposed hierarchical attention-based mechanism can successfully make predictions while also being interpretable, which can help make more informed decisions for social analysis.

Causal Pathway Extraction from Web-Board Documents

The proposed approach enables causal-pathway extraction with high accuracy from the documents and proposes transitive closure and a dynamic template to indicate and represent the implicit mediators with the explicit ones.

Causal Web Determination from Texts

The research aim is to determine a causal web from downloaded guru web-board documents. The causal web which benefits a diagnosis service assistant of a problem-solving system consists of several

Domain Adaptative Causality Encoder

The characteristics of dependency trees and adversarial learning are leveraged to address the tasks of adaptive causality identification and localisation and present a new causality dataset, namely MedCaus, which integrates all types of causality in the text.



Generating Event Causality Hypotheses through Semantic Relations

This work proposes a method of hypothesizing unseen event causalities from knownevent causalities extracted from the web by the semantic relations between nouns, and shows that this method outperforms a state-of-the-art hypothesis generation method.

Toward a Better Understanding of Causality between Verbal Events: Extraction and Analysis of the Causal Power of Verb-Verb Associations

This paper proposes a set of knowledge-rich metrics to learn the likelihood of causal relations between verbs, and generates a knowledge base which identifies three categories of verb pairs: Strongly Causal, Ambiguous, and Strongly Non-causal.

Causal Relation Extraction

This paper presents a supervised method for the detection and extraction of Causal Relations from open domain text, and identifies the syntactic patterns that may encode a causation, then uses Machine Learning techniques to decide whether or not a pattern instance encodes a causation.

Building a Corpus of Temporal-Causal Structure

A corpus of 1000 event pairs for both temporal and causal relations is annotated, focusing on a relatively frequent construction in which the events are conjoined by the word “and”, suggesting that simple grammatical cues may be insufficient for identifying the more difficult temporal and causals.

An Analysis of Causality between Events and its Relation to Temporal Information

An annotation framework to capture causality between events, inspired by TimeML, and a language resource covering both temporal and causal relations are presented, which are then used to build an automatic extraction system for causal signals and causal links between given event pairs.

Generating Natural Language Inference Chains

A new task is proposed that measures how well a model can generate an entailed sentence from a source sentence and takes entailment-pairs of the Stanford Natural Language Inference corpus and trains an LSTM with attention, and applies this model recursively to input-output pairs, thereby generating natural language inference chains.

Automatic Detection of Causal Relations for Question Answering

A method for the automatic detection and extraction of causal relations and an inductive learning approach to the automatic discovery of lexical and semantic constraints necessary in the disambiguation of causal Relations that are then used in question answering.

Creating Causal Embeddings for Question Answering with Minimal Supervision

This work argues that a better approach is to look for answers that are related to the question in a relevant way, according to the information need of the question, which may be determined through task-specific embeddings, and implements causality as a use case.

Human Judgements on Causation in French Texts

It is argued that the statistical association of features of causation from theoretical previous work with causation itself in human judgements in an annotation experiment leads to coherent annotation guidelines, since it allows for a κ = 0.84 agreement between the majority of the annotators answers and the authors' own educated judgements.

Cause-Effect Relation Learning

An automated bootstrapping procedure which can learn and produce with minimal effort a cause-effect term repository, and to filter out the erroneously extracted information, incorporates graph-based methods.