Cristina Puente

Extraction, analysis and representation of imperfect conditional and causal sentences by means of a semi-automatic process

Causality is not only a matter of causal statements, but also of conditional sentences. In conditional statements, causality generally emerges from the entailment relationship between the antecedent and the consequence. This entailment is frequently vague and uncertain in nature. In this article, we present a method of retrieving crisp and imperfect conditional and causal sentences identified by some linguistic patterns. These sentences are pre-processed to obtain both single cause-effect structures and causal chains. The result is displayed automatically in an imperfect causal graph by means of a Java application. The causal graph shows the strenght of the causal link labelling it with a fuzzy quantifier and the intensity of the cause or effect nodes with linguistics hedges. The knowledge base used to provide automatic information based on causal relations was some medical texts, suited for the described process.

Extraction of Conditional and Causal Sentences from Queries to Provide a Flexible Answer

This paper presents a flexible retrieval method for Q/A systems based on causal knowledge. Causality is not only a matter of causal statements, but also of conditional sentences. In conditional statements, causality generally emerges from the entailment relationship between the antecedent and the consequence. In this article, we present a method of retrieving conditional and causal sentences, in particular those identified by the presence of certain interrogative particles. These sentences are pre-processed to obtain both single cause-effect structures and causal chains. The knowledge base used to provide automatic answers based on causal relations are some medical texts, adapted to the described process. Causal paths permit qualifications in terms of weighting the intensity of the cause or the strength of links connecting causes to effects. A formalism that combines degrees of truth and McCulloch-Pitts cells enables us to weight the effect with a value and thereby obtain a flexible answer.

Retrieving Crisp and Imperfect Causal Sentences in Texts: From Single Causal Sentences to Mechanisms

Causality is a fundamental notion in every field of science. In empirical sciences, such as physics, causality is a typical way of generating knowledge and providing explanations. Usually, causation is a kind of relationship between two entities: cause and effect. The cause provokes an effect, and the effect is derived from the cause, so there is a relationship of strong dependence between cause and effect. Causality and conditionality are closely related. One of the main topics in the field of causality is to analyze the relationship between causality and conditionality, and to determine which causal relationships can be formulated as conditional links. In this work a method has been developed to extract causal and conditional sentences from texts belonging to different genres or disciplines, using them as a database to study imperfect causality and to explore the causal relationships of a given concept by means of a causal graph. The process is divided into three major parts. The first part creates a causal knowledge base by means of a detection and classification processes which are able to extract those sentences matching any of the causal patterns selected for this task. The second part selects those sentences related to an input concept and creates a brief summary of them, retrieving the concepts involved in the causal relationship such as the cause and effect nodes, its modifiers, linguistic edges and the type of causal relationship. The third part presents a graphical representation of the causal relationships through a causal graph, with nodes and relationships labelled with linguistic hedges that denote the intensity with which the causes or effects happen. This procedure should help to explore the role of causality in different areas such as medicine, biology, social sciences and engineering.

Mining answers for causal questions in a medical example

The aim of this paper is to approach causal questions in a medical domain. Causal questions par excellence are what, how and why-questions. The ‘pyramid of questions’ shows this. At the top, why-questions are the prototype of causal questions. Usually why-questions are related to scientific explanations. Although cover law explanation is characteristically of physical sciences, it is less common in biological or medical knowledge. In medicine, laws applied to all cases are rare. It seems that doctors express their knowledge using mechanisms instead of natural laws. In this paper we will approach causal questions with the aim of: (1) answering what-questions as identifying the cause of an effect; (2) answering how-questions as selecting an appropriate part of a mechanism that relates pairs of cause-effect (3) answering why-questions as identifying ultimate causes in the answers of how-questions. In this task, we hypothesize that why-questions are related to scientific explanations in a negative and a positive note: (i) as previously said, scientific explanations in biology are based on mechanisms instead of natural laws; (ii) scientific explanations are generally concerned with deepening, providing explanations as detailed as possible. Thus, we conjecture that answers to why-questions have to find the ultimate causes in a mechanism and link them to the prior cause summarizing the intermediate nodes in order to provide a comprehensible answer. The Mackie´s INUS causality offers a theoretical support for this solution.

Causality and imperfect causality from texts: A frame for causality in social sciences

The focus of this paper is the study of causality in both its crisp and approximate forms. Crisp causality is characterized by some properties and modalities and is related to semantic implications. This paper presents a program that extracts causal and conditional sentences with causal content from several texts. The examples extracted show that, even in scientific texts, causality can be imprecise or imperfect, as shown by the linguistic modifiers or the fuzzy quantifiers embedded in them. Quantum mechanics introduces imprecision in physics, but social sciences are the disciplines that show more circumstantial and imperfect links between cause and effect. In social sciences, there are two theoretical paradigms to understanding imperfect causality: (i) cause as an ‘ideal type’, from Weber, (ii) cause as a ‘family resemblance predicate’, from Anscombe, a follower of Wittgensteins philosophy. Our work provides two short exemplifications of these paradigms using the causal or conditional sentences retrieved from texts of different genres.

Mining Temporal Causal Relations in Medical Texts

Causal sentences are a main part of the medical explanations, providing the causes of diseases or showing the effects of medical treatments. In medicine, causal association is frequently related to time restrictions. So, some drugs must be taken before or after meals, being ‘after’ and ‘before’ temporary constraints. Thus, we conjecture that frequently medical papers include time causal sentences. Causality involves a transfer of qualities from the cause to the effect, denoted by a directed arrow. An arrow connecting the node cause with the node effect is a causal graph. Causal graphs are an imagery way to show the causal dependencies that a sentence shows using plain text. In this paper, we will provide several programs to extract time causal sentences from medical Internet resources and to convert the obtained sentences in their equivalent causal graphs, providing an enlightening image of the relations that a text describes, showing the cause-effect links and the temporary constraints affecting their interpretation.

Evaluation of causal sentences in automated summaries

This paper presents an experiment to show the importance of causal sentences in summaries. Presumably, causal sentences hold relevant information and thus summaries should contain them. We perform an experiment to refute or validate this hypothesis. We have selected 28 medical documents to extract and analyze causal and conditional sentences from medical texts. Once retrieved, classic metrics are used to determine the relevance of the causal content among all the sentences in the document and, so, to evaluate if they are important enough to make a better summary. Finally, a comparison table to explore the results is showed and some conclusions are outlined.

Summarizing information by means of causal sentences through causal graphs

Abstract The objective of this work is to propose a complete system able to extract causal sentences from a set of text documents, select the causal sentences contained, create a causal graph in base to a given concept using as source these causal sentences, and finally produce a text summary gathering all the information connected by means of this causal graph. This procedure has three main steps. The first one is focused in the extraction, filtering and selection of those causal sentences that could have relevant information for the system. The second one is focused on the composition of a suitable causal graph, removing redundant information and solving ambiguity problems. The third step is a procedure able to read the causal graph to compose a suitable answer to a proposed causal question by summarizing the information contained in it.

Summarizing Information by Means of Causal Sentences Through Causal Questions

The aim of this paper is to introduce a system able to configure an automatic answer from a proposed question and summarize information from a causal graph. This procedure has three main steps. The first one is focused in the extraction, filtering and selection of those causal sentences that could have relevant information for the system. The second one is focused in the composition of a suitable causal graph, removing redundant information and solving ambiguity problems. The third step is a procedure able to read the causal graph to compose a suitable answer to a proposed causal question by summarizing the information contained in it.

Compressing the representation of a causal graph

The aim of this paper is to introduce procedure capable of compressing the representation of causal graphs, removing the redundant information introduced. In previous works we have presented several algorithms to extract causal information from text documents by means of a semi-automatic process. As result we obtained a causal graph connecting concepts related to a given topic.