Challenges and some potential strategies for relating engineering issues with their causes in text

Abstract

This work is targeted towards achieving knowledge reuse by acquiring diagnostic knowledge about assembly from documents, and feeding it back for subsequent use in the design/assembly planning stage. Towards this objective, this paper looks at some challenges that have been identified in finding in text causal relations between issues and their causes. The analysis is carried out from the perspective of using the logical form of the text, which provides a computer-based representation of natural language text. By analyzing examples, a set of generalizations are made about the structure of possible classes into which text expressing both issues and causes may be classified. Then we show a simple example for which identifying the cause and effect is possible. However, we also show that it gets more difficult for more complex cases. Hence a method based on pattern matching in text is proposed and explained with further examples. Future directions are then discussed.