On the Use of Anaphora Resolution for Workflow Extraction

On the Use of Anaphora Resolution for Workflow Extraction. Schumacher, P., Minor, M., & Schulte-Zurhausen, E. In Bouabana-Tebibel, T. & Rubin, S. H., editors, Integration of Reusable Systems, of Advances in Intelligent Systems and Computing, pages 151–170. Springer International Publishing, January, 2014.

Paper abstract bibtex

In this chapter we present three anaphora resolution approaches for workflow extraction. We introduce a lexical approach and two further approaches based on a set of association rules which are created during a statistical analysis of a corpus of workflows. We implement these approaches in our generic workflow extraction framework. The workflow extraction framework allows to derive a formal representation based on workflows from textual descriptions of instructions, for instance, of aircraft repair procedures from a maintenance manual. The framework applies a pipes-and-filters architecture and uses Natural Language Processing (NLP) tools to perform information extraction steps automatically. We evaluate the anaphora resolution approaches in the cooking domain. Two different evaluation functions are used for the evaluation which compare the extraction result with a golden standard. The syntactic function is strictly limited to syntactical comparison. The semantic evaluation function can use an ontology to infer a semantic distance for the evaluation. The evaluation shows that the most advanced anaphora resolution approach performs best. In addition a comparison of the semantic and syntactic evaluation functions shows that the semantic evaluation function is better suited for the evaluation of the anaphora resolution approaches.

@incollection{schumacher_use_2014,
	series = {Advances in {Intelligent} {Systems} and {Computing}},
	title = {On the {Use} of {Anaphora} {Resolution} for {Workflow} {Extraction}},
	copyright = {©2014 Springer International Publishing Switzerland},
	isbn = {978-3-319-04716-4 978-3-319-04717-1},
	url = {http://link.springer.com/chapter/10.1007/978-3-319-04717-1_7},
	abstract = {In this chapter we present three anaphora resolution approaches for workflow extraction. We introduce a lexical approach and two further approaches based on a set of association rules which are created during a statistical analysis of a corpus of workflows. We implement these approaches in our generic workflow extraction framework. The workflow extraction framework allows to derive a formal representation based on workflows from textual descriptions of instructions, for instance, of aircraft repair procedures from a maintenance manual. The framework applies a pipes-and-filters architecture and uses Natural Language Processing (NLP) tools to perform information extraction steps automatically. We evaluate the anaphora resolution approaches in the cooking domain. Two different evaluation functions are used for the evaluation which compare the extraction result with a golden standard. The syntactic function is strictly limited to syntactical comparison. The semantic evaluation function can use an ontology to infer a semantic distance for the evaluation. The evaluation shows that the most advanced anaphora resolution approach performs best. In addition a comparison of the semantic and syntactic evaluation functions shows that the semantic evaluation function is better suited for the evaluation of the anaphora resolution approaches.},
	language = {en},
	number = {263},
	urldate = {2014-02-27},
	booktitle = {Integration of {Reusable} {Systems}},
	publisher = {Springer International Publishing},
	author = {Schumacher, Pol and Minor, Mirjam and Schulte-Zurhausen, Erik},
	editor = {Bouabana-Tebibel, Thouraya and Rubin, Stuart H.},
	month = jan,
	year = {2014},
	keywords = {Artificial Intelligence (incl. Robotics), Computational Intelligence, information extraction, Anaphora resolution, Process oriented case-based reasoning, Workflow extraction},
	pages = {151--170},
	file = {Snapshot:C\:\\Users\\Minor\\Zotero\\storage\\EIGHFZ7D\\978-3-319-04717-1_7.html:text/html},
}

Downloads: 0