Extracting and enriching workflows from text

Abstract

This paper is on a workflow extraction framework which 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 NLP (Natural Language Processing) tools to perform information extraction steps automatically. In detail, the paper presents on the step of anaphora resolution to enrich the workflow extracted so far. 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. The results of the approaches are compared to each other. For the evaluation, we use 37 workflows which have been created by a human expert.

Comparing and Aligning Process Representations

“…Therefore, we here take a conservative approach when imposing ordering restrictions. In particular, we recognize that textual descriptions generally describe actions in a chronological order [248]. For this reason, our rationale is to only allow for the generation of alignments that respect a chronological description.…”

Supporting Legal Requirements in the Design of Public Processes

“…These contain the set of components, management rules and instructions regarding a public administrative procedure [5][6] [20]. They also regulate strictly how to create a certain output [4].…”

“…The emerged processes are then visualized using a subset of UML diagrams. In the same sense, there exists the works of [5] which propose framework for extraction and feeding processes from legal texts. The framework applies pipes and filters architecture and uses NLP tools to perform information extraction steps.…”

Ontology-based workflow extraction from texts using word sense disambiguation

“…This can be supported by reasoning methods like Process Oriented Case-Based Reasoning (POCBR) [2] which is a newly merged branch from the Case Based Reasoning (CBR) field. Because case engineering is a complicated and a costly process, automatic case-based acquisition from texts has been a major focus research in the last decade [3,4,5].…”

“…However, their model have only successfully extracted ∼ 42% of these documents. In [4], the authors propose to use a full fledged NLP solution to extract and enrich workflows from texts. Their frame-based Information Extraction (IE) approach to resolve anaphoras in texts uses frame patterns to trigger phrases with slots filled with content extracted from texts.…”