Files in this item



application/pdfKANG-DISSERTATION-2017.pdf (11MB)
(no description provided)PDF


Title:Extraction of formal manufacturing rules from unstructured English text
Author(s):Kang, Sungku
Director of Research:Dutta, Debasish
Doctoral Committee Chair(s):Dutta, Debasish
Doctoral Committee Member(s):Ferreira, Placid; Kim, Harrison Hyung Min; Patil, Lalit; Rangarajan, Arvind
Department / Program:Mechanical Sci & Engineering
Discipline:Mechanical Engineering
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Rule extraction
Semantic technology
Natural language processing (NLP)
Abstract:Semantics-based approaches—founded on the idea of explicitly encoding meaning separately from the data or the application code—are being applied to manufacturing, for example, to enable early manufacturability feedback. These approaches rely on formal, i.e., computer-interpretable, knowledge and rules along with the context or semantics. On the other hand, manufacturing knowledge has been maintained primarily in the form of unstructured English text. It is considered impractical for engineers to author accurate, formal, and structured manufacturing rules. Previous efforts on extracting semantics from unstructured text in manufacturing have focused exclusively on basic concept names and hierarchies. In this context, this dissertation focuses on the development of a semantics-based framework for acquiring more complex manufacturing knowledge, primarily rules, in a formal form, from unstructured English text such as those written in manufacturing handbooks. This dissertation includes the following specific research tasks. First, it studies the problem in manufacturing domain, proposes the formal rule extraction framework, and demonstrates its feasibility. Second, it extends the framework to complement standard Natural Language Processing (NLP) techniques with manufacturing domain knowledge to resolve ambiguities, called as domain-specific ambiguities, that are due to manufacturing-specific meanings implicit in the English text. Finally, this dissertation extends the framework to identify the cases that need input text validation, and provide the relevant feedback to the user to modify the input text for the extraction of correct rules. This research also demonstrates the extensibility of the framework. Specifically, the framework was initially developed using the subset of a manufacturing handbook only including milling, metal stamping, and die-casting sections, and then applied to the rest of the manufacturing processes including 30 sections in forming, machining, casting, molding, assembling, and finishing chapters in the book. Case studies are performed to demonstrate the feasibility of the framework on the dataset of 133 sentences. First, the feasibility of the rule extraction framework is shown by extracting correct rules from approx. 57% of the sentences. Second, the effectiveness of ambiguity resolution by complementing standard NLP techniques with manufacturing domain knowledge is demonstrated by an increasing the correct rules to 70%. Lastly, for the remaining 30% of the cases that need input text validation, relevant feedback is provided to the user to modify the input text for the extraction of the correct rules. It is expected that this research will facilitate the development of formal manufacturing knowledge including complex manufacturing rules. It will thus address an important barrier that has prevented a larger scale application and the adoption of semantic technologies in the field of manufacturing, especially for semantics-based manufacturability analysis.
Issue Date:2017-12-06
Rights Information:Copyright 2017 SungKu Kang
Date Available in IDEALS:2018-03-13
Date Deposited:2017-12

This item appears in the following Collection(s)

Item Statistics