Project: Information Extraction from Work Orders for Knowledge Graphs
Description
Company: Marel
Location: Boxmeer
Background
Work order descriptions provided by engineers are of utmost importance in various industries as they serve as essential documentation for maintenance, repairs, and other technical tasks. These descriptions provide detailed information about the required work, including the scope, specifications, and any necessary materials or tools. However, integrating this vast amount of information and making it easily searchable can pose a significant challenge. The descriptions may vary in format, language, and level of detail, making it difficult to establish a standardized system for organizing and retrieving them. Additionally, engineers often use technical terminology and jargon that may not be readily understandable to others, further complicating the process. Still, work orders contain vital information that (when aggregated) can give many insights into procedures/defects/processes. One potential solution to address this challenge is to integrate this information into a Knowledge Graph (KG), structured databases that capture factual information.
Problem
Obtaining relevant information from work order texts poses significant challenges. This often starts from the get go where defining what information (entities or relations) are relevant is generally undefined from prior work due to the uniqueness of use-cases. For the actual extraction off-the-shelf libraries or models are often focused on more general domains (Wikipedia, social media) and thus rarely work for company-specific industrial applications. Standardizing and normalizing the extracted information is another challenge, as work orders often exhibit diverse naming conventions and terminologies. Meeting these demands requires a combination of domain expertise, advanced technologies, and a robust data management framework to successfully obtain a reliable and meaningful Knowledge Graph from work order texts.
Still, having this information extracted and placed in Knowledge Graph opens up a wide range of valuable applications. It can provide insights into historical maintenance activities, facilitate predictive maintenance, identify early signs of equipment failure or support decisionmaking by providing a comprehensive view of equipment, assets, and their relationships.
Research direction
Existing work on this shows potential for the domain of oil and gas industry [1] which tries to tackle both the Information Extraction side as well as visual results for applications. This raises questions how applicable this work is for other domains, such as the food processing industry. To answer this, one will need to find out whether the same principles hold or how they differ, but how one can still arrive at a functional KG.
Goals & Deliverables
- Identification of relevant entities/relations and study similarity between existing works
- Apply, build and verify IE methods for extracting information from Work orders in food processing equipment
- Build and demonstrate the resulting knowledge graph from the extracted data
Student requirements
- Having passed (or intending to take) Text Mining (2AMM30) is a pre
- Pragmatic but research oriented, with the intention to publish the results/findings in a journal paper
- Selecting and pre-processing the data for this project is available as an internship (see here: [Link]). Students who complete the internship will be prioritized for the project.
About Marel
Marel is a global company specializing in advanced food processing equipment and systems for the poultry, meat, and fish industries. They offer a comprehensive range of solutions for various stages of food production. Marel's technologies and automation solutions are designed to enhance efficiency, improve product quality, and ensure food safety. They also provide software solutions and data analytics services to help food processors monitor and optimize their operations. With a strong emphasis on innovation and sustainability, Marel collaborates with industry partners to drive advancements in food processing methods.
[1] Stewart M, Hodkiewicz M, Liu W, French T. MWO2KG and Echidna: Constructing and exploring knowledge graphs from maintenance data. Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability. 2022;0(0). doi:10.1177/1748006X221131128
Details
- Supervisor
-
Mykola Pechenizkiy
- Secondary supervisor
-
Zeno van Cauter
- External location
- Marel, Boxmeer
- Interested?
- Get in contact