Challenges for Information Access in Multi-Disciplinary Product Design and Engineering Settings
-
Upload
dirk-ahlers -
Category
Technology
-
view
267 -
download
0
Transcript of Challenges for Information Access in Multi-Disciplinary Product Design and Engineering Settings
Challenges for Information Access in Multi-Disciplinary Product Design and Engineering Settings
Dirk Ahlers∗, Mahsa Mehrpoor#, Kjetil Kristensen#, John Krogstie∗
∗ Department of Computer and Information Science# Department of Engineering Design and Materials NTNU – Norwegian University of Science and Technology Trondheim, Norway
ICDIM 2015, Jeju, South Korea
2
Background Topics
• KBE (Knowledge-based Engineering)• IR (Information Retrieval)• Information Seeking• Recommender Systems (RecSys)
• Knowledge Management• Engineering and Design Toolchains• Collaborative Work
3
Application Domain
• Information Access in large engineering and design projects
• Large and multi-disciplinary project teams, long project durations
• Vast mix of tools and heterogeneous data sources
• Professional Search Environment
[Pictures source Wikipedia:User Tannkrem https://en.wikipedia.org/wiki/File:Aker_Spitsbergen.JPG]
4
Research Issues
• KBE as rule-based engineering– Capture and reuse of design knowledge
• Information-seeking workflow in knowledge-intense engineering
• Heterogeneous non-textual data sources• Bridging the gap between IR/RecSys and KBE
5
Challenges
• Limited information space• Heterogeneous sources• Knowledge separation• Inconsistent (use of) metadata• Insufficient connections between documents• Sparsity of document space and user interactions• Searcher are domain experts, search tasks are highly
complex• Much interaction outside of the system
6
Conceptual View
[D. Ahlers and M. Mehrpoor, Semantic Social Recommendations in Knowledge-Based Engineering, SP 2014]
7
Information Needs and Search Tasks
• Information Seeking behaviour expanded with KBE tasks• Expand from a system-centric view and take user
context and tasks into account• Widely varying tasks and contexts• IR/KM/RecSys is only an inner task in a nested context
8
Information Seeking
• s
9
Contextual Features
• Important aspects in user information seeking
• Can be used to enhance search and recommendation
• Domain-specific• Ontology-supported query
matching
10
System Framework
• Implementation of both RecSys and IR aspects
• User profiles• Domain Ontology• Relevance feedback• Adapted standard tools for
(text) indexing• Information extraction and
estimation for non-textual documents
11
Conclusion
• Challenging research area:– Manufacturing Engineering Information Access
• Use of context of the entire document collection• Ontology-supported matching of documents and
tasks/processes• Support for recommendation, search, navigation• Future work
– Refined system implementation– Deeper document analysis, Use of detailed ontology– User evaluations
13
Further reading
[M. Mehrpoor, J. A. Gulla, D. Ahlers, K. Kristensen, S. Ghodrat, and O. I. Sivertsen, Using Process Ontologies to Contextualize Recommender Systems in Engineering Projects for Knowledge Access Improvement, in ECKM2015.][D. Ahlers and M. Mehrpoor, Everything is Filed under ‘File’ – Conceptual Challenges in Applying Semantic Search to Network Shares for Collaborative Work, in Hypertext 2015.][M. Mehrpoor, A. Gjærde, and O. I. Sivertsen, Intelligent services: A semantic recommender system for knowledge representation in industry, in ICE 2014.][D. Ahlers and M. Mehrpoor, Semantic Social Recommenda- tions in Knowledge-Based Engineering, in SP 2014: Workshop on Social Personalisation at Hypertext 2014.]