trieval. Section 3 shows the overview of our approach. Section
4 designs a DR ontology based on the extended IBIS-based DR
representation. Section 5 proposes our approach of performing
ontology-aided indexing. Section 6 provides three user-friendly
modes for querying and gives the corresponding query process-
ing. After that, implementation of a prototype system is given in
Section 7. Finally, Section 8 summarizes our proposed work and
discusses future works.
2. RELATED WORK
2.1 DR Representation
A good representation schema is vital to enabling effec-
tive design and reuse [1]. Research on DR representation has
been reported since the 1970s. Most of the DR representation
approaches are argumentation-based approaches, and the typi-
cal model is issue-based information system (IBIS) [4], which
uses issues, positions, arguments and relationships between them
to represent DR. Several software tools which allow engineer-
ing designers to record DR have been implemented based on
IBIS. For example, Conklin et al. [5] developed graphical IBIS
(gIBIS), and Bracewell et al. [6–8] implemented Design Ratio-
nale editor (DRed). In addition, McCall [9] proposed the Proce-
dural Hierarchy of Issues (PHI) model, which broadens the scope
of the concept of “issue” in IBIS. Another argumentation-based
model is question, option and criteria (QOC) [10], which is a
kind of semi-formal notation of design space analysis. Liu et
al. [11, 12] proposed an issue, solution and artifact layer (ISAL)
model for DR representation and rationale information discovery
from design archival documents.
2.2 DR Retrieval
There are several works dedicated to DR retrieval in re-
cent years. In general, DR retrieval works can be classified into
two main categories: text-based retrieval and ontology-based re-
trieval.
Most of current DR retrieval methods are text-based. Liang
et al. [13] proposed a DR search and retrieval system which fo-
cuses on interactive user interface design. There are three ba-
sic functions: the view functions enable engineering designers
to intuitively navigate DR repository; the search functions sup-
port designers to retrieve relevant DR from multiple aspects; and
the analysis functions suggest some useful DR insights. Kim
et al. presented two methods for the retrieval of DR captured
using DRed. The first approach uses NLP techniques to anno-
tate rationale records with 9 selected semantic relations [2]. The
second approach recommends relevant pieces of DR by analyz-
ing the design task models of design reuse [14]. Also for DRed
files, Wang et al. developed a keyword-based retrieval tool at
first [15], and then proposed a new DR retrieval system making
use of the implicit structures in DRed graphs [16]. The general
problem about the text-based retrieval is that various DR records
have semantics such as types, relationships and structures, etc.,
however, text-based retrieval is very hard to take full advantage
of the semantics, especially the implicit ones.
In comparison with text-based retrieval, ontology-based re-
trieval makes better use of the semantics embedded in DR
records by utilizing ontology. Lim et al. [17, 18] proposed an
information search and retrieval framework based on the seman-
tically annotated multi-facet product family ontology, and ex-
emplified how they can derive new product variants based on
the designer’s query of requirements via the faceted search and
retrieval of product family information. L
´
opez et al. [19] pre-
sented NDR ontology to describe non-functional requirements
(NFR) and DR knowledge, and multi-facet search was imple-
mented through executing SPARQL
1
queries over the semantic
catalogues of NFR. However, these approaches are far from be-
ing practical since they require a relatively complex query lan-
guage. A scalable alternative to query construction from simple
queries is semantic indexing, in which semantic data in RDF
2
knowledge bases is indexed in a structured way and directly
available to be searched with simple queries such as keyword-
based query. In information retrieval domain, Kara et al. [3]
presented an ontology-based information extraction and retrieval
system in the soccer domain, in their work, a keyword-based re-
trieval approach using semantic indexing was proposed.
In summary, the research on DR retrieval is still in its in-
fancy. Our work presents an ontology-based DR retrieval ap-
proach, which takes full advantage of the semantics in DR
database through ontology-aided indexing and supports flexible
query modes.
3. OVERVIEW OF APPROACH
Figure 1 shows the overview of our ontology-based DR re-
trieval approach. It could be seen that our approach contains
three main parts, i.e. the DR database, the online processing and
the offline processing. Here we give a brief description of each
part respectively.
The DR database stores all the necessary data involved in
both online processing and offline processing, including the DR
records captured by designers, the DR ontology designed ac-
cording to the extended IBIS-based DR representation as well
as the semantic rules defined. Moreover, the index generated by
ontology-aided indexing is also stored in this database. In this
work, each DR record is stored as a file whose content is rep-
resented in a structured way, similar to that in the proposed DR
representation in section 4.1.
The online processing starts when a user inputs a query and
ends with exporting pieces of DR records fulfilling the query.
1
http://www.w3.org/TR/rdf-sparql-query/.
2
http://www.w3.org/TR/PR-rdf-syntax/.
2 Copyright
c
2013 by ASME
Downloaded From: http://asmedigitalcollection.asme.org/ on 01/13/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use