OGSABased Product Innovation Application Platform

1
OGSA-Based Product Innovation Application Platform

• Guoxian Lou
• Grid Computing Lab of Zhejiang University

2
1.Introduction
In the Grid Environment, the product
innovation application platform will have two
important feature OPEN and SEMANTICS. We
use OGSA to construct the open grid environment
and use RDF to represent the entities of paten
and product in order to make the elements of
product innovation application platform contain
semantics.
3
2.Architecture of OGSA-Based Product Innovation
Product Innovation Application Platform
Semantic Browser
ApplicationFramework
Semantic search
Patent Grid-Service
Questionnaire Grid- Service
Grid-Service Layer
Questionnaire Knowledge-Base Grid
Patent Knowledge-Base Grid
Virtual DB Service
Grid Layer
DB-Grid
patent
questions
product
4
3.Using OGSA to construct the Grid Environment(1)
In the Internet environment the resources
used in the product innovation application
platform is widely distributed. For example the
patents base is one of the key resource of
product innovation, and there exist tens even
hundreds of patents databases around the world.
They may reside in different computer system and
may be accessed from different product innovation
application platform. What is needed is a common
way for the application layer to access data from
any data source like a device or a database. To
achieve it, we choose the Grid computing model,
and take OGSA as the common accessing way.
5
3.Using OGSA to construct the Grid Environment(2)
Building on both Grid and Web services
technologies, the Open Grid Services Architecture
(OGSA) defines mechanisms for creating, managing,
and exchanging information among entities called
Grid Services. As in the architecture of
the platform, there are two grid service
components. Each of them can provide several
services. For example each of them provides a
service that can get the ontology of patent or
requirement domain respectively.
6
4. RDF-Based Semantic Representation
In order to make the effective search on the
patents we should take some measures to make the
traditional search containing semantics. It must
be a more effective method than the classical
one. To construct an effective search containing
semantics (also named semantic search), we
believe, the semantic reconstruction from the
bottom data is essential. We use the Resource
Description Framework (RDF) and RDF Schema1,2
for representing information including the
semantic information of the patent and product.
7
4.1 Knowledge Representation of Patent(1)
To describe the semantics of a product
actually, data should be understandable for
machines. Therefore, semantics are required to
provide the information needed to understand and
analyze data. RDF gives a framework and schemas
for the implementation of data representation
including patents.
8
4.1 Knowledge Representation of Patent(2)
B
Patent Ontology
Performing Operation Transporting
C
A
Human Necessities
Chemistry Metallurgy
Patent
D
TextilePaper
Fixed Constructions
E
IPC
9
4.1 Knowledge Representation of Patent(3)
FURNITUREDOMESTIC ARTICLES OR APPLIANCES
SUCTION CLEANERS IN GENERAL
A47
AUTO WALKING
DOMESTIC WASHING OR CLEANING SUCTION CLEANERS
IN GENERAL
WASHING CARPET
L
Suction cleaners
5/00 7/00 9/00
10
4.1 Knowledge Representation of Patent(4)
We consider the leaf nodes in the IPC as a
domain, and define a set of function words in
every domain. Organize the function words under
the IPC, we get the ontology of the function as
the figure below showing.
Semantic relationship between function keys ex.
rdfseealso
function
domain
5.3
11
4.1 Knowledge Representation of Patent(5)
Attributes of patent

• Application Number
• Application Date
• Publication Number
• Publication Date
• International Classification
• Abstract
• Inventor Name
• Applicant(s) Name
• Address

Add a function attribute and use RDF to represent
the patent with full semantics
RDF/XML
5.1
12
4.2 Knowledge Representation of Product(1)
5.1
13
4.2 Knowledge Representation of Product(2)
A product can be defined as anything that can
satisfy some requirements. Based on the
requirements, a product also can be divided into
several components, each of which satisfies one
or more requirements. And a component may make
use of one or more patents. In such semantic
aspect, a product can be considered to have two
properties that are components and requirements.
14
5. Semantic Search Methods
Based on the semantic representation of the
patent and product, we present three semantic
search methods on the patents for product
innovation.
15
5.1 Semantic Collision
As in Figure 2 and Figure 3 saying, a patent
can perform several functions and a product
should satisfy some requirement that is composed
of several functions. When we began to develop a
new product, we had made a domain location for
the product. In the specified domain we search
the corresponding patents that can perform the
functions that make up of the requirement. These
patents are used by the components of which the
product consists. Get all these patents, they are
the result of the semantic collision method.
16
5.2 Semantic Cluster(1)
In Figure 4 we can find that in a specified
domain there exists relationship among different
functions. Using such relationship we create the
semantic cluster method.
17
5.2 Semantic Cluster(2)
The relationship between a function and its
domain is captured by the predicate domain_of.
Between the function1 and domain1,
domain_of(function1, domain1) means function1 is
belong to domain1. In the early example, we
suppose function f1, f2 and f3 are belong to
domain d1, so we can describe it as
following domain_of(f1, d1). domain_of(f2,
d1). domain_of(f3, d1).
18
5.2 Semantic Cluster(3)
The relationship between a patent and its
performing function is captured by the predicate
perform.8 Between the patent1 and function1,
perform(patent1, function1) means patent1 can
perform function1. Suppose patent p1, p2, p3
perform f1, f2, f3 respectively, we can describe
it as following perform(p1, f1). perform(p1,
f1). perform(p1, f1).
19
5.2 Semantic Cluster(4)
In addition we should use a predicate to describe
the relationship between functions in the same
domain. We use the predicate relation_in_domain
to capture it. Between the function1 and
function2, relation_in_domain (function1,
function) means there exists a relationship
between function1 and function2. As the early
example we describe the relationship between f1
and f1 as following relation_in_domain(f1,f2)
20
5.2 Semantic Cluster(5)
Since there exists relationship among f1,
f2 and f3, we think that patents performing f2
and f3 can also be used by the new product. And
we name the method to find the patents by f2 and
f3 as semantic cluster.
21
5.3 Semantic Association(1)
In Figure 4 we can also find that there exists
relationship among functions in different
domains. Semantic association is created based on
such relationship. For example, function A is one
of the functions that composed of the products
requirement, and there exist relationship between
function A and function B. But A and B is in
different domain. Function B and C also have
relationship, and C is belong a domain that is
different from A and B. Although B and C is
belong to a different domain from A, we think
that the patents performing B or C may be useful
for A.
22
5.3 Semantic Association(2)
23
Thank you!
24
IPC
A?-??????(?????) B?-????? C?-????? D?-?????
E?-?????(?????) F?-???? G?-?? H?-??
4.1