Development of Multi-lingual Ontology Based Country Specific Search Engine for E-Commerce

toadspottedincurableInternet and Web Development

Dec 4, 2013 (3 years and 4 months ago)

60 views


-

1
-




CIS 700B Master’s Project Proposal


Development of Multi
-
lingual
Ontology

Based Country
Specific Search Engine for E
-
Commerce



Submitted to the

Department of Computer Science

College of Computing Sciences

New Jersey Institute of Technology



in Pa
rtial Fulfillment of

the Requirements for the Degree of


Master of Science


By


Ravish Khanna



APPROVALS


Proposal Number:


000
-
426
-
128


Agree to Advice
:


Professor James Geller



(Project Advisor)


Date Submitted:

_________________________


App
roved by:




_________________________


(MS CS Committee)





Date Approved:

__________________________









-

2
-



ABSTRACT


An application,

Multilingual
Ontology

Research Ecommerce
(MORE) system

has bee
n
developed

[5]

by
utilizing
w
eb translation services a
nd the Google search engine API.

The MORE system lets the user type any search terms

in his/her own language but
searches the Internet with another language and lets

the user add Ecommerce terms into
t
he search string so that the search results

are from
Ecommerce

Websites only.

This
system also caters to country specific search and summary translation.

Protégé

is an
extensible, platform
-
independent environment for creating and editing
ontologies

and
kno
wledge bases.
The project is

currently
running on
Ontology

editor
,

Protégé

version
3.1.1

but now we need

to

upgrade

it to the new beta version
Protégé

3.2
.
Currently
while
running the project using Protégé
3.2

it

is

having some server side
issues. So now th
e
ontology created in the version Protégé version 3.1.1 will have to

be

imaged/recreated
into the new

Protégé version 3.2.





-

3
-


Table of contents


1

Introduction ……………………………….
…………………………………..
4


2

Overview/pr
evi
ous Work……………………………………………………...4


3

Existing Architec
ture………………………………………………………….
5


3.1

Ontology

C
reation……………………………………………………….. 7


4

Proposed

Wo
rk……………………………………
…………………………..9


4.1

Work Description………………………………………………………
....9


4.2

Deliverables……………………………………………………………
...12


4.3

Task Break d
own Structure……………………………………………
...13

R
eferences…………………………
……
…………………………………...14








-

4
-


1

Introduction


Ontology

is a
data model

that represents a
domain

and is used to
reason

about the objects
in that domain and the relations between them.

Our project uses
Protégé

3.1.1 to build the
Ecommerce
Ontology
.
With the help of

Protégé

API the English
Ontology

can be
translated into the desired language
Ontology
.

The project is currentl
y facing server side
issues such as

not being able to communicate with the client using the new beta version

3.2
.

The

negative term


feature
wil
l also be

implemented
that finds

t
hose terms that are
more not relevant

for review of products and then exclude those terms.


1.1
Overview/Previous Work

Ecommerce has brought the world closer
together
and
made
more
products
accessible
today. Everything can

be bought/sold over the internet. It is a booming business and still
has
a lot of potential growth left in it. Sellers would like their products or services to
extend to any country’s buyers in the world, but they cannot create their Websites in
other lan
guages. A service with the capability of helping both sellers and buyers of
different languages to reach each other will solve this problem to a considerable extent.



The project allows us to

append country terms to the given user query so that the searc
h
is more
specific
. So

the user can get the results in the desired language. For example

if
the user wants to do a car search related to a specific country then he can select India in
the country option and can get the result in English, Ge
rman, Russian o
r Spanish

language
.
[5]


-

5
-


3
Existing
Architecture


Currently there is an interface

[5]

for searching with three categories of search Simple,
Advanced and Regular

is using version 3.1.1 of
Protégé
. Simple and Advanced searches
are used to search for Ecommerce

Websites while the Regular feature is used for a
normal web search.

The user enters a term in the search field and then selects
one among
the five

input languages

listed in the dropdown menu.

The user can
then
select
the output
language
and
the interface
displays the result in the output language
chosen.

The user can
also select the search to be

country specific by select the country name in the country
drop down menu

[8]
.

With the Simple search, basic Ecommerce terms such as Credit card
is appended to the

users input term for searching in Ecommerce Websites only.
Using

Advanced Ecommerce search the user can select terms that can be appended to the given
input term which will reduce the search space only to those relevant Ecommerce web
pages.
There is also
a help section (link) which gives a detailed explanation about the
working of the MORE search system.
Figure 2 shows the current user interface.


The tri
-
lingual
Ontology

of e
-
commerce terms has also been created as a first draft, and
users may append thos
e terms (in the correct language) to their own search terms. Also,
the
Ontology

based chart of country specific goods that are available has been made
.







-

6
-


Figure 1
: Screen shot of existing interface





3.1

Ontology

Creation

The
Ontology

covers generic

Ecommerce terms, independent of the product. Many terms
have been collected for the Ecommerce domain.

The
Ec
ommerce
Ontology

has been
created as shown
partially
in Fig
ure
2
.



-

7
-


Figure 2
: Rooted Directed Acyclic Graph for Partial Ecommerce
Ontology


























-

8
-


Figure 3:
Flowcha
r
t for Ecommerce Searching Interface


Explanation of the flowchart
:
-

Many steps are involved in our Multilingual
Ontology Research Ecommerce system, as sh
own in the flowchart in Figure 3
.

First, for t
he step 1 from the
flowchart, project

needed to create a multilingual
Ecommerce

Ontology.
Project is using
OWL, developed by t
he World Wide Web
Consortium

as the

Ecommerce Ontology creating

Language
. OWL is categorized into
three sub
-
languages: OWL
-
Lite, OWL
-
DL and OWL
-
Full.

It

also
required

some
specialized

software for

construct
ing

the Web Ontology in
the OWL language, which
included
:


Protégé [3], developed at Stanford University
,
a great graphical interface tool

for building Ontology classes. It has useful plug
-
ins, such

as the Protégé
-
OWL

Plug
-
in
, which can save the Ontology in OWL format.


RA
CER
, a stand alone semantic reasoner for RDF/OWL, is another great tool

which
helps to compute the inferred Ontology class hierarchy and to check

the consistency of the Ontology. W
hen working with Protégé, it’s easy to check

for errors while building the Ontology with the Protégé.

Project required the creation of

several
Ontology’s

for the

mult
ilingual Ontology system;
it is
a
chieve
d by

using a central
Ontology with translator to au
tomatically synchronize different

versions of the Ontology.
A
ll terms used in the

domain into other languages
ar
e manually translated and
save
d
in a
database.
A
program
is written which uses the

central version of the

Ontology and the
translated terms

to

c
reate

a unique Ontology for different languages.


-

9
-

In step 2, the user’s input term(s)

are translated

into the selected output language.

Project

use
s
existing Web translation services to process the input term(s). For this step, a

program

is written
to submi
t the term(s) to the Web translation service and process the
returned

results.
It uses JAVA as the

programming language.

In step 3,
a program

is written

to translate the Ontology from one

lang
uage to another
language. It uses

the Protégé JAVA API [3] to ac
cess the

Ontology

server.

In step 4, the user’s input term and selected Ecommerce terms
are submitted
to the search

engine.

It then uses

the Google API [6
] as the
tool to implement this process
together into
a user

interface. It uses

Java Server Pages

to
integra
te all mentioned programs in the

interface. We s
et up an Apache Tomcat server

as our Web server to run our JSP and

Java
Servlets.



Proposed

Work


My contribution to this project will address the following issues:



1) I will imple
ment the project using
Protégé

3.2 version on the server side.

2)

I will also implement the ‘negative term’ feature

that finds those terms that
are more not relevant for review of products and then exclude those terms

like
if the user is ordering a magazin
e reviewing cars then the user can order the
magazine not the car there.

3)
I will write code t
o

narrow the search.

For

example if the user types ‘car’ in
the search column then it will also display the set
of

terms related to it(car)
like ‘used cars’,’ fa
st cars’,

sports cars


etc.







-

10
-


4
.1

Work description

Research
:

I will

study in depth the working
of the existing architecture to understand

how the terms
are appended to the search string
which
will r
equire me to carry out the follo
wing:

1.

L
earn how simi
lar E
commerce W
ebsites
carries

out product specific searches and
display
s

their results.

For example amazon.com provides
us category

wise search
like grocery, beauty etc along with various quick links that user can
select.
EBay(another Ecommerce site) prov
ides an additional link called specialty sites
which allows the user to view most frequently visited categories.


2.

Unders
t
a
nd the workings of
Protege
3.2 and
Protégé

3.1.1.
U
se
Protégé
, developed
at Stanford University, as a graphical interface tool for
editi
ng

the
Ontology
. It
has useful plug
-
ins, such as the
Protégé
-
OWL plug
-
in, that can save the
Ontology

in OWL format. [3]


4.1
.1

Design
:

This phase will involve creating the blueprint of my work. The following are the tasks I
will accomplish during the Desig
n phase:


1.

Create modules of code that can

keep history of recent searches.

2.

Create modules of code
that can allow user to narrow down the search.

3.

Create the module for upgrading the project to version 3.2 of
Protégé
.


-

11
-

4.

Have a skeleton of my project in terms o
f what technologies to use, what code
will reside in which tier, etc.




4.1.2

Coding
/Implementation

1.

I will c
omplete all server
-
s
ide and client side programming, debug code and
running
the project

using Protégé version

3.2
.

2.

For
negative term feature
D
evelopment, I will use

java.


4.1.3
Integration/
Testing
:

Once the code has been generated, the software testing will begin. I will test my user
interface and Ontologies for various combinations of

inputs to check that the product
works as expected and to m
ake my code stable and

robust.

Once that is done, I will
integrate my code into the existing interface and continue testing.
I will deploy my

functionality on AFS servers at

NJIT. I will also do rigorous
integration
testing
i
n this

last
and final phase in
order to make sure that the Product is robust and user friendly.





-

12
-

4.2

Deliverables:

The following are the modules I will work on:

1.

Communicato
r module for

Ontology

Server:


Input: Product name.


Output: Product name with E
-
Commerce search t
erms appended to it

along with


s
et of
terms related

to the searched term
, along the history section

using
Protégé



Beta 3.2 version.

2.

Negative term module


Input: Product name.


Output:


List
those terms that a
re more typical for review of products and then


exclude those terms
.




4.3

Task Break d
own Structure


Week

Tasks to be accomplished

1

Research relevant technical papers and journals. Understand the central concept of
the project. Read

papers and technical tutorials about multilingual web browsers
and search engine functionality.

Understand

the workings of
Protégé

3.1.1.
[1]


2

Understand the workings for
Protégé

3.2
.
[2]



3
,4

5,6

Writing the code for upgrading it to the newer version

3.2
.

Working on negative term
implementation
.


-

13
-



7,8

Testing the code on the server end and debugging the errors
.


9

Working on the search optimizer module.


10

Debug code and test it.

11

Writing the code to implement the history section

12

Debug code.

Start testing in order to make code stable and robust.


13.

I will deploy my functionality on AFS servers in NJIT. I will also do rigorous
testing in this last and final phase in order to make sure that th
e Product is robust,
and user friendly.


14

Writ
e project report and do
integration

testing.



















-

14
-





References



[1
]
Protégé

Tool
,
http://
Protege
.stanford.edu/
,

Oct 19
, 2006
.


[2
]

Protégé

Tutorial
,

http://www.co
-
ode.org/resources/tutorials/
Protege
OWLTutorial.pdf
,

Oct 19
, 2006
.


[3
] Kuo
-
chuan Huang, Development of Web Ontologies as Data Exchange and

Decision Support Tools

to Facilitate Economic Cooperation between Ukraine and

USA
, MS Project Proposal

submitted to NJIT
, CS Department, 2006
.


[4
] James Geller’s homepage
,
http://web.njit.edu/~geller/what_is_an_Ontology.html,
Oct17
, 2006




[5
]
Arjun Rajaraman
,
Development of M
ulti
-
lingual
Ontology

Based Country Specif
ic
Search Engine for E
-
Commerce,
Oct15, 2006.


[6]
http://www.google.com/apis/reference.html, Oct16, 2006.