SOA IMPLEMENTATION IN PASSENGER TRANSPORT

MARKETING AIDED DECISION SYSTEM

Jianxiong Wang and Chunhuang Liu

Institute of Computing Technology, China Academy of Railway Science, Daliushu Rd, Haidian District, Beijing, China

Keywords: Business Intelligence, Cognos, Passenger Transport Marketing aided Decision, Cube, Service-Oriented

Architecture.

Abstract: Business Intelligence (BI) technology is researched in the first part of this paper, and application with

Service-Oriented Architecture (SOA) is built with IBM Cognos BI components to implement Railway

Passenger Transport Marketing aided Decision System, which providing intellectual query and analysis

solution for China Ministry of Railway and all railway bureaus in China. Finally, A development flow of

multi-dimension cube and its application of analyzing railway passenger data is presented to describe the

flexibility and high reliability.

1 INTRODUCTION

Railway Passenger Transport Marketing aided

Decision System (RPTMD) is used for collecting

passenger transporting information and processing

data, which can reflect railway transport capacity,

passenger send count and income, etc. There are

many kinds of complex business data，so Business

Intelligence technology is introduced into this

system to provide a united data view, and user

administration function based on role with light

directory access protocol (LDAP). The China

ministry of railway, bureau and station are

authorized to visit data and analyze railway transport

information to predict the passenger flow trends，

which can help the supervisor to make an optimized

decision. Here we make a research on Cognos BI

platform based on RPTMD.

2 BUSINESS INTELLIGENCE

Business Intelligence (BI) was first supposed by

Howard Dresner of Gartner Company in 1996. BI is

an enterprise solutions integrating data warehouse,

OLAP and data mining, which aims to collect data

from different business system. Extracting,

transforming and then loading into data warehouse,

which is an ETL process (Kimball and Ross, 2002).

Via query and analysis tools, data mining and OLAP

tools to analyze, which can product knowledge to

support managers to make decision. Nowadays, BI

products include data warehouse, data extracting

tools, OLAP tools and presentation tools and their

integration (Han and Kamber, 2001).

2.1 Cognos

Cognos is a BI platform built on WEB application,

which composes three components: BI Modelling,

BI Server for Reporting, BI Server for Analysis. It

also can be divided into four logical layers (Volitich,

2008).

 Data source layer: Data source layer includes

business database, data market, data

warehouse, cube and other non-relational data.

 Model layer: This layer provides two

functions, one is defining logical model using

metadata to make user apprehend easily, the

other is to build cube with pre-defined

dimension and set permissions to all of the

roles.

 Application server layer: Both BI Server for

Reporting and BI Server for Analysis publish

model and cube by Web services, query model

is linked to cube by “drill-through”.

 User layer: User layer is a virtual layer to

describe authenticated users who can query

and analyze.

380

Wang J. and Liu C..

SOA IMPLEMENTATION IN PASSENGER TRANSPORT MARKETING AIDED DECISION SYSTEM.

DOI: 10.5220/0003568903800383

In Proceedings of the 13th International Conference on Enterprise Information Systems (ICEIS-2011), pages 380-383

ISBN: 978-989-8425-53-9

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

2.2 Integrate BI with Decision Support

System

Recently years, BI technology is commonly used in

decision support system. Decision is a complex

process that persons participate in and take many

aspects into account, such as passengers’ demand,

flow prediction and marketing analysis, make a

certain aim with restrictions and search relative

cases in decision history. In this system a

generalized model is built by analyzing relative

cases. Integrating BI and decision support system

makes human and computer complementary. (For

instance, man-machine conversation is often used to

diagnose the boundary condition when solving an

ambiguous boundary problem.)

BI is also able to simulate reasoning mechanism

on the basis of demand ， through man-machine

conversation and related rules. Furthermore, BI

leads decision makers to select appropriate model.

Decision makers can get information by so-called

information push and pull technology, navigate the

system, query and analyze data and authorize the

roles.

3 APPLY BI IN RPTMD

3.1 Function

Passenger Transport Marketing aided Decision

service to Ministry of Railway, railway bureaus and

stations. The users are staff of passenger transport

company in China, who demand for:

 Easily access by Web.There are 18 bureaus

and thousands of stations in china. However,

Data are gathered in Ministry of Railway,

system must be Browser and Server

architectural to be convenient for maintain.

 Loosely coupling software architecture.

Changing on demand and service oriented

application are required in modern enterprise,

which appeals for software designer building

loosely coupling software and publishing

business with accessible services. It’s

preferred to develop software with SOA

application composition (Newcomer and

Lomow, 2004).

 Flexible analysis basis on schema and

multiple representations. Railway passenger

transport indexes are flexible. For example,

passenger send count can be viewed in many

dimensions, such as ticket type, seat type,

train type, depart time, etc. Otherwise, there’re

also many layers (time dimension includes

year, quarter and month). In a word, business

data is a multi-dimensional information and

suitable to described by super cube (or multi-

dimensional cube). In a word, every cube is a

schema reflecting business in reality.

3.2 Network Topology

A clustered network structure is used for getting

through fastigium. Business logic and Cognos portal

applications are running on Web application server,

several BI servers support BI analysis service with

cluster structure and scheduled by Web application

server. A content database is installed in content

manager server, which initialized by Cognos

configuration tools. LDAP service is running on

Directory server to supply directory managing and

certification based on role.

Figure 1: Network topology.

Under control of a directory manager server,

users are able to browse the permitted directory.

When users logged in the website, the request is

dispatched to directory server, and a session

including user information is stored in the

namespace.

After a signal of self-help query or browse a

report is sent，the request dispatches among these

BI servers, load balance is kept by gateway with

weighted distribution strategy. Finally, the request

will be sent to a relatively SOA composition

application development.

As Cognos supports Web Service interfaces and

open service contracts, a SOA System can be

composite with Cognos. As Javier et al. (2010)

content that Business functions are provided with

SOA models, and application developers can loosely

access all the available services, shorten the

SOA IMPLEMENTATION IN PASSENGER TRANSPORT MARKETING AIDED DECISION SYSTEM

381

developing cycle and reduce expenses and

emphasize more on business problems. As shown in

figure 2, Cognos kernel service and Passenger

Transport Marketing aided Decision service are

registered in service base with WDSL. When a user

gateway, the client gateway will search for costumed

server relative services in sub-system, such as

viewing reports based on report service, report data

server and batch report service, or designing query

based on report data service and managing users and

defining function options based on Content manager

service. However, nothing service detail is necessary

but service contract for developers.

Figure 2: SOA synthesize application development.

4 DEVELOPMENT

A developing case of Cube is introduced to explain

Cognos Multi-Dimensional Analysis development

technique, and the develop flow is shown in figure 3.

Figure 3: Multi-Dimensional Analysis development.

4.1 Data Source

A connect preference is set up to define destination

business database or plain file. There are two ways

to define, one is to select “running metadata wizard”

option in Namespace by Framework Manager, the

other is to open Cognos Connection interfaces, then

select “directory” option in tools menu, click icon

“new datasource” in “datasource” page to create a

new datasource. Oracle, MS SQLServer, Sybase and

ODBC are supported by Cognos.

4.2 Metadata

A Metadata model project can be created by using

Framework Manager with define its name and

saving path. Choose the metadata with type

“datasource”, and Import the database table for

working. After that there’re follow steps is necessary:

 Define table relationship. This step is to define

connection condition between imported data

tables. When a table is selected and after

choosing the option “creating relationship”,

another table should be chosen in Query

Subject to make relationship. One or more

table field will be involved in relationship.

There are four kinds of relationship including

1:1, 1: n, 0:1 and 0: n. The former two kinds

represent a common join relationship in

database and the later two kinds represent an

outer join relationship.

 Create view. When a Namespace is selected,

right button of mouse should be clicked and

menu option “Create” should be selected to

create a Query Subject. Firstly, the view name

and type should be defined, and then some

useful fields of database table should be

dragged in the query subject and renamed to a

virtual name.

 Assign externalize method. Sometimes some

view has to be published as external file. For

example, the Query Subject should be

selected, and modify its Externalize Method to

IQD, it’s a file format which means external

query define.

4.3 Publish Metadata

Firstly, packages category should be selected to

create a package. Secondly, package name is needed

and related views should be chosen. Finally,

“Publish Packages” menu option can be selected

while selecting the package. If an IQD file should be

publish as an alternate, an optional checkbox titled

“Generate the files for externalized query subject”

should be checked and it’s saving path is

indispensably.

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

382

4.4 Create Model

Now a database view is created and can be use for

report and query. However, to an extend OLAP

application, multi-dimension model is pressed for.

As shown in figure 4, firstly, a model name should

be defined by PowerPlay Transformer tool and

datasource type should be assigned by inserting data

source or creating an IQD file. Secondly, dimension

and index are defined by dragging database fields

into “Dimension map” work zone and dragging

database fields into “Measures” work zone. Finally,

a Cube definition should be insert into cube by

“insert cube” operation in PowerCubes interface.

Figure 4: Create Cube.

4.5 Cube

Cognos Cube is including metadata and values data.

Metadata is the attributes data of values data and

values data are visual data like dimension, layer and

class. Cube can be imported into Framework as data

source, then a cube file (*.mdc) will be published on

web, and the cube will be integrated into model by

PowerPlay Transformer tool as a model file (*.mdl).

As described in the cube design document, data for

query will be load into datasource from data

warehouse and dimension view is built which shape

the basis of cube. Several IQD datasources are

supported in the above process.

Here we make a experiment of Spring festival

Passenger transport in China, date and bureau

conclude the dimension, the former consists of

fastigium, festival and statistic date and the later

consists of bureau, branch bureau and station. Index

consists of passenger send num, bureau send num,

student send num, pre-sale num, seat utilization rate

and income, etc. All the indexes can be integrated

into a cube as shown in figure5. Users can connect

to the portal of RPTMD to create report, analysis

and query by themselves.

Figure 5: Publish Cube.

5 CONCLUSIONS

RPTMD has provided a large amount of data report

and analysis resource for China ministry of railway

and all railway bureaus in China. Especially in

Spring festival and so-called “Golden week”,

RPTMD has produced reports which offered

scientific evidence for passenger transport

department of railway aided on train plan, transport

capacity scheduling, assign ticket reasonably. SOA

architecture makes it possible to build loosely

coupling system and publish business with

accessible services. Furthermore, versatile

representation of query and report meets different

users’ demands and open analysis tool provides a

friendly develop condition. A farther research of BI

will be hold for deeply data mining and extracting

available information to guide passenger transport

plan, which will lead RPTMD embedded into

business flow.

REFERENCES

Ralph, Kimball., Margy, Ross., 2002. The data warehouse

toolkit. John Wiley & Sons, inc. New York. 2

edition.

J, Han., M, Kamber., 2001. Data Mining: Concepts and

Techniques. Morgan Kaufmann, San Francisco, CA.

Dan, Volitich., 2008. IBM Cognos 8 Business Intelligence:

The Official Guide. McGraw-Hill, Inc. New York.

Eric, Newcomer., Greg, Lomow., 2004.12. Understanding

SOA with Web Services. Addison-Wesley.

Francisco, Javier., Calle, Esperanza, Albacete., Enrique,

Sánchez., 2010. Cognos: A Natural Interaction

Knowledge Management Toolkit. Lecture Notes in

Computer Science, 2010, Volume 5723/2010, 303-304.

SOA IMPLEMENTATION IN PASSENGER TRANSPORT MARKETING AIDED DECISION SYSTEM

383