THE DESIGN AND IMPLEMENTATION OF THE INTEGRATED

DECISION SUPPORT SYSTEM ON LABOR MARKET

Dongjin Yu

1,2

Hangzhou Dianzi University, Hangzhou, 310018, China

Zhejiang Gongshang University, Hangzhou, 310018, China

Shixin Feng

Information Center of Labor and Social Security, Hangzhou, 310003, China

Guangming Wang

Zhejiang Gongshang University, Hangzhou, 310018, China

Keywords: Decision Support System, Labor Market, Service Oriented Architecture, OLAP.

Abstract: Nowadays labor resources in China face the fierce situation in their employment. The decision support

system on labor market helps the government to have a sound grip of the composition, migration and trend

of regional labor resources’ supply and requirement . This paper proposes an integrated architecture of the

decision support system on regional labor markets, which leverages the Service Oriented Architecture. The

real system implemented on this architecture, called as LMDSS, is also presented to show the features of

the multi-dimension analysis of job introducing.

1 INTRODUCTION

Since 1990s, significant changes have taken place in

the labor market of China. According to Chinese

statistical yearbooks, the number of registered

unemployed persons in urban areas rises steadily

from 7.70 million in 2002, to 8.00 million in 2003,

8.27 million in 2004, 8.39 million in 2005 and 8.47

million in 2006 at the registered unemployment rate

no less than 4.0% approximately each year (National

Bureau of Statistics of China. (ed), 2002 to 2007).

With the implementation of so-called Golden

Security Project launched since 2001, many cities in

China have established the Labor Market

Information Systems (LMIS), which provide the

capabilities of employment related information

services. LMIS focuses on business functions such as

unemployment registering, job introducing,

professional training and unemployment fund

allocation. Although LMIS usually stores a large

amount of historical business data, it does not

support decision making. However, the decision

makers in government agencies usually want to

discover the composition and its migration for both

the regional unemployment and employment.

Questions frequently arise as: which group is most

vulnerable to unemployment? Which kind of

irregular employment is most popular? Only with

these information at hand, could they adopt the

effective and positive employment policies.

Decision Support System (DSS) represents the

promise of enhancing managerial effectiveness by

improving the quality of decision making, in terms of

depth of analysis, breadth of synthesis, timeliness,

and systemic perspective (Gonzhlez, 1993). Since the

concept of DSS was first launched by Gorry and

Scott Morton (Goory and Morton, 1971), substantial

research has been conducted, especially in its

different application fields. Study on DSS with the

area of labor market commonly focuses on the

selection of candidates for recruitment. For instances,

Nina gives a Decision Support System based on

fuzzy logic model for Human Resources Appraisal

and Selection (HRAS) (Ruskova, 2002), and

Liang-Chih Huang constructs a novel model for

evaluation of managerial talent in DSS through fuzzy

535

Yu D., Feng S. and Wang G. (2008).

THE DESIGN AND IMPLEMENTATION OF THE INTEGRATED DECISION SUPPORT SYSTEM ON LABOR MARKET.

In Proceedings of the Tenth International Conference on Enterprise Information Systems - AIDSS, pages 535-538

DOI: 10.5220/0001705805350538

 SciTePress

neural network (Huang and Huang, 2004). The

Recommender System for job seekers and employers

is another application field given much attention.

Tobias Keim presents a unified multilayer

framework supporting the matching of individuals

for recruitment in (Keim, 2007). However, the

analysis of composition, migration and trends for

regional labor market is less concerned.

This paper introduces the framework and its

implementation of decision support system for labor

market, called LMDSS. Based on Services Oriented

Architecture, LMDSS accomplishes the interaction

with other related information systems. It is

constructed on the data warehouse, and leverages the

technology of Online Analytic Processing (OLAP)

and data mining. OLAP helps ad-hoc query via data

slice, dice, drill up, drill down and swap, while data

mining helps discover the implicit and valuable

information such as patterns, trends and

characteristics.

The paper proceeds as follows. Section 2 shows

the overall architecture of LMDSS, as well as its

features. Section 3 discusses its characteristic on

services orientation. The implementation of LMDSS

is presented in Section 4, especially on it’s

multi-dimension analysis of job introducing. Finally,

in Section 5, some concluding remarks and

interesting open issues are sketched.

2 THE ARCHITECTURE OF

LMDSS

The overall architecture of LMDSS mainly consists

of the framework, meta-data management module,

data collection module, data online analysis module

and data mining module as showing in Fig 1.

Figure 1: The Architecture of LMDSS.

The framework provides the running-time

environment, supporting portal, workflow

management, security control and knowledge

sharing, while the meta-data management module

maintains the related models, algorithms and rules.

The core of system implements data collection,

analysis and mining. It integrates the records via the

ETL (Extract-Transform-Load) engine from the

related operational systems, the survey results and

the economic leading index published in yearbooks,

and then constructs the data warehouse. Currently,

LMDSS stores the following themes of data:

registering of unemployment, unemployment

funding, enterprise employment, individual

job-hunting records, and the urban and rural labor

resources.

The data analysis module focuses on the

multi-dimensional drilling of certain regional labor

market measures, like the number of posts wanted,

the amount of unemployment fund paid, etc. The data

mining module offers the medium-short-term

forecast of the overall labor resources’

supply-demand, the medium-long-term forecast of

it’s components, and also the profit-loss analysis of

unemployment fund. In addition, certain key features

could be discovered through the process of data

classification, clustering and association.

The resource layer in Fig. 1 is composed of the

data warehouse providing the standard JDBC

interfaces and other data sources located by JNDI,

including the model base for schemes of themes, the

algorithm base for data mining and the rule base for

constraint. For system administrating, the interface

layer provides the desktop application based on

Eclipse. Meanwhile, the browser based pages are

also given for data presentation.

3 DESIGN OF SERVICES

ORIENTED INTEGRATION

LMDSS is not independent and needs to interact with

other systems like the labor market operational

system, the social security operational system and so

on. For instances, LMDSS retrieves post-wanted

records from the labor market operational system and

unemployment payment records from the social

security operational system. Besides, operational

systems reuse the algorithms provided by LMDSS to

implement their own business analysis.

Traditional DSSs are somewhat difficult to

integrate new functions or connect to other systems

ICEIS 2008 - International Conference on Enterprise Information Systems

536

due to the deficiency in openness (Chi et al., 2005).

The specific adapters are usually developed to

connect all these heterogeneous systems. Instead,

LMDSS leverages the Service Oriented Architecture

to achieve the application coupling. Service Oriented

Architecture (SOA) is a paradigm for organizing and

utilizing distributed capabilities that may be under

the control of different ownership domains. It

provides a uniform means to offer, discover, interact

with and use capabilities to produce desired effects

consistent with measurable preconditions and

expectations.

Using the standard SOA approach, LMDSS

provides open access to its OLAP engine with

multi-dimensional data through web services. In

LMDSS, the execution of MDX (Multidimensional

Expressions) statements is provided as web services

via XMLA (XML for Analysis) protocol, which

specifies a set of XML message interfaces that use

the industry standard Simple Object Access Protocol

(SOAP) to define data access interaction between a

client application and an analytical data provider

working over the Internet. Once receiving requests of

multi-dimensional queries, the OLAP engine in

LMDSS parses the MDX statement into SQL

statement and then access the data warehouse.

Finally, the OLAP engine reformats the result with

SOAP and returns the messages back to the client.

In addition, the core data mining algorithms are

encapsulated as SCA (Service Component

Architecture) components. Therefore, with the help

of WSDL defined interfaces, LMDSS provides the

capabilities of on-demand invoking of it’s inner

mining algorithms for outside systems.

4 THE IMPLEMENTATION OF

MULTI-DIMENSION ANALYSIS

The data online analysis module in LMDSS is

implemented through OLAP technology. The

multi-dimension data set is organized as cubes,

measures and dimensions, while the dimensions are

further expressed by hierarchies and levels. Once

fetched from the data warehouse, the result set is

temporarily kept in the server’s memory. If the new

query could be calculated directly from the data in

memory, no more reading of warehouse is needed.

Moreover, the frequently used joins of fact tables and

dimension tables are materialized in advance. If

possible, the SQL statements are rewritten to access

the materialized tables instead of the underlying fact

tables and dimension tables. Experiments have

shown the performance of data retrieving could be

significantly improved.

Fig. 2 gives the screen shot for the analysis of job

introducing in LMDSS, which presents 3 dimensions

as unit types, data issued and education levels, and

also 2 measures as the number of the wanted and the

number of the successfully introduced.

Figure 2: The screen shot for the multi-dimension analysis

of job introducing.

5 CONCLUSIONS

The decision support system for labor market need to

be architecturally flexible and extensible. LMDSS

leverages the SOA and aids the decision-makers by

presenting multi-dimension analysis. The data source

are mainly the records retrieved from the labor

market information systems, plus the statistical data

from surveys and yearbooks.

Because of the huge amount of related

operational records and their distribution, the model

of distributed data analysis need to be further studied.

For the next step, LMDSS is expected to be migrated

to the grid platform, and Globus is thought to be the

ideal candidate.

ACKNOWLEDGEMENTS

The work is supported by Zhejiang Provincial Key

Project with grant No. 2007C21044, of China.

THE DESIGN AND IMPLEMENTATION OF THE INTEGRATED DECISION SUPPORT SYSTEM ON LABOR

MARKET

537

REFERENCES

National Bureau of Statistics of China. (ed), 2002, 2003,

2004, 2005, 2006, 2007. China Statistical Yearbook:

2002, 2003, 2004, 2005, 2006, 2007. China Statistical

Press. Beijing.

Gasth T. Gonzhlez, 1993. Information architectures and the

successful implementation of decision support systems.

In Proceedings of IEEE Systems Man and Cybernetics

Conference.

G.A. Goory, and M.S. Scott Morton, 1971. A framework

for management information systems. In Sloon

Management Review, vol. 13, no. 1.

Nina At. Ruskova, 2002. Decision Support System for

Human Resources Appraisal and Selection. In First

International IEEE Symposium “Intelligent Systems”.

Liang-Chih Huang, Kuo-Shu Huang, 2004. Applying

Fuzzy Neural Network in Human Resource Selection

System. In NAFIPS '04.

Tobias Keim, 2007. Extending the Applicability of

Recommender Systems: A Multilayer Framework for

Matching Human Resources. In HICSS'07,

Proceedings of the 40th Annual Hawaii International

Conference on System Sciences.

Jia-Yu Chi, Ling Sun, Xue-Guang Chen, etc, 2005.

Architecture and Design of Distributed Decision

Support System Based on Agent Grid, In Proceedings

of the Fourth International Conference on Machine

Learning and Cybernetics.

ICEIS 2008 - International Conference on Enterprise Information Systems

538