22-05-2014, 10:48 AM
Social Security and Social Welfare Data Mining: An Overview
Social Security and Social .pdf (Size: 743.37 KB / Downloads: 44)
Abstract
The importance of social security and social welfare
business has been increasingly recognized in more and more coun-
tries. It impinges on a large proportion of the population and affects
government service policies and people’s life quality. Typical wel-
fare countries, such as Australia and Canada, have accumulated a
huge amount of social security and social welfare data. Emerging
business issues such as fraudulent outlays, and customer service
and performance improvements challenge existing policies, as well
as techniques and systems including data matching and business
intelligence reporting systems. The need for a deep understand-
ing of customers and customer–government interactions through
advanced data analytics has been increasingly recognized by the
community at large. So far, however, no substantial work on the
mining of social security and social welfare data has been reported.
For the first time in data mining and machine learning, and to the
best of our knowledge, this paper draws a comprehensive overall
picture and summarizes the corresponding techniques and illustra-
tions to analyze social security/welfare data, namely, social security
data mining (SSDM), based on a thorough review of a large number
of related references from the past half century. In particular, we
introduce an SSDM framework, including business and research
issues, social security/welfare services and data, as well as chal-
lenges, goals, and tasks in mining social security/welfare data. A
summary of SSDM case studies is also presented with substan-
tial citations that direct readers to more specific techniques and
practices about SSDM.
INTRODUCTION
MACHINE learning and data mining are increasingly used
in business applications [21], and in particular, in public
sectors [94]. A distinct public-sector area is social security and
social welfare [121] which suffers critical business problems,
such as the loss of billions of dollars in annual service delivery
because of fraud and incorrect payments [121], [134]. People
working in different communities are increasingly interested
in “what do social security data show” [93] and recognize the
value of data-driven analysis and decisions to enhance public
service objectives, payment accuracy, and compliance [101],
Technical Perspective
From the technical perspective, the main issues that have
been addressed in the literature focus on several areas, includ-
ing problem analysis, process and policy modeling, business-
oriented analysis, correlation analysis, infrastructure support,
and data-driven analysis.
1) Problem analysis: From time to time, we find papers
discussing or debating the issues of reform [56], cri-
sis [6], [13], issues for policies [55], privatization [71],
uncertainty [92], optimization [97], fraud [32], [88], and
effect on economy [30], society, capital market [31], hu-
man resources [90], [93], etc.
2) Process and policy modeling: Different approaches, e.g.,
empirical analysis, time-series analysis, quantitative com-
parative analysis, and equilibrium analysis [42], have been
used and developed to design, simulate, and evaluate pol-
icy, pension, benefit [7], process and their effects, as well
as their optimization, choice [65], and performance rat-
ing [62] including accuracy [45].
Social Security Data Mining Goals
We summarize the main data mining goals in the social secu-
rity area into the following five classes, according to our under-
standing and practices of key entities, problems, and challenges
in social welfare business by data mining: 1) overpayment-
centric analysis; 2) customer-centric analysis; 3) policy-centric
analysis; 4) process-centric analysis; and 5) fraud-centric anal-
ysis [121]. They are shown in Fig. 3 and are explained briefly
in the following.
1) Payment-Centric Analysis: Overpayments or govern-
ment customer debt are a major concern in social security
government services [119]. Overpayment/debt-centric analysis,
therefore, emerges as a major objective of SSDM. Its goals
consist of the deep understanding of the distributions of over-
payments across business lines, the cause and effect of over-
payments, and the evolution and changes of overpayments in
the life of government customers. In addition, issues that are
related to payment accuracy also involve underpayment anal-
ysis, and alignment and gap analysis between customer earn-
ing/employer payment and government payment. The findings
from payment-centric analysis contribute to government cus-
tomer debt recovery, debt prevention, and debt prediction, as
well as better customer service quality.
CONCLUSION
With the occurrence of the global financial crisis, more and
more governments have realized the necessity of enhancing so-
cial security services objectives and quality. Data mining and
machine learning can play a critical role, as we have demon-
strated in mining Australian social security data for debt pre-
vention, recovery, customer analysis, etc., during the past few
years. However, as the literature review shows, mining social
security (and public sector) data are still an open field for busi-
ness applications in data mining and machine learning. Very
few references have been publicized. In this paper, for the first
time in the community, we present a picture of studies on social
security issues and summarize the key concepts, goals, tasks,
and challenges of SSDM, based on our experience and knowl-
edge accumulated through conducting data mining in Australian
social security data.