Gaimei Miao and Juanqiong Gou
School of Economics and Management, Beijing Jiaotong University, Haidian District, 100044 Beijing, China
Keywords: Metadata, Integration, Yearbook.
Abstract: This paper summarizes the development of data integration and the present situation of the development in
domestic and abroad. It also analyzes the yearbook data, find that different years, different regions of the
data has very big differences in the structure, name, dimension, data type, and other aspects, and the current
data integration technology cannot be effectively integrate yearbook data. However, metadata exactly can
well solve the problems, using metadata not only can realize the effective integration about local yearbook
data, but also can achieve different platforms, different subsystem, different software sharing yearbook data.
So, metadata is helpful for effectively improving yearbook data utilization rate.
The worldwide Data Integration project index drawn
by Patrick Ziegler, as shown in figure 1, describes
the development of data integration.
The traditional data integration technology
mainly includes many multiple database System and
the federal database System. With the development
of the distributed network technology, in order to
increase the processing of Web data and semi-
structured data and integrate data sources which
have new forms, new technologies occurred , such
as the data integration system based on agent , the
integration technology based on ontology, etc. As
web service technology developed, researchers
launched a research based on the web service
integration technology.
Compared with the overseas research, domestic
research about the data integration technology
started relatively late, however, it is developed very
quickly. Relatively speaking, the researches that
southeast university computer science and
engineering researchers did was earlier that they
developed versatile based on a distributed
heterogeneous data sources integration system
prototype — CORBA, aiming at integrating data
from different data sources in the way of playing at
the same time. Chinese people's university
researchers paid more attention to doing research on
the question of Web enquires and semi-structured
data model.
To sum up, at present both at home and abroad,
there is as yet a promising potential and
development space in the research of the integration
of heterogeneous data.
The mentioned representative research works as
above, almost all didn't refer to the data in-
consistency solution. And metadata can well solves
the problem; therefore, here we did a research on
how to achieve statistical yearbook data integration
by metadata, in order to achieve better results.
The article fully analyzes the different years,
different regions of the Yearbook, and thinks about
metadata management strategy. Finally it definite
the yearbook with four metadata (Yearbook, Special
Topic, Report, Statistics Field), they include the
relationship between layers. That is a yearbook
includes a number of topics, a topic includes several
reports, a report includes a lot of Statistics Field.
Meanwhile, it further refines the various
metadata. it uses the metadata (Yearbook ID,
Yearbook Name, Type, Start Date, End Date,
Publisher, Date of Purchase, Note) to define
Yearbook; uses the metadata (Special Topic ID,
Miao G. and Gou J..
DOI: 10.5220/0003586205690573
In Proceedings of the 13th International Conference on Enterprise Information Systems (PMSS-2011), pages 569-573
ISBN: 978-989-8425-56-0
2011 SCITEPRESS (Science and Technology Publications, Lda.)
Figure 1: Data Integration project index.
Belonged Yearbook ID, Chinese name, English
name, notes) to define Special Topic; uses the
metadata(Report ID, Belonged Special Topic ID,
Report Chinese name, Store name, notes) to define
Report; uses the metadata (SF_ID, Belonged Report
ID,SF-Name, SF-S-name, SF-Unit, SF-Level, SF-
Row-span, SF-Column-span, SF-Extend, notes) to
define Statistics Field. Through these definitions, it
can make the reports with same structure.
In order to realize isomorphic management of the
yearbook data from different years, different
regions, transforming the tables with different levels,
different structure into the two-dimensional structure
and storing it in Oracle database is critical. In the
integration of the yearbook, it uses the rows of
spanning, the columns of spanning, the rows of
extensions to solve this problem, and has achieved
good effect.
In order to complete mapping, it need to get out
relevant field from mass data table, then store it in
one or a few tables, as figure 2 shows.
Figure 2.
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
It is too fussy to put forward extract field from
many tables one by one, therefore it needs to be
dealt with it in a structural processing, when
unstructural fields switch into structural fields, we
need to deal with the following conflicts:
(1) Naming conflicts
Description: if two elements E1 and E2 express
the same entity, but different names, it is happened
the conflicts of naming synonym. When the
elements E1 and E2 have different names but means
different entity or concept it will occur the conflict
of homonymy. For example, using permanent
population at the end of years in the table T1 means
the number of permanent population; at the table T2,
using the total population expresses the number of
permanent population, then it will occurs the conflict
of class name synonymous between long-term
populations at the end of the years. Absolute
(number all workers) means the absolute number
of the workers' salary in the table A1; Absolute
(number/ all workers) means the absolute average
wages in the table A2. The conflict of class name
homonymy word is happened.
The forms of relational model conflicts: Using
class in the relation bag represents entity or concept,
so the name conflicts in the model performance class
names with the same form and synonymous
Solving strategy: the Abnormity synonyms
conflicts tip the synonyms checked by HUB
according to input the synonyms by users;
Homonymy conflicts find the conflict according to
matching class-name users choose the method to
solve the conflicts
(2) Date type conflicts
Description; suppose that A1 and A2 describe
the character of the same entity but have different
date model A1 and A2 have date model conflicts.
For example, the date model of agricultural output is
plastic in the table 1. The date model of agricultural
output is character type in the table2.
The form of the conflicts of relation model;
relevance of column and date type expresses the date
type of a field. So the conflict is happen, when data
type of relevant field is different.
Solving strategy: Matching means the same
character of column, which is relevant for the date
type, users can choose date type.
(3) The conflict of Data dimension
Description: Suppose A1and A2 describe the
same features of the same entity, but date dimension
is different between A1and A2, they exist the
conflict of date dimension. Such as using meter
expresses height in the table 1; using inch expresses
height in table 2.It cannot know the date dimension
in the date dictionary. That is to say, database
metadata does not provide dimension of semantics.
The form of conflict model; there is no the
express way of the date dimension, when model
integration, it does not check the date dimension.
Suppose that users understand the conflicts of the
date dimension in the two tables, user solve it by
(4) Numerical range and Precision conflict
Description: related objects equivalent data
elements have different range and accuracy settings.
For instance, in the table T1, Agricultural
output value of the unit price is six-figure, the two-
figure behind point, such as 1000.82;in the table
2,the unit price of the total agriculture production is
five- figure, the behind of the point is one- figure,
such as 1200.5.
The form of the relation model conflict;
attribute setting in the relation bag express the scope
and precision which is list in the date base, the
conflicts of scope and precision in the model express
the inconformity of the attribute setting of column.
Solving strategy: users make sure the scope and
precision according to the need of statistical
analysis, and get rid of noise data.
(5) The description of constraints conflict
Description: related objects equivalent data
elements have different examples constraint. Such as
the age of the adult in the tableT1 must be over 18
years old, and it is above 20 years old in the table
The form of the relation model conflict; the
relevant for the element in the relation bag and
constraint is based on the constraints of the element.
Such as the relevance of the column and constraint,
the attribute of constraint .body is above 18 years
old .which means examples must be above 18.In the
model, the constraints conflict is the same element
of the expression conflict which is based on the
Solving strategy: matching the same element,
which is relevant for the constraint, if the relevant
expression is different, users decide whether it has
constraints conflict and conflict resolution or the
conflict solved by user is only the constraints
conflict which is possible to occur. The specific
expression meaning is solved by users.
(6) The primary key conflicts
Description; Established in related objects of
different only marks. Such as, the primary key is
numbers which is in the table T1 (numbers, years,
trade … ),the primary key is trade in the table T2
(numbers, years, trade … ).
The form of relation model conflict; the
relevance among tables, columns, primary keys is
used in the relation bag, which marks a column as
primary key. Primary key conflict in the model is
built up the relation among the same table, different
columns, and primary keys.
Solving strategy: we should check column
which is related with primary key, if column is
different, users choose the way of solving conflict.
(7) Structural conflicts
Description; the same entity or concept use different
representation methods, one represents entity, the
other one represents features. Such as table T1
(numbers, years, animal husbandry and fishery
output…) represents animal husbandry and fishery
output; it is the attribute of table T2 (numbers, years,
absolute number / (animal husbandry and fishery
index) (last year is equal to 100/), animal husbandry
and fishery).
The form of relation model conflict; table in the
relation bag represents the table which is in the date
base, column represents line which is in the date
base, the structural conflict in the model express
table which is in the model of address, in a model
address is column.
Solving strategy: matching table class name
and column class name which is under the different
table ,we can find out the possible structure conflict,
but the complexity of the class name across the
namespace matching time is big, and there exists a
lot of similar name which is under the different
namespace, HUB submit mass of the possible
conflicts, which is estimate by users .however, the
structural conflict is seldom, so the benefit it brings
is little, therefore we do not check the structural
conflict. Suppose that users realize the structural
conflicts of two tables, they solve structural
conflict, otherwise there are too many date in the
integrated table.
Unified metadata standard is the most need of
Yearbook data of cross-platform sharing, and the
CWM (Common Warehouse Model), which the
international object management group proposed to,
is just such kind of metadata exchange standards in
the fields of data warehouse and business analysis,
which realizes the exchange of metadata in different
software tools, as figure 3 shows.
Figure 3.
CWM is in fact a kind of exchange technology,
aiming at promoting the metadata exchange
activities between different software tools. In the
data integration based on CWM, I use a metadata
store tool to store the metadata used in the exchange,
and by accessing metadata store components directly
have the access to the exchange metadata needed,
adopting metadata store tool can lessen the workload
involved in the exchange of metadata. At the same
time if every kind of tool use CWM that is the
general format to input and output metadata, all
software can understand each other and exchange
metadata, but also can exchange metadata with all
other tools.
Create XML exchanging documents according to
the document type definition in the metadata
integration, which can directly convert to the object
related to the metadata tools, realizing the exchange
of metadata in the condition of CWM as sharing
The essence of CWM metadata exchange,
namely the exchange of class and associated
examples, is the ability of exchanging in any middle
format which can make them mutual agreement
aiming at CWM. Once started the CWM metadata
transforming mechanism, it would send the metadata
of public format to any a kind of tool, not need to
specify the concrete tools to create this exchange.
Therefore, we can use a metadata store as the
medium of CWM metadata exchange.
Due to the CWM metadata store is a base
relational database, however the meta store extracted
out based on the CWM modeling is object-oriented,
therefore the first problem needed to solve is to
ensure meta store of CWM keep object-oriented
logic structure and make them map to the relation
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
This paper makes full use of the characteristics of
the metadata, firstly it makes the data of yearbook
isomorphic management locally; Secondly by
mapping and extracting the field needed, we can
make evaluation and analysis using the yearbook
data; Finally, using the CWM exchange technology,
we complete the data sharing among different
platforms, different subsystem and software. That is,
it improves yearbook data usage rate effectively.
This paper was supported by “the Fundamental
Research Funds for the Central Universities
John Poole, Dan Chang, Douglas Tolbert, David Mellor et
al. Common Warehouse Model Developers Guide
Wiley Publishing, 2003.
Dr. Daniel, T. Chang. Common Warehouse Meta-model
(CWM), UML and XML. Meta Data Conference,
2000, 6(4): 1 9-23.
William Rub, Enterprise Application Integration. John
Wiley &Sons2002
Daniel TChang, Common Warehouse Meta-model
(CWM), UML and XML Meta Data Conference,
March 19·23, 2000.