Extended Change Identification System
Parimala N.
1
and Vinay Gautam
2
1
School of Computer & Systems Sciences, Jawaharlal Nehru University, New Delhi 110067, India
2
Chandigarh Engineering College, Landran, Mohali, Punjab 140307, India
Keywords: Multi-version Metadata, Metadata, Ontology, Data Warehouse.
Abstract: Schema evolution leads to multiple versions of the data warehouse schema. We address the issue of whether
the information required by the decision maker is present in some version of the data warehouse or not by
checking all the versions for the existence or the absence of the required information. The user specifies the
sought information using business terms. We build Delta Ontology which captures the ontology information
in terms of mapping between business terms and schema terms. The Delta Ontology is built for the
modifications to the schema as it evolves. We propose an algorithm to search for the information in the
latest version of E-Metadata and the Delta Ontology. Our algorithm lists all the versions where the
information is available giving precedence to finding the information in a single version over across
versions. The decision maker is also informed if the information is totally missing.
1 INTRODUCTION
We are looking at the definition of the warehouse
per se and analyzing its adequacies to meet the needs
of the user. If the current data warehouse does not
meet the needs of the user then it implies that there
is a gap between the information content of the
warehouse and the information that is needed by the
decision maker. Further, the latest warehouse
schema could have evolved over time. That is, there
could be changes made to the schema more than
once. It is possible that the information sought by
the user, which is absent in the current schema,
existed in some earlier version of the schema and
has been subsequently deleted. If it is possible to
trace the changes, then this trace can help the user in
analyzing the manner in which the gap has arisen.
Consider an example of an Insurance Schema
represented as a star schema shown in Figure 1. The
Multi-dimensional schema has Policy_holder,
Policy, Claim and Time as dimensions. Policy
Revenue is the fact with Premium_dollar and
Coverage_period as measures. Let the schema in
Figure 1 be the latest version, version3. Let us
assume that an attribute Holder_gender was present
in version 1 and was subsequently deleted. It is,
therefore, not present in version2 and version3.
Consider the case where it is needed to get “the
revenue that was generated gender wise for each
city”. If we look at the latest version of the schema
as given in Figure 1,we see that ‘revenue’ is defined
as premium_dollar and there is no dimensional
attribute ‘gender’ in version3. However, the
attribute, ‘gender’, was defined as Holder_gender in
the earlier versions. Further, the attribute ‘city’ is
part of the attribute Holder_address. In other words,
three things can be observed.
1. The decision maker is expressing the information
using business terms which are not the same as
schema terms.
2. The information may not be directly represented.
In the above example, the attribute city is not
directly represented but is part of the attribute
address. Different cases arise when information
is not directly represented. These have to be
identified.
3. The sought information could be missing.
Information could be missing either because it is
not defined in the latest version and existed in an
earlier version since the schema has evolved over
time or because it is altogether absent.
We briefly outline the solutions to the problems
listed above.
1. Ontology is defined consisting of business as
well as schema terms and a mapping between
these. The schema terms of all the schema
versions are grouped into two groups – the terms
N., P. and Gautam, V.
Extended Change Identification System.
In Proceedings of the 11th International Conference on Evaluation of Novel Software Approaches to Software Engineering (ENASE 2016), pages 51-58
ISBN: 978-989-758-189-2
Copyright
c
2016 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
51
that have remained constant across versions and
those that have participated in the evolution.
Delta Ontology (DO) has ontology entries for the
changed schema terms across versions.
2. A set of rules are defined to identify the form in
which the information is present in the latest
version or existed in the earlier versions.
3. The third issue is addressed by checking whether
the information existed in any earlier definition
of the schema.
Figure 1: Insurance Warehouse Schema.
Our system, known as the eXtended Change
Identification System (X-CIS), is shown in Figure 2.
First, we build the DO. Next, the Information
Processing component searches for the terms
specified by the user across schemata. The
ontological entries for the constant terms are
available in the latest version of E-Metadata (section
2.2). Thus, instead of a blind search across versions,
we search in the latest E-Metadata and DO.
Figure 2: X-CIS Architecture.
The contribution of the paper is three fold:
Firstly, we show the manner in which Delta
Ontology is built. Secondly, we propose Delta Scan
Algorithm (DSA) to search for the requested
information across data warehouse versions. Thirdly,
we identify the forms in which the requested terms
are present.
1.1 Related Work
Conventional approaches for the management of
changes to a multidimensional schema and the
contents (which is the data warehouse) can be
broadly classified into two categories namely,
schema evolution where the changes are made to the
multidimensional structure without retaining the
existing definition (Blaschka et al., 1999; Kaas et al.,
2004; Bebel et al., 2006) and version extension,
where all the versions are maintained (Body et al.,
2002; Shazad et al., 2005; Golfarelli, 2006).The
evolution in a data warehouse schema affects the
data warehouse metadata as described in (Vaduva
and Dittrich, 2001; Pan, 2010).
Different versions of data warehouse metadata
have been maintained and queried for different
purposes. In (Wrembel and Bębel, 2007) the user is
informed about the missing attributes in a query by
means of meta-information that is attached to the
result. The user is also informed about the changes
in the structure. In (Leja et al., 2010) the query
language, MVDWQL, supports two types of queries,
namely content queries and metadata queries
through a GUI. Multiple version data can be queried
using the former. The history of evolution can be
sought by executing a metadata query. A set of
traces which relate multidimensional elements and
data sources is maintained in (Maté and Trujillo,
2014). With the help of these traces, changes to the
schema when data sources change are easily
incorporated. In all these systems, the user queries
are restricted to using schema terms.
In this paper, we explore the schema definitions
in the different versions and the history of changes
to find the versions where the information sought by
the user is present. The user's need is expressed
using business terms. Towards this, we build delta
ontology using the changes reflected in the multi
version metadata. Our work is different in that the
information sought by the user is expressed in
business terms and not warehouse schema names.
Secondly, we do not query multiple metadata
versions but build Delta Ontology which records
only the changes.
The rest of the paper is organized as follows.
Definitions are given in section 2.In section 3, the
creation of Delta Ontology is explained. The X-CIS
architecture is explained in section 4. Section 5 is
the concluding section.
2 DEFINITIONS
In this section, we define the terms used in this
paper.
ENASE 2016 - 11th International Conference on Evaluation of Novel Software Approaches to Software Engineering
52
2.1 Analysis Component
The information needed by the decision maker is as
given in (Parimala and Gautam, 2010). It is
expressed using two terms ‘what’ and ‘how’ where
a)‘what’ or “what is to be analyzed" describes the
data to be analyzed. It refers to the measures which
represents the factual data.
b)‘how’ describes the business perspective under
which data analysis is to be performed. It specifies
the dimensional attributes along which the measure
is to be analyzed.
2.2 E-Metadata
E-Metadata consists of the technical metadata of the
data warehouse schema and the ontology as defined
in (Parimala and Gautam, 2011). The technical
metadata is extracted from the metadata of a
warehouse schema. The ontology for the terms in the
technical metadata is built using the WordNet.
Figure 3 shows the partial E-Metadata entry for
Policy_Holder. The entry says, for example, that
customer is a synonym of Policy_Holder.
<Dimension-3-Policy_Holder>
<Synset-of-Dimension-3>Syn-
customer</Synset-of-Dimension-3>
<Term-of-Dimension-3>Meronyms-</Term-
of-Dimension-3>
<Hierarchy-of-Dimension-3>Hiper-Enter
Hypernym of Policy_Holder</Hierarchy-
of-Dimension-3>
</Dimension-3-Policy_Holder>
Figure 3: E-Metadata Entry.
3 DELTA ONTOLOGY
Delta Ontology (DO) is expressed using OWL. DO
has six ontology classes which are DVersion,
Operation_type, DWTerm, Label, Synset and Term.
DVersion contains the version of a delta file.
Corresponding to each term of an entry in the delta
file (as shown above) we define three classes namely
Operation_type, DWTerm and Label. In addition,
two new classes Synset and Term are introduced
where Synset class denotes the domain terms which
have the same sense as the terms in Label class and
Merset class denotes meronyms of terms in the
Label class.
The relation between the classes in the ontology
is represented as OWL Property. Relations in DO
show the associations between the classes. These
relations are Oper_DWT, Oper_Label, Label_DWT,
Label_Synset ,Label_Merset and DVersion_Oper.
Oper_DWT is a relation between the classes
Operation_type and DWTerm which shows whether
a DWTerm is added or deleted. The rest of the
relations are self explanatory.
3.1 Delta Ontology Development
We create a repository of the changes that have
taken place. As brought out in (Pan, 2010; Saddad et
al., 2008), the changes that the user can make to the
technical metadata schema are insertion, deletion
and renaming of any of a dimension, a dimensional
attribute and a fact attribute.
A new version of the technical metadata reflects
the changes to the data warehouse schema. The
Delta file, D
i,i+1
, shows the difference between
technical metadata version i, TM
i
and technical
metadata version i+1,TM
i+1
(Gautam and Parimala,
2012). The Delta Ontology is built by extracting the
information from the delta file. The changes can be
addition, deletion or renaming of schema names. In
the new version of the technical metadata, the newly
added elements are inserted; the deleted ones are not
reflected. As far as rename is concerned the new
name is added to the new version and the old name
exists in the previous version. Thus, an entry in the
delta file is of the form:
<Operation_type DWTerm = Label [DWTerm =
Label2]>
Here, Operation_type is one of insertion, deletion or
rename. It reflects the operation used to perform the
movement from one version to the next. DWTerm is
either a Fact, Dimension, Measure or Dimension
Attribute. Label is a term from a warehouse schema
such as city, policy etc. As an example of an entry
consider
Example 1: <Insert Attribute=’Quarter’>
which says that the attribute ‘Quarter' has been
inserted.
The optional specification, ‘DWTerm = Label2’,
is valid only when Operation_type is Rename. For
e.g. the following entry in the delta file shows that
‘Holder_sname’ is renamed as ‘Holder_lastname'.
Example 2: <Rename Attribute =
”Holder_sname” Attribute=”Holder_lastname”>
The Delta Ontology Development Process
(DODP) starts by creating an instance of DVersion.
The instance contains the version of the delta file
Extended Change Identification System
53
D
i,i+1
. Next, Java API is used to extract the terms,
referred to as a ‘token’, from the delta file D
i,i+1
.
.
In the next step, the base class for each ‘token’,
from among the DO classes (section 4), is identified.
The following rules are used to identify the base
class for a given token.
R1: If (token = Operation_type) then
“Base Concept of token is
Operation_type”
R2: If (token = DWTerm) then “Base
Concept of token is the DWTerm”
R3: If (token = Label) then “Base
Concept of token is the Label”.
The token itself, is added as an instance of the base
class to which it belongs. If the ‘Operation_type’ is
‘Rename, then it is treated as if the old Label is
deleted and the new Label has been inserted.
Therefore, the system will create two instances of
Operation_type which are ‘Delete’ and ‘Insert’.
Consider Example 1. The extraction process will
extract three tokens namely Insert, Attribute and
Quarter. The token ‘Insert’ is added as an instance
of Operation_type’, ‘Attribute’ as an instance of
‘DWTerm’ class and ‘Quarter' as an instance of
‘Label’ class. In Example 2, both
‘Holder_sname'and ‘Holder_lastname’ become
instances of Label class. Since the operation is
rename, two instances of Operation_type’, ‘Insert’
and ‘Delete’ are created.
Subsequently, the domain concepts are added
using WordNet. In this step, the WordNet is used to
extract terms (Synnonym and Meronyms) for the
tokens belonging to the class ‘label.’ Synonyms are
words with similar meaning and meronyms show
partof relationship of a token with WordNet terms.
The synonyms are added as instances of Synset; the
meronyms are added as instances of Term in the
DO.
The above information can be added provided
the token is found in the WordNet (Canas et al.,
2003; Miller and Hristea, 2006; Miller, 1995). If the
token is not found in the WordNet, then the Data
Warehouse Administrator (DWA) is asked to
specify an equivalent term which can be found in the
WordNet. The approach of asking for an equivalent
term was prompted by studying different example
schemas. (http://merc.tv/img/Figure./ Adventure
WorksDW2008.pdf, http://www.information-
management-architect.com /star-schema .html). The
schema term CalenderYear has no entry in
WordNet. However, the word Year is present in
WordNet. If the DWA can specify, Year as an
equivalent term then it enhances the ontology.
Figure 4: Delta Ontology after Updates.
Next, the relations between the instances are
established. All the instances created above are
linked to their respective classes. All the instances
are also linked to the DVersion instance. Further, all
the instances of the Label class are linked to the
corresponding instance of Operation_type. Figure 4
shows DO for Example 2.
4 X-CIS ARCHITECTURE
Once the Delta Ontology is built, eXtended Change
Identification System (X-CIS) identifies the form
and the versions where the requested terms are
present. The X-CIS architecture is shown in Figure
2. The input to the system is 'what' and 'how' terms
specified as the analysis component. X-CIS uses the
latest version of E-Metadata and DO to search for
the information. The output of the system is either
the versions where the data is available or a
suggestion for a change in the warehouse schema, if
the term is absent.
The manner in which E-metadata is searched is
as given in (Parimala and Gautam, 2010). Here, we
explain the Delta Scan Algorithm (DSA) which
looks for the terms in the DO. Before explaining
DSA, the guidelines for identifying the forms in
which the terms may appear is given.
4.1 Forms of Terms
The existence of an analysis component term in the
DO has two aspects to it. The first is whether the
term is directly or indirectly available. The second is
the version where it is available since the search
spans across versions. We consider each of these in
turn before explaining the order in which the result
is presented to the user.
Directly Available
If the ‘what’ term is defined as a measure in the E-
Metadata, we say that it is directly available. The
corresponding statement is
ENASE 2016 - 11th International Conference on Evaluation of Novel Software Approaches to Software Engineering
54
WS1: is a measure
Similarly, the ‘how’ term is directly defined if it
is the name of a dimensional attribute. The
corresponding statement is
HS1: is a name of an attribute of a dimension.
Indirectly Available
If a term is not directly available, then, it may be
defined in the schema but not as expected or it may
be an ontological equivalent. In these cases, we say
that the term is indirectly available. The different
ways in which ‘what’ can be indirectly available is
as follows:
WS2: is a synonym of a measure
WS3: is the name of the fact table.
WS4: is a dimension name.
WS5: is a name of an attribute of a dimension.
The different ways in which ‘how’ is indirectly
available is as follows:
HS2: is a synonym of a dimension attribute.
HS3: is a meronym of a dimension attribute.
HS4: is a dimension name.
HS5: is a fact.
HS6: is a fact attribute (measure).
Table 1: Result Options
Version Possibilities
Consider, now, the second aspect of finding the
‘what’ and ‘how’ terms across versions. Both the
terms may be defined in the same version, either
directly or indirectly or they me be defined in
different versions, again directly or indirectly. Table
1 shows these different possibilities. If the 'what' or
the ‘how’ term is directly available in version i then
it is denoted as D
i
; ID
i
implies that the term is
indirectly available in version i. N says that the term
is not available.
Message
The result of searching a ‘what’ term is expressed as
WS
k
V
i
= ‘what term is available as WS
k
in version i
and that of ‘how’ term is HS
k
V
i
= ‘ how term is
available as HS
k
in version i’.
Result Order
When a term is available in more than one version,
instead of just listing the versions, we order the
result based on our belief that users would prefer to
find ‘what’ and ‘how’ in the same version to finding
‘what’ and ‘how’ across versions. This is because if
the information is found in the same version then the
user can query the corresponding data warehouse.
X-CIS will display the information according to the
preference order given below.
Preference 1: Both are directly available in the same
version (Sl. No. 1 of Table 1).
Preference 2: Both are directly available across
versions (Sl. No. 2 of Table 1).
Preference 3: Same or across versions, one of them
is directly available (Sl. No. 3 and 4 of Table 1).
Preference 4: One of them is available either directly
or indirectly and the other is not available (Sl. No. 5
and 6 of Table 1.)
If the information is not available in any version (Sl.
No. 7 of Table 1), then we suggest that changes may
have to be made to the current version.
4.2 Delta Scan Algorithm
Let V
1
,V
2
,…,V
n
be the E-Metadata versions with
V
n
as the latest version. DSA has two parts. The first
is the creation of the two matrices WAvail and
HAvail. The second is to scan the matrices and
inform the user about the evolution.
In the rules given below, IsEqual( ) is a Boolean
function which compares two terms and returns true
if they refer to the same term; otherwise it returns
false (Parimala and Gautam, 2010).
The rules given below set the values in the
matrices WAvail and HAvail. The entries in Delta
Ontology for a ‘what’ or a ‘how’ term specify the
operation(Insert or Delete ) which created the term
which is an instance of the Label class. The
operation is linked to an instance of DVersion. If the
operation is Insert, then it is available in the higher
version; if it is Delete, then the term is available in
the lower version. The version instance is of the
form D
i,i+1
. Thus, if the operation is Insert, then it
implies that the term is now available in the schema
version i+1. On the other hand, if the operation is
Delete, then it means that the term was available in
schema version i. In the former case, (i+1)th row of
the matrices is set to 1 and ith row in the latter case.
Consider, next, the columns. A column in WAvail is
set depending on the considerations of a match
between ‘what’ term and other schema terms as
given in section. Similar considerations for the
‘how’ term determine which column of HAvail is set
to 1. If any rule fires, then the remaining rules are
skipped.
Extended Change Identification System
55
4.2.1 Matrix Creation
Rules for ‘what’
1) If the ‘what’ term is an instance of
‘Label’ class and related version
DVersion = i,i+1 then
If
IsEqual(‘what’,‘Fact_Attribute’)then
If Operation_type =‘Insert’ then
setWAvail[i+1,1] to 1
If Operation_type =‘Delete’ then
setWAvail[i,1] to 1
If IsEqual(‘what’,‘Fact’) then
If Operation_type =‘Insert’ then
setWAvail[i+1,3] to 1
If Operation_type =‘Delete’ then
setWAvail[i,3] to 1
If IsEqual(‘what’,‘Dimension’) then
setWAvail[i+1,4] to 1
If Operation_type =‘Insert’ then
setWAvail[i+1,4] to 1
If Operation_type =‘Delete’ then
setWAvail[i,4] to 1
If IsEqual(‘what’,’Dim_Attribute’)
then
If Operation_type =‘Insert’ then
setWAvail[i+1,5] to 1
If Operation_type =‘Delete’ then
setWAvail[i,5] to 1
2) If the ‘what’ term is an instance of
‘Synset’ class then the related
instance of ‘Label’ is picked up. If
related version DVersion = i,i+1
then
If IsEqual(‘related Label
instance’, ‘Fact_Attribute’) then
If Operation_type =‘Insert’ then
setWAvail[i+1,2] to 1
If Operation_type =‘Delete’ then
setWAvail[i,2] to 1
Rules for ’how’
1) If the ‘how’ term is an instance of
‘Label’ class and related version
DVersion = i,i+1 then
If IsEqual(‘how’, ‘Dim_Attribute)
then
If Operation_type =‘Insert’ then
setHAvail[i+1,1] to 1
If Operation_type =‘Delete’ then
setWAvail[i,1] to 1
If IsEqual (‘how’,’Dimension’) then
If Operation_type =‘Insert’ then
setWAvail[i+1,4] to 1
If Operation_type =‘Delete’ then
setWAvail[i,4] to 1
If IsEqual (‘how’,’Fact’) then
If Operation_type =‘Insert’ then
setWAvail[i+1,5] to 1
If Operation_type =‘Delete’ then
setWAvail[i,5] to 1
If IsEqual (‘how’, ’Fact_Attribute’
then
If Operation_type =‘Insert’ then
setWAvail[i+1,6] to 1
If Operation_type =‘Delete’ then
setWAvail[i,6] to 1
2) If the ‘how’ term is an instance of
‘Synset’ class then the related
instance of ‘Label’ is picked up. If
related version DVersion = i,i+1
then
If IsEqual (‘related Label
instance’, ‘Fact_Attribute’) then
If Operation_type =‘Insert’ then
setWAvail[i+1,2] to 1
If Operation_type =‘Delete’ then
setWAvail[i,2] to 1
As an example, let us say that the user wants to
analyze ‘Revenue’ ('what') along ‘gender' ('how'). As
given in section 1, ‘Revenue’ is a synonym of the
measure Premiun_dollars which is available in all
the three versions of the schema. 'gender' as
'Holder_gender' was available only in version 1. The
resulting matrices are shown below.
Table 2: WAvail Matrix.
Table 3: HAvail Matrix.
4.2.2 Response
Using the matrices built in the previous section, it is
possible to inform the user about ‘what’ and ‘how’
terms. Recall that in section 3, we showed that the
result is of the form WS
k
D
i
or HS
k
D
i
. The mapping
between the result and the matrices is given below:
The matrices are repeatedly scanned to list the
result in terms of preferences.
ENASE 2016 - 11th International Conference on Evaluation of Novel Software Approaches to Software Engineering
56
Preference1 :If WAvail[i,1]=1 and
HAwail[i,1]=1 then message is WS
1
V
i
and
HS
1
V
i
, 1i n
Preference2 :If WAvail[i,1]=1 and
HAwail[j,1]=1 then message is WS
1
V
i
and
HS
1
V
j,
1in, 1j n, i j
Preference3 :If WAvail[i,1]=1 and
HAwail[i,k]=1 then message is WS
1
V
i
and
HS
k
V
i
,1in, 2k6
If WAvail[i,k]=1 and HAwail[i,1]=1
then message is WS
k
V
i
and HS
1
V
i
,1in,
2k5
Preference4 :If matrices WAvailand
HAvail are all zeroes then the message
is ‘what’ term and 'how' ‘are not
available’
If matrix WAvail is all zeroes and if
HAvail[i,k]=1 for 1k5 then the
message is ‘what’ term ‘is not
available’ and HS
k
V
i,
1in
If matrix HAvail is all zeroes and and
if WAvail[i,k]=1 for 1k6 then the
message is ‘how’ term ‘is not
available’ andWS
k
V
i,
1in
4.2.3 Time Complexity of DSA
DSA algorithm scans and compares the attributes of
the latest E-Metadata version and the DO entries
once. The number of comparisons is the number of
attributes in the latest E-Metadata (say m) + no. of
entries in the Delta Ontology (say k). Therefore,
complexity= O (m+k)
In order to analyze the complexity, we compare it
with the complexity of an algorithm in the absence
of DO. In this case, an algorithm to find ‘what’ and
‘how’ terms, would have to scan and compare each
attribute of each and every version of E-Metadata.
Therefore,
Complexity = average no. of attributes in a E-
Metadata version (m) * number of E-Metadata
versions (n)
= O (m*n)
Let us analyze n and k.
Case1: If k is very large and is equal (n-1)*m then
the complexity of DSA is the same as complete scan
algorithm.
Case 2: If k << (n-1)*m then the complexity DSA is
much lower.
Case 1 will occur if the number of changes is as high
as the E-Metadata itself. This is a highly unlikely
scenario. Case 2 will occur most of the time.
It must be noted that in both the cases matrices
have to be built to give the complete picture to the
user and further, the time taken to scan the matrices
would be the same for both the algorithms.
5 CONCLUSION
The aim is to find whether the multiple warehouse
versions cater to the needs of the business user. The
needs are expressed using ‘what’ and ‘how’ terms.
These may be business terms and not necessarily
schema terms. We have built Delta Ontology to
capture the mapping between business terms and
schema terms. The Delta Ontology itself is built
using the differences in the metadata of the
warehouse schema as it undergoes changes. All the
versions where the terms are available is picked up
and listed according to the preference order.
Java is used for building the prototype of the
system and SQL Server 2005 is used as back end
tool to store metadata of the warehouse.
It may be argued that special data structures for
sparse matrices can be used to store the contents of
WAvail and HAvail. Since the matrices are not very
large the time taken to scan them is not very high.
As far as populating them is concerned, DSA is
better in terms of time complexity than complete
scan of all versions.
It may be noted that we have not defined a query
language to query multi versions of the metadata.
Query language is appropriate when different ad hoc
queries are to be framed. In our case, we search only
for ‘what’ and ‘how’ terms. Therefore, an
appropriate GUI is built to accept these terms.
In our system, it is possible for the user to know
whether the missing information in the current
schema was available in an earlier version or not.
This, we believe, will help the user decide whether
or not to change the current data warehouse schema.
It is, also, possible to see whether earlier versions
were more in tune to the decision maker’s needs. It
will also give a feedback on the evolution of the
schema vis-a-vis it satisfying query needs.
REFERENCES
Bebel, B., Królikowski, Z. and Wrembel, R., 2006.
Managing evolution of data warehouses by means of
nested transactions. In Advances in Information
Systems (pp. 119-128). Springer Berlin Heidelberg.
Blaschka, M., Sapia, C. and Höfling, G., 1999. On schema
evolution in multidimensional databases. In Data
Warehousing and Knowledge Discovery(pp. 153-164).
Springer Berlin Heidelberg.
Body, M., Miquel, M., Bédard, Y. and Tchounikine, A.,
2003, March. Handling evolutions in multidimensional
structures. In Data Engineering, 2003. Proceedings.
19th International Conference on (pp. 581-591).
IEEE.
Extended Change Identification System
57
Cañas, A.J., Valerio, A., Lalinde-Pulido, J., Carvalho, M.
and Arguedas, M., 2003, October. Using WordNet for
word sense disambiguation to support concept map
construction. In String Processing and Information
Retrieval(pp. 350-359). Springer Berlin Heidelberg.
Gautam, V. and Parimala, N., 2012. E-Metadata
versioning system for data warehouse schema.
International Journal of Metadata, Semantics and
Ontologies, 7(2), pp.101-113.
Golfarelli, M., Lechtenbörger, J., Rizzi, S. and Vossen, G.,
2006. Schema versioning in data warehouses:
Enabling cross-version querying via schema
augmentation. Data & Knowledge Engineering, 59(2),
pp.435-459.
Kaas, C., Pedersen, T.B. and Rasmussen, B., 2004.
Schema evolution for stars and snowflakes. In
Proceedings of the Sixth International Conference on
Enterprise Information Systems.
Leja, W., Wrembel, R. and Ziembicki, R., 2009. On
Querying Data and Metadata in Multiversion Data
Warehouse. Data Warehousing Design and Advanced
Engineering Applications: Methods for Complex
Construction: Methods for Complex Construction,
p.206.
Maté, A. and Trujillo, J., 2014. Tracing conceptual
models' evolution in data warehouses by using the
model driven architecture. Computer Standards &
Interfaces, 36(5), pp.831-843.
Miller, G.A., 1995. WordNet: a lexical database for
English. Communications of the ACM, 38(11), pp.39-
41.
Miller, G.A. and Hristea, F., 2006. WordNet nouns:
Classes and instances. Computational linguistics,
32(1), pp.1-3.
Pan, D., 2010. Metadata version management for DW 2.0
environment'.Journal of Convergence Information
Technology, 5(3).
Parimala, N. and Gautam, V., 2010. CIS: Change
Identification System. In KEOD (pp. 347-350).
Parimala, N. and Gautam, V., 2011. Extended Metadata
for Data Warehouse Schema. In ENASE (pp. 254-259).
Saddad, E., El-Bastawissy, A., Rafea, M. and Hegazy, O.,
2008, March. Multiversion queries in
multidimensional structures. In Proceedings of the 6th
International Conference on Informatics and Systems
(INFOS'08).
Serra, I. and Girardi, R., 2011. A process for extracting
non-taxonomic relationships of ontologies from
text. Intelligent Information Management, 3, 119.
Shahzad, M.K., Nasir, J.A. and Pasha, M.A., 2005. CEV-
DW: Creation and Evolution of Versions in Data
Warehouse. Asian Journal of Information Technology,
4(10), pp.910-917.
Vaduva, A. and Dittrich, K.R., 2001. Metadata
management for data warehousing: between vision and
reality. In Database Engineering and Applications,
2001 International Symposium on. (pp. 129-135).
IEEE.
Wrembel, R. and Bębel, B., 2007. Metadata management
in a multiversion data warehouse. In Journal on data
semantics VIII (pp. 118-157). Springer Berlin
Heidelberg.
ENASE 2016 - 11th International Conference on Evaluation of Novel Software Approaches to Software Engineering
58