Comparative Analysis of State-of-the-Art Spatial Data Warehouse

Meta-models

Catching the Expressive Power of SDW Schemas!

Alfredo Cuzzocrea

and Robson do N. Fidalgo

ICAR-CNR and University of Calabria, Arcavacata di Rende, Italy

Center for Informatics, Federal University of Pernambuco, Recife, Brazil

Keywords: Spatial Data Warehouse and Metamodel.

Abstract: In this paper we provide a comparative analysis of the Spatial Data Warehouse Metamodel (SDWM)

proposal against three state-of-the-art Spatial Data Warehouses (SDW) meta-model proposals. Results of

this analysis allow us to conclude that the SDWM proposal exposes a higher expressive power of the

comparison approaches, and, in addition to this, it allows us to obtain more concise and compact SDW

schemas when compared with the schemas provided by the comparison approaches.

1 INTRODUCTION

Spatial Data Warehouse Metamodel (SDWM) (Del

Aguila et al., 2011; Cuzzocrea & Fidalgo, 2012a;

Cuzzocrea & Fidalgo, 2012b) has been provided

recently with the goal of effectively supporting the

modeling of Spatial Data Warehouses (SDW)

(Bédard et al., 2001; Zghal et al., 2003) by adding

several contributions. Among these, relevant ones

concern with separating the SDW conceptual

modeling from the OLAP data cube (Gray et al.,

1997) conceptual modeling, supporting SDW

complex constructs modeling, and, finally,

stereotyping attributes and measures as spatial

objects directly.

Another nice contribution due to this line of

research is represented by the proposal of a software

environment, called SDWCASE, which allows us to

model a SDW according to the SDW’s design and

modeling principles, in a user-friendly manner.

As a further research effort along the so-depicted

line of research, in this paper we provide a

comparative analysis of the SDWM proposal against

three state-of-the-art SDW meta-model proposals

(Fidalgo et al., 2004; Malinowski & Zimányi, 2007;

Glorio & Trujillo, 2008).

Results of this analysis allow us to conclude that

the SDWM proposal exposes a higher expressive

power of the comparison approaches, and, in

addition to this, it allows us to obtain more concise

and compact SDW schemas when compared with

the schemas provided by the comparison

approaches.

The remaining part of this paper is organized as

follows. In Section 2, we provide an overview of the

SDWM proposal. In Section 3, we introduce a

running example focusing on a SDW of homicide

cases for the secretary of the Public Safety Office of

Pernambuco/Brazil. Next Sections 4-6 are devoted

to the comparative analysis of SDW with the

comparison approaches: (Fidalgo et al., 2004)

(Section 4), (Malinowski & Zimányi, 2007) (Section

5), and (Glorio & Trujillo, 2008) (Section 6). In

Section 7, we provide the results of the comparative

analysis that is the main contribution of our research.

Finally, in Section 8, we provide conclusions and

future work of our research.

2 SDWM IN A NUTSHELL

SDWM (Del Aguila et al., 2011; Cuzzocrea &

Fidalgo, 2012a; Cuzzocrea & Fidalgo, 2012b) is a

meta-model that embeds the following significant

features: (i) disassociating DW dimensional

modeling from OLAP data cube modeling; (ii)

representing the spatiality of a SDW by directly

stereotyping attributes/measures as spatial types,

rather than stereotyping dimension/fact tables as

spatial or hybrid objects; (iii) capturing whether the

302

Cuzzocrea A. and do N. Fidalgo R..

Comparative Analysis of State-of-the-Art Spatial Data Warehouse Meta-models - Catching the Expressive Power of SDW Schemas!.

DOI: 10.5220/0004455903020309

In Proceedings of the 15th International Conference on Enterprise Information Systems (ICEIS-2013), pages 302-309

ISBN: 978-989-8565-60-0

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

geometry of a spatial attribute/measure can be

normalized and/or shared; (iv) supporting the

following DW modeling techniques: degenerated

dimensions, many-to-many relationships (bridge

tables), role-playing dimensions, which are

typically-hard modeling cases (Bédard et al., 2001;

Zghal et al., 2003); (v) providing a set of stereotypes

with pictograms that aim at being concise and user-

friendly; (vi) being used as a basic meta-model for

the CASE tool SDWCASE that supports the

modeling of logical SDW schemas, as well as, given

an input SDW schema, checking whether the

schema is syntactically valid.

In Figure 1, the UML class diagram of SDWM is

shown. Here, three relevant enumerations are

introduced: Cardinality, DataType and

GeometricType. Cardinality is used to define whether

a relationship is of kind many-to-one, one-to-many or

many-to-many. In turn, DataType and GeometricType

represent the primitive or spatial data types supported

by SDWM, respectively. Moreover, SDWM exposes

five main meta-classes: Schema, Table,

Relationship, DimensionColumn and FactColumn.

Schema is the root meta-class that corresponds to the

drawing area for a SDW schema. For this reason,

Schema is a composition of zero or more Table and

zero or more Relationship. Finally,

DimensionColumn and FactColumn are just a set of

different types of column.

Figure 1: SDWM UML class diagram.

Besides the previous constructors, SDWM is also

characterized by the following eight specialized

meta-classes (see Figure 1): Fact, Dimension,

Bridge, SpatialMeasure, DegenerateDimension,

ConventionalMeasure, SpatialAttribute and

ConventionalAttribute. These meta-classes address

the main concepts supported by the SDWM

modeling approach. On the basis of this approach, a

Table is specialized in Fact, Dimension or Bridge,

which capture the concepts of (SDW) fact table,

dimension table and a bridge table, respectively. A

FactColumn is specialized in SpatialMeasure,

DegenerateDimension and ConventionalMeasure,

which correspond to a spatial feature type, a

descriptive attribute and a measurable attribute,

respectively. A DimensionColumn is specialized in

SpatialAttribute and ConventionalAttribute, which

represent a spatial feature type and a descriptive

attribute, respectively. Furthermore, a Fact is a

composition of zero or more FactColumn and zero

or more ConventionalAttribute. In turn, a Dimension

and a Bridge are a composition of zero or more

DimensionColumn.

In order to capture tables that are source and

target in a relationship, SDWM introduces two

different associations, named as Source and Target,

respectively. Furthermore, since a dimension can

play different roles (role-playing dimensions),

SDWM introduces the attribute Role to support this

specialized modeling case. Other important SDWM

attributes are: Name, isNormalized, isShared,

hasDescription, Type and Size. Name is used to label

a meta-class. IsNormalized is used to define whether

the position (geometry) of a spatial measure/attribute

has to be normalized in a different table from its

location (description). IsShared is used to define

whether the position of a spatial attribute/measure

has to be shared among several spatial

attributes/measures (to this end, it is necessary to

define the same name and the same geometric type).

HasDescription is used to define whether the

location of a spatial measure has to be stored

(contrary to a SpatialAttribute, which must have a

position and a location, the location of a

SpatialMeasure is optional). Type is used to

associate a type (from the collection of allowed

SDWM types). Finally, Size is used to define the

length of a conventional attribute, a degenerated

dimension or a conventional measure.

SDWM makes use of stereotypes with

pictograms in order to increase its expressive power

and visualization capabilities, namely: Fact Table

, Dimension Table , Bridge Table ,

Conventional Attribute , Conventional Measure

, Degenerated Dimension , Spatial Attribute

, Spatial Measure , Relation , Integer ,

String , Date , Real , Point , Line ,

Polygon , Multipoint , Multiline ,

Multipolygon , Collection . The combined

action of these stereotypes allows us to design “rich”

SDW schemas.

ComparativeAnalysisofState-of-the-ArtSpatialDataWarehouseMeta-models-CatchingtheExpressivePowerofSDW

Schemas

303

3 RUNNING EXAMPLE: THE

HOMICIDE SDW

In order to assess the effectiveness of the proposed

SDWM approach, we developed a complete case

study focused on a SDW of homicide cases for the

secretary of the Public Safety Office of

Pernambuco/Brazil. This originated quite a complex

schema. A fragment of this schema designed by

means of SDWCASE is shown in Figure 2.

Figure 2: Fragment of the homicide SDW according to

SDWM.

As shown in Figure 2, the homicide SDW is

characterized by the following DW objects. Four

dimension tables: Date, Victim, Defendant and Arm,

which are stereotyped with . One bridge table:

Arms, which is stereotyped with . One fact table:

Homicide, which is stereotyped with . Two role-

playing dimensions: Case_Record_Date and

Homicide_Date. Two many-to-many relationships:

one between Homicide and Victim and another

between Homicide and Defendant. Four one-to-

many relationships: two between Homicide and

Date, one between Homicide and Arms, and one

between Arms and Arm. One degenerated

dimension: Case_Number, which is stereotyped with

. Two conventional measures:

Defendant_Quantity and Victim_Quantity, which are

stereotyped with . One spatial measure: Place,

which is stereotyped with . Twenty-seven

conventional attributes, such as Year,

Weighting_Factor, Name and Date, which are

stereotyped with , , and ,

respectively. Twelve spatial attributes, such as

Country, District and Address, which are

stereotyped with , and ,

respectively.

It is worth to noticing that, in the schema of

Figure 2, spatial attributes Country, Region, State

and City are defined as normalized (depicted with

bold font) and shared (depicted as Italic font).

According to the SDWM modeling approach, this

means that these spatial attributes have their

geometries stored in a table different of the table

containing their descriptions, and their geometries

can be reused between the dimensions Victim and

Defendant. This solution aims at reducing the

overall spatial data volume of the homicide SDW.

4 COMPARISON WITH

(FIDALGO ET AL., 2004)

(Fidalgo et al., 2004; Times et al., 2009; da Silva et

al., 2010) introduce a framework, a meta-model and

a CASE tool for modeling SDW. The proposed

meta-model is depicted in Figure 3 (da Silva et al.,

2010) while the logical model of a SDW focusing on

meteorology data (which makes use of their

proposed CASE tool) is showed in Figure 4 (Times

et al., 2009).

Figure 3: SDW meta-model proposed by (Fidalgo et al.,

2004).

As shown in Figure 3, the meta-model proposed

by (Fidalgo et al., 2004; Times et al., 2009; da Silva

et al., 2010) introduces classical constructs useful to

model a SDW, i.e.: attributes, measures, degenerated

dimensions, primary keys, foreign keys, common

measures, spatial measures, geographical

dimensions, conventional dimensions, hybrid

dimensions and fact tables. However, this models is

not totally complete as some important constructs

are still missing. Particularly, these constructs are:

many-to-many relationships (bridge tables), role-

playing dimensions and spatial attributes. These

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

304

constructs, indeed, are very useful to model real-life

SDW, like the case of our running example focusing

on the homicide SDW (see Figure 2). As a

consequence, we can infer that the meta-model by

(Fidalgo et al., 2004; Times et al., 2009; da Silva et

al., 2010) does not provide a full-support for modern

SDW.

Figure 4: Meteorology SDW according to (Fidalgo et al.,

2004).

As mentioned above, Figure 4 show the logical

model of the meteorology SDW designed by means

of the CASE tool by (Fidalgo et al., 2004; Times et

al., 2009; da Silva et al., 2010), which, obviously,

adheres to their proposed meta-model. Here, one fact

table is defined, i.e. Meteorology, and nine

dimensional tables, i.e. Hydrographic Basin, Basin

Location, Time, Data Collection Platform (DCP),

DCP Location, State, Meso Region, Micro Region,

City. As shown in Figure 4, fact table Meteorology

has only conventional measures (i.e., precipitation

and wind_speed) and dimension tables have both

conventional (i.e., nm_basin in Hydrographic Basin

and year in Time) and spatial (i.e., state in State and

dcp_location in DCP Location) attributes.

Moreover, the use of dimensions stereotyped with

spatial pictograms (i.e., Meso Region and Micro

Region) does not provide a concise/short notation,

as, for each spatial concept, one dimension is

introduced. It is worth to notice that this approach

pollutes the SDW schema as it results in an

excessive and redundant number of spatial concepts

immersed in the schema.

Figure 5 shows the meteorology SDW of Figure

4 modeled by means of SDWCASE according to

SDWM. This originates a SDW schema that is

equivalent to the schema of Figure 4. As an

Figure 5: Meteorology SDW according to SDWM.

alternative to the modeling due to (Fidalgo et al.,

2004; Times et al., 2009; da Silva et al., 2010) (see

Figure 4), in the SDWM modeling (see Figure 5) we

introduce city, micro_region, meso_region and state

as spatial attributes directly and their geometries are

normalized and shared between the dimensions

Hydrographic Basin and DCP. As is clearly follows

from the comparison between the two schemas in

Figure 4 and Figure 5, the use of spatial attributes

improves representation of the meteorology SDW

(and, in turn, its “visual quality”) by achieving a

more compact one, as six dimensions are no longer

modeled (i.e., Basin Location, DCP Location, State,

Meso Region, Micro Region, City).

On the other hand, from Figure 4 it also follows

that the meta-model proposed by (Fidalgo et al.,

2004; Times et al., 2009; da Silva et al., 2010) does

not mix the DW modeling concepts with the OLAP

data cube ones, similarly to the proposed meta-

model SDWM. This is, indeed, a positive

contribution.

Summarizing, from the results of this analysis we

can conclude that, with respect to the meta-model

proposed by (Fidalgo et al., 2004; Times et al., 2009;

da Silva et al., 2010), the proposed meta-model

SDWM is capable of achieving equivalent SDW

schemas in a much more compact and concise way,

thanks to the fact it can immerse spatial attributes

(and their geometries) directly into dimensions,

hence the expressive power of the SDWM proposal

is clearly higher.

5 COMPARISON WITH

(MALINOWSKI & ZIMÁNYI,

2007)

Malinowski & Zimányi (2007; 2009) propose a

SDW meta-model that introduces dimensions,

hierarchies, levels and measures, which all can be

ComparativeAnalysisofState-of-the-ArtSpatialDataWarehouseMeta-models-CatchingtheExpressivePowerofSDW

Schemas

305

spatial or not by simply declaring a Boolean

specifying state variable called Spatiality. The SDW

logical models adhering to such a meta-model are

represented as suitable extensions of the classical ER

model. Figure 6 (Malinowski & Zimányi, 2009)

shows the UML class diagram for the meta-model

proposed by Malinowski & Zimányi (2007; 2009).

As it follows from Figure 6, this meta-model mixes

DW modeling concepts with OLAP data cube

modeling ones, and it does not provide support for

the following DW modeling techniques: degenerated

dimensions, bridge tables (many-to-many

relationships) and role-playing dimensions. As a

consequence, the meta-model by Malinowski &

Zimányi (2007; 2009) does not allow a full

modeling of the homicide SDW of the running

example (see Section 3). In fact, this meta-model

neither allows specifying whether the geometry of a

spatial attribute can be normalized and/or shared nor

provides support for modeling the previously-

mentioned DW modeling techniques (which are

frequent cases in real-life SDW settings). Moreover,

to the best of our knowledge, there is no a CASE

tool based on this meta-model. This is another

relevant limitation of the proposal by Malinowski &

Zimányi (2007; 2009).

Figure 6: UML class diagram of the SDW meta-model

proposed by (Malinowski & Zimányi, 2007).

Figure 7 (Malinowski & Zimányi, 2007) shows

the logical model of a highway SDW according to

the proposal due to Malinowski & Zimányi (2007;

2009). As shown in Figure 6, the SDW schema

introduces one fact table, called Highway

Maintenance, with is characterized by conventional

and spatial measures (e.g., No. cars and Common

area), and dimensions/levels (e.g., Highway and

Highway Segment) with conventional and spatial

information (e.g., Road condition and State). From

Figure 7, it follows that representation of levels as

entities (it should be reminded that in this case ER

extensions are considered), besides being not

correspond to an intrinsic concept of DW, it does not

provide a concise representation/notation, as,

according to this approach, it is necessary to create

an entity for each level, hence polluting the SDW

schema significantly.

Figure 7: Highway SDW according to (Malinowski &

Zimányi, 2007).

Figure 8 shows the highway SDW of Figure 7

modeled by means of SDWCASE according to

SDWM. Again, the two schemas are equivalent.

Since, from Figure 7, it is not possible to know

whether the geometry of a spatial object is

normalized or shared, in the logical model of the

highway SDW according to SDWCASE of Figure 8,

we simply define all geometries as not normalized

and not shared. As a consequence, the SDW schema

of Figure 8 is characterized by significant

redundancy of geometric information, which, in

turn, increases the whole data volume of the final

SDW and, in addition to this, the SDW

administration itself becomes more difficult.

Similarly to the case of the comparison of SDWM

with the meta-model proposed by (Fidalgo et al.,

2004; Times et al., 2009; da Silva et al., 2010) (see

Section 4), here we again observe the clear

advantages deriving from immersing spatial

attributes (and their geometries) into dimensions

directly. In fact, in the highway SDW according to

SDWM (see Figure 8) we employ four dimensions

only whereas in the highway SDW according to

Malinowski & Zimányi (2007; 2009) (see Figure 7)

seven dimensions are necessary to represent the

same knowledge. This comparison clearly is in favor

of SDWM, which is capable of representing the

same knowledge in a more compact and concise

way.

Summarizing, from the results of this analysis we

can conclude that, with respect to the meta-model

proposed by Malinowski & Zimányi (2007; 2009),

the proposed meta-model SDWM exposes a clearly-

higher expressive power, as the meta-model due to

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

306

Malinowski & Zimányi (2007; 2009) does not

address some important DW modeling techniques

(i.e., degenerated dimensions, bridge tables and role-

playing dimensions) and also it does not allow to

specify whether the geometry of a spatial attribute

can be normalized and/or shared among dimensions,

thus preventing a full modeling of our running

example on the homicide SDW. In addition to this,

the meta-model due to Malinowski & Zimányi

(2007; 2009) introduces an entity for each

dimensional level, hence it clearly reduces the

clarity and the comprehensibility of final SDW

schemas (this drawback is much more evident with

real-life SDW schemas that are usually characterized

by high numbers of dimensions and dimensional

attributes). Finally, Malinowski & Zimányi (2007;

2009) do not propose any CASE tool adhering to

their meta-model, like the SDWM proposal.

Figure 8: Highway SDW according to SDWM.

6 COMPARISON WITH (GLORIO

& TRUJILLO, 2008)

Glorio & Trujillo (2008; 2009) extend the UML

meta-model in order to define a UML profile

enriched with a set of stereotypes for dimensions,

facts, conventional measures, spatial measures,

degenerated dimensions, conventional levels and

spatial levels. Figure 9 (Glorio & Trujillo, 2009)

shows the obtained SDW-aware UML profile.

Moreover, based on the proposed UML profile, they

also build a CASE tool whose components adhere to

their meta-model. Figure 10 (Glorio & Trujillo,

2009) shows the logical models of a sale SDW

according to the proposal due to Glorio & Trujillo

(2008; 2009). As shown in Figure 10, the SDW

schema mixes DW modeling concepts with OLAP

data cube ones and it does not provide support for

the following DW modeling techniques: role-playing

dimensions and bridge tables. Also, it is not possible

to define whether the geometry of spatial attributes

can be normalized and/or shared. As a consequence,

similarly to the other two meta-model proposals due

(Fidalgo et al., 2004; Times et al., 2009; da Silva et

al., 2010) (Section 4) and Malinowski & Zimányi

(2007; 2009) (Section 5), the meta-model by Glorio

& Trujillo (2008; 2009) does not allow a full

modeling of the homicide SDW of the running

example (see Section 3).

Figure 9: SDW-aware UML profile by (Glorio & Trujillo,

2008).

Looking again to Figure 10, we observe that the

sale SDW schema modeled according to the

methodology by Glorio & Trujillo (2008; 2009)

exposes one fact table, called Sales, which is

characterized by conventional measures only (e.g.,

cost and total) and three dimensions: Store, Product

and Client, equipped with levels having

conventional and spatial attributes (e.g., category

and geometry_polygon). This means, again, mixing

concepts from different contexts (i.e., DW and

OLAP). Also, similarly to the previous meta-models

investigated in our analysis, the solution by Glorio &

Trujillo (2008; 2009) does not provide a concise and

compact notation. In fact, again one class (it should

be reminded that in this case UML extensions are

considered) is introduced for each dimensional level.

Figure 11 shows the sale SDW of Figure 10

modeled by means of SDWCASE according to

SDWM. Again, the two schemas are equivalent.

Similarly to the case of the highway SDW (see

Section 5), in the logical model of the sale SDW

according to SDWCASE of Figure 11, we simply

define all geometries as not normalized and not

shared, hence again obtaining redundancy, high data

volumes and difficult SDW administration. Just like

the previous case, we observe that, contrary to this,

the SDWM meta-model offers a more concise and

compact solution (five dimensions in the SDWM’s

case – see Figure 11 – vs eight dimensions in the

ComparativeAnalysisofState-of-the-ArtSpatialDataWarehouseMeta-models-CatchingtheExpressivePowerofSDW

Schemas

307

Glorio & Trujillo's (2008; 2009) case – see Figure

10).

Figure 10: Sale SDW according to (Glorio & Trujillo,

(2008).

Summarizing, from the results of this analysis we

can conclude that, with respect to the meta-model

proposed by Glorio & Trujillo (2008; 2009), the

proposed meta-model SDWM exposes a clearly-

higher expressive power, along with more concise

and compact schemas, according to similar

consideration given in Section 4 and Section 5.

Figure 11: Sale SDW according to SDWM.

7 COMPARATIVE ANALYSIS

RESULTS

In this Section, we provide the results of the

comparative analysis on the proposed meta-model

SDWM against the related ones we discussed in the

previous Sections.

Table 1 summarizes the results of our analysis.

As it follows from Table 1, all comparison

approaches allow us to design spatial measures.

However, no proposal except the SDWM one

addresses spatial attributes. As a consequence,

comparison approaches do not support determining

whether the geometry of a spatial attribute should be

normalized and/or shared among different

dimensions. As highlighted in previous Sections, it

should be recalled here that normalizing and sharing

spatial attributes has the beneficial effect of reducing

the whole data volume of SDW and making the

SDW administration simpler.

Moreover, among comparison approaches, only

the proposal by (Fidalgo et al., 2004; Times et al.,

2009; da Silva et al., 2010) (i) disassociates DW

modeling concepts from OLAP data cubes modeling

concepts, and (ii) addresses the degenerated

dimension modeling technique. Unfortunately, for

what regards bridge tables and role-playing

dimensions, no comparison approach addresses

these yet-useful constructs.

Also, contrary to the SDWM approach, where

we make use of suitable spatial attributes to

represent spatial information, (Fidalgo et al., 2004;

Times et al., 2009; da Silva et al., 2010) model

spatial information as dimensions, Malinowski &

Zimányi (2007; 2009) as ER entities, and Glorio &

Trujillo (2008; 2009) as UML classes.

Finally, as highlighted throughout the paper,

comparison approaches are clearly not capable of

providing concise and compact SDW schemas like

the SDWM approach.

Table 1: Results of the comparative analysis among SDW

meta-modeling approaches.

(Fidalgo

et al.,

2004)

(Malinowski

& Zimányi,

2007)

(Glorio &

Trujillo,

2008)

SDWM

DW vs OLAP

Modeling

YES NO NO YES

CASE

Tool

YES NO YES YES

Degenerated

Dimensions

YES NO YES YES

M-N

Relationships

(Bridge Tables)

NO NO NO YES

Role-Playing

Dimensions

NO NO NO YES

Spatial

Attributes

NO NO NO YES

Spatial

Measures

YES YES YES YES

Concise

Notation

NO NO NO YES

Normalized

Geometry.

NO NO NO YES

Shared

Geometry.

NO NO NO YES

Yes (%) 40 10 30 100

NO (%) 60 90 70 0

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

308

8 CONCLUSIONS AND FUTURE

WORK

In this paper, we have provided a comparative

analysis of the SDWM approach (Del Aguila et al.,

2011; Cuzzocrea & Fidalgo, 2012a; Cuzzocrea &

Fidalgo, 2012b) against the state-of-the-art SDW

meta-model proposals (Fidalgo et al., 2004;

Malinowski & Zimányi, 2007; Glorio & Trujillo,

2008). Results of our analysis clearly state that the

SDWM proposal exposes a higher expressive power

and allows us to obtain more concise and compact

SDW schemas.

Future work is oriented towards enriching

SDWM with novel aspects such as security and

privacy of SDW, in line with recent results in the

context of security and privacy of DW and OLAP

(e.g., (Cuzzocrea & Bertino, 2011; Cuzzocrea et al.,

2012; Cuzzocrea & Saccà, 2012)).

REFERENCES

Bédard, Y., Merrett, T., & Han, J. 2001. Fundamentals of

spatial data warehousing for geographic knowledge

discovery. Geographic data mining and knowledge

discovery, vol. 2, pp. 53-73.

Del Aguila, P. S. R., Fidalgo, R. N. & Mota, A., 2011.

Towards a more straightforward and more expressive

metamodel for SDW modeling. In Proceedings of the

ACM 14th international workshop on Data

Warehousing and OLAP. New York, NY, USA: ACM,

pp. 31-36.

Cuzzocrea, A., & Bertino, E., 2011. Privacy Preserving

OLAP over Distributed XML Data: A Theoretically-

Sound Secure-Multiparty-Computation Approach.

Journal of Computer and System Sciences 77(6), pp.

965-987.

Cuzzocrea, A., Bertino, E., & Saccà D., 2012. Towards a

theory for privacy preserving distributed OLAP. In:

EDBT/ICDT Workshops 2012, pp. 221-226.

Cuzzocrea, A. & Fidalgo, R. N., 2012a. Enhancing

Coverage and Expressive Power of Spatial Data

Warehousing Modeling: The SDWM Approach. In A.

Cuzzocrea & U. Dayal, eds. Data Warehousing and

Knowledge Discovery. Springer Berlin / Heidelberg,

pp. 15-29.

Cuzzocrea, A. & Fidalgo, R. N., 2012b. SDWM: An

Enhanced Spatial Data Warehouse Metamodel. In M.

Kirikova & J. Stirna, eds. CAiSE Forum. CEUR-

WS.org, pp. 32-39.

Cuzzocrea, A., & Saccà, D., 2012. A Theoretically-Sound

Accuracy/Privacy-Constrained Framework for

Computing Privacy Preserving Data Cubes in OLAP

Environments. In: OTM Conferences 2012, pp. 527-

548.

Fidalgo, R. N. et al., 2004. GeoDWFrame: A Framework

for Guiding the Design of Geographical Dimensional

Schemas. In Y. Kambayashi, M. Mohania, & W. Wöß,

eds. Data Warehousing and Knowledge Discovery.

Springer Berlin / Heidelberg, pp. 26-37.

Glorio, O. & Trujillo, J., 2008. An MDA Approach for the

Development of Spatial Data Warehouses. In I.-Y.

Song, J. Eder, & T. Nguyen, eds. Data Warehousing

and Knowledge Discovery. Springer Berlin /

Heidelberg, pp. 23-32.

Glorio, O. & Trujillo, J., 2009. Designing Data

Warehouses for Geographic OLAP Querying by Using

MDA. In O. Gervasi et al., eds. Computational

Science and Its Applications – ICCSA 2009. Springer

Berlin / Heidelberg, pp. 505-519.

Gray, J. Chaudhuri, S., Bosworth, A., Layman, A.,

Reichart, D., Venkatrao, M., Pellow, F. & Pirahesh H.,

1997. Data Cube: A Relational Aggregation Operator

Generalizing Group-by, Cross-Tab, and Sub Totals.

Data Mining and Knowledge Discovery 1(1), pp. 29-

53.

Malinowski, E. & Zimányi, E., 2009. Advanced Data

Warehouse Design: From Conventional to Spatial and

Temporal Applications (Data-Centric Systems and

Applications), Springer.

Malinowski, E. & Zimányi, E., 2007. Logical

Representation of a Conceptual Model for Spatial Data

Warehouses.

GeoInformatica, 11(4), pp.431-457.

da Silva, J. et al., 2010. Modelling and querying

geographical data warehouses. Information Systems,

35(5), pp.592-614.

Times, Valéria Cesário et al., 2009. A Metamodel for the

Specification of Geographical Data Warehouses. In S.

Kozielski et al., eds. New Trends in Data Warehousing

and Data Analysis. Springer US, pp. 1-22.

Zghal, H. B., Faïz, S. & Ghézala, H. B., 2003. CASME : A

CASE Tool for Spatial Data Marts Design and

Generation. In Proceedings of Design and

Management of Data Warehouses. Berlin, Germany,

pp. 1-11.

ComparativeAnalysisofState-of-the-ArtSpatialDataWarehouseMeta-models-CatchingtheExpressivePowerofSDW

Schemas

309