Designing Uniform Database Representations for Cloud Data

Interchange Services

Alina Andreica

Babes-Bolyai University, Cluj-Napoca, Romania

Keywords: Equivalence and Simplification Algorithms, Data Interchange, Cloud Services, Database Representation,

Pattern Matching, Software Design.

Abstract: The paper proposes design principles for data representation in cloud data interchange services among

various information systems. We apply equivalence algorithms and canonical representation in order to

ensure the uniform representation in the cloud database. The solution we describe, proposed to be provided

within cloud architectures, brings important advantages in organizational communication and cooperation,

with important societal benefits. The generic design principles we apply bring important advantages in the

design of the cloud interchange services.

1 INTRODUCTION AND

WORKING FRAMEWORK

Within the framework of software design principles,

based on systematic techniques, abstract patterns and

adequate tools for problem solving, we propose

means of using simplification and equivalence

algorithms for modelling data representation. We

apply these techniques in designing data interchange

service among various information systems within

cloud environments.

Equivalence algorithms can be implemented in an

abstract manner, based on category theory (Andreica

et al, 2012). Applying generic techniques is useful

both for design reasons and for solving specific

problems based on mathematical models (Andreica et

al, 2012).

Interoperability is the capability of different

systems to share functionalities or data (Olmedilla et

al, 2006). System interoperability has been dealt with

by means of various models (Morris et al, 2004) and

has been extensively researched for business

processes (Ziemann, 2010).

In (Andreica et al, 2015) we overview

interoperability layers, principles and tools. Ones of

the most relevant are: The Electronic Data

Interchange (EDI) model (Adams et al, 2002), the

XML standard (XML standard, 2015), RosettaNet

(Rosetta, 2015), ebXML (OASIS, 2015) standards.

Knowledge discovery, inference, logic are enabled

by semantic interoperability.

Knowledge sharing over computer information

systems, a major task for ensuring interoperability, is

based on the Conceptual Knowledge Processing

paradigm (Stumme and Wille, 2000).The Open

Internet of Things standards (OpenIoT, 2015) may

also be used as an efficient framework for data

interchange.

Within section 2 we address means of

implementing simplification and equivalence

algorithms on various entities, including hierarchical

structures. In section 3 we propose techniques for

solving specific pattern matching problems using

equivalence algorithms. Section 4 addresses

principles of data interchange between information

systems using a cloud database that retains data in a

canonical representation. We propose structures for

the entities and attributes to be used in the local and

cloud databases. Conclusions reveal the most

important topics presented in the paper and future

research and development directions.

2 EQUIVALENCE ALGORITHMS

Within this section we present means of

implementing equivalence algorithms (Buchberger

and Loos, 1982) on various entities, including

hierarchical structures. We use the implementation

framework that we have introduced in (Andreica et

al, 2012).

An equivalence relation ‘~’ verifies reflexivity,

554

Andreica, A.

Designing Uniform Database Representations for Cloud Data Interchange Services.

DOI: 10.5220/0006354205820587

In Proceedings of the 7th International Conference on Cloud Computing and Services Science (CLOSER 2017), pages 554-559

ISBN: 978-989-758-243-1

symmetry, transitivity properties (Buchberger and

Loos, 1982).

Since the data volume that has to be processed is

usually large, the most efficient way of organizing it

is using a relational database, in which entities are

retained in dedicated tables. Database structuring

principles for processing equivalent entities are

introduced in (Andreica et al, 2012).

We also use tables for retaining the entities on

which equivalence algorithms are applied and we

process entities belonging to the dedicated tables that

retain those entities based on the following principle:

IsSpecificEntity(d) :=









otherwisefalse

EntityTbldtrue

]id_entity[_,

Hierarchical data structures are often necessary to

be processed in a database; such structures may be

retained in relational databases by means of

ascendant / successor pointers in dedicated tables

(Andreica et al, 2010). Principles for retaining and

processing hierarchical structures and a comparison

of their processing techniques are presented in

(Andreica et al, 2010).

In (Andreica et al, 2012) and (Andreica et al,

2010) we present means of processing hierarchical

structures at database level, using a dedicated table

for retaining the corresponding entities and a 4

pointers representation technique: ascendant,

descendant, predecessor (same level), successor

(same level).

In (Andreica et al, 2012) we discuss the case

study of equivalent disciplines and in (Andreica,

2016) – the example of an expert system for plant

therapy

In (Andreica et al, 2010) we detail the above

mentioned principles for processing modules of

didactic activities. The implementation uses stored

procedures parameterized with the level value. The

system uses a MS SQL database, the hierarchical

structures which model curricula information being

processed by means of stored procedures – see

(Andreica et al, 2010) for details.

Postorder type n-ary tree evaluation algorithms

using the above described tree representation are

implemented in order to parse the hierarchy of

entities.

Some efficiency studies we have performed on

processing hierarchical structures at database level

are presented in (Andreica et al, 2010).

In the hierarchical entity structure, leaf entities

are retained in dedicated tables, based on the

principle stated below:

IsLeafEntity(m) :=

, : [d]

true d m IsSpecificEntity

false otherwise









For example, in (Andreica et al, 2012) we process

hierarchies of modules in which all non-leaf modules

consist only of modules.

Let d1, d2 be two entities. We use the notation

‘~’ for describing the equivalence of the two entities

d1 ~ d2; this relation may have various significances

in various case studies – see (Andreica et al, 2010),

(Andreica et al, 2012). In each case, we have to

check whether the relation is an equivalence one

since by verifying if it complies reflexivity,

symmetry, transitivity properties.

The canonical representative of an entity

equivalence class is important since it will be further

used in pattern matching rules – see section 3.

We implemented the simplification algorithm for

determining the canonical representative for a given

entity class (Andreica et al, 2015). Based on this

algorithm, we may also test the equivalence of two

entities by verifying they have the same canonical

representative.

By generically denoting with ‘’ an equivalence

relation for categories of entities, we may state that:

e1  e2  (d1  e1, ! d2  e2 : d1 ~ d2 ) 

(d2  e2, ! d1  e1 : d1 ~ d2 )

For a leaf category of entities e, we consider

Canonic (e) = {Canonic(d) | de} – the set of

canonical representative for the contained entities. It

can be shown that two leaf equivalent entities have

the same sets of canonical representatives.

For a category of entities we can recursively

compute its canonical representative set as:

Canonic(e) =











otherwisee

etyIsLeafEntie

},ed | d){Canonic(e

][},d | ){Canonic(d

Intuitively, the canonical set for a category of

entities is obtained by “flattening” its category sub-

tree and computing the union set of all canonical sets

corresponding to its descendant leaf entities.

Generically, we may state:

Canonic (e) = {Canonic(d) | de}

3 PATTERN MATCHING

PRINCIPLES

Within this section we refer to mappings between

two elements as correspondences between the two

elements that may be occur in various cases, usually

named as pattern matching.

We implement pattern matching rules for

equivalent entities by reducing the mapping between

Designing Uniform Database Representations for Cloud Data Interchange Services

555

two elements, belonging to the two equivalence

classes that are to be mapped, to mapping their

canonical representatives, as described below:

Let e

 E class of equivalent entities, em

 EM

class of equivalent mapped entities, e

– the

canonical representative of class E and em

– the

canonical representative of class EM. Then we

reduce a mapping of two entities e

, em

belonging

respectively to the equivalence classes E, EM to the

mapping between the two canonical representatives

 E, em

 EM – see Figure 1:

-> em

 e

-> em

, where e

 E, em

 EM

Figure 1: Pattern Matching Scheme for Equivalence

Classes.

We may as well use the equivalent mappings:

-> em

where e

 E,

(any element of E may be mapped into the

canonical element of EM)

-> em

, where em

 EM

(there is a mapping between the canonical

element of E and any element of EM)

For the case of equivalence classes with

hierarchical representations – see Figure 2 – the

canonical representatives are the roots of the

corresponding trees (Figure 2). Parsing algorithms

for finding the canonical representatives generally

use the ascendant pointer – see section 2.

We can use the above described rules in

managing pattern matching problems on equivalence

classes that occur in the design of expert systems.

We note that for the database representation it is

important to improve table access speed table by

indexing the tables in respect with the search id; this

principle is very useful to be applied as well in

managing hierarchical representations at database

level, which are frequently processed in order to find

the canonical representative.

Figure 2: Pattern Matching on Hierarchical Structures of

Equivalence Classes.

4 PRINCIPLES FOR BUILDING

THE CLOUD UNIFORM

DATABASE

REPRESENTATION

The data interchange architecture (Andreica et al,

2015) provides data exchange services between

various information systems or entities using cloud

services and a cloud database for mapping and

handling the exchanged information – see Figure 3.

Data exchange may be performed both in XML

relational database formats; for XML format, the

corresponding database representation (Andreica et

al, 2012) is generated into the cloud database. The

following sequences are pursued: data to be

exchanged is marked in the source database using

dedicated tables and columns, mapped into the cloud

database, sent and retained into the cloud database –

see the structure proposed below. For the destination

system / database, which sends data requests, a data

mapping is also performed and required data is sent

from the cloud database into the destination one.

The data exchange may use multi-criteria agents

implemented in the cloud environment both for

performing necessary mappings and for handling

communication.

The Cloud Data Interchange Services will be

designed for supporting automatic data exchange in

various fields, with important communication

efficiency benefits (Andreica et al, 2015).

The design principles are based on agent system

development, the communication between agents

being designed as a multi-agent system, since the

multi-agent architecture (Weiss, 1999) provides

many advantages, such as: decentralization,

extensibility, robustness, maintainability, flexibility.

CLOSER 2017 - 7th International Conference on Cloud Computing and Services Science

556

Figure 3: Data Interchange Model Using Cloud Services.

The agent architecture includes self-adapting

communicating objects, which work on distributed

datasets, supporting both the exchange and the

analysis of distributed sources (Andreica et al,

2015).

Multi-agent systems (MAS) are appropriate for

modelling heterogeneous interactions, based on

flexible autonomous actions aiming at achieving

specific goals. Ontologies and agent technologies

may be combined in order to successfully enable

heterogeneous knowledge sharing (Andreica et al,

2015).

Figure 4: Agent Architecture Model for Data Interchange

in Cloud Frameworks.

We use the multi-agent model proposed in (Faulkner

et al, 2014) for modelling the agent interaction. The

agent architecture implements local agents for

gathering data from the communicating systems.

Within the cloud environment, a multi-criteria agent

handles information mapping between source and

destination systems, using a generic information

structure defined in the cloud database, as a

canonical representation – see Figure 4.

This agent can be viewed as a mediator type of

agent. A mediator may be defined as a system which

refines, in a specific way, information from one or

more sources (Wiederhold , 1992). A mediator

embeds the knowledge which is necessary for

processing a specific type of information and may

also convert data to a common format (Chawathe et

al , 1994). This mediator agent definition (Andreica

et al, 2015) frame is given in Definition 1.

Agent:{ Local

Interface:

Data[require(local_representation)]

…

Effector[provide(exchanged_items)]

KnowledgeBase:

Source_Exchanged_Data

LocalDB_into_CloudDB

…

Capabilities:

Handle_Exchange_Request

… }

Definition 1: Mediator definition frame

These mediators model the interface between the

destination systems and the cloud environment,

having queries as inputs and returning objects via

the interface layer. The proposed mediators are

primarily hosted in the cloud environment, but they

can also be used within client information systems

(Andreica et al, 2015). They send dedicated queries

in order to obtain appropriate objects from the

remote sources, via the cloud environment.

In order to enable the exchange of various data

between two information systems, using dedicated

databases, with different structures, we set the

canonical representation on the cloud database. The

mapping process is user assisted since it requires

human input.

We further propose a database mpdel for

processing the entity equivalence from tables in

various local databases using a uniform caninical

representation in the clod database.

Let {E

, E

, ..., E

} be the class of entities to be

processed from a local database:

, A

, ..., A

]

, A

, ..., A

]

....

, A

, ..., A

]

We note min=min{Card(E

), ..., Card(E

) }.

If min < n ie. i{1,...,m}: Card(E

)<n, there

exist entities E

, i{1,...,m} which do not represent

Designing Uniform Database Representations for Cloud Data Interchange Services

557

all attributes that are taken into account into the

canonical representation. In this case, we consider

the corresponding attributes from the entities E

i{1,...,m} to be Null.

If min > n, there exist entities E

, i{1,...,m}

which contain attributes that are not represented into

the canonical representation.

We further discuss the attribute equivalence from

a local database, ie. the columns contained in the

entity tables.

We note with A

the canonical representative for

the class {A

, A

, ... A

} containing attributes

from the entities {E

, E

, ..., E

}

We note with A

the canonical representative for

the class { A

, A

, ... A

} containing attributes

from the entities {E

, E

, ..., E

}

....

We note with A

the canonical representative for

the class { A

, A

, ... A

} containing attributes

from the entities {E

, E

, ..., E

}

Then the entity E

, A

, ..., A

] will be the

canonical representative for the class of entities {E

, ..., E

}

When mappings between the entities and

attributes from the various information systems

databases are performed, we create equivalence

tables in each local information system database

containing:

EntityEquiv[id_entities, LocalEntity_TableName,

CloudEntity_TableName, Obs]

AttributeEquiv[id_attributes, id_entities,

LocalAttribute, CloudAttribute, Type, Matched,

Permissions, UpdatedFrom, Date, Significance,

Obs]

where

Type – is the type of the attribute

Matched is a logical value, representing whether the

matching was performed

Permissions{R, W} representing the permissions

on the attribute, Read or Write

Significance is the significance of the Attribute

Obs = observations

UpdateList[id_attributes, UpdatedFrom, Date, Obs]

where

UpdatedFrom = the source database from which the

update was performed (via the cloud database)

Date = the date of the update

Observations:

1. Back-ups for local databases are obviously

recommended. In the cases in which local

database administrators choose, dedicated tables

or even databases (copies) may be used for the

interchange process and then synchronizations

can be managed locally.

2. In the case in which more cloud environments

are used, the id of the cloud database has also to

be retained.

The cloud database will contain the following

equivalence structure:

CloudEntitiesEquiv[id_ent, source,

LocalEntity_TableName, CloudEntity_TableName,

Significance, Obs]

CloudAttributeEquiv[id_attributes, id_entities,

LocalAttribute, CloudAttribute, Type, Significance,

Matched, Permissions, Obs]

CloudUpdateList[id_attributes, UpdatedFrom,

Date, Obs]

// the LocalAttribute with id_entities id is

updated in the corresponding local database from the

source specified in UpdatedFrom, at the specified

Date

The fields have the same significances as

described above, but are retained in the cloud

database. The Date of the update is the date when

the data is sent into a local database within the

interchange process.

Observations:

1. The cloud table CloudEntitiesEquiv contains an

entry for each entity / table from the local

databases which are involved in the interchange

process.

2. The cloud table CloudAttributeEquiv contains

an entry for each atttribute from the local

databases which are involved in the interchange

process.

5 CONCLUSION AND FUTURE

WORK

The paper proposes data representation and design

principles for performing data interchange between

various information systems databases by means of

cloud services.

Simplification and equivalence algorithms are

used in order to ensure canonical data representation

in the cloud database and data correspondence in the

data exchange process. We propose a specific

structure for the entities / tables and attributes that are

to be exchanged in the local and cloud databases.

The generic manner in which we implement

simplification and equivalence algorithms on various

CLOSER 2017 - 7th International Conference on Cloud Computing and Services Science

558

entities, including hierarchical ones, represented at

database level, ensures generality and applicability in

various cases. Pattern matching rules and canonical

representatives are used in the cloud database.

We reveal the advantages of applying the

algebraic equivalence algorithm and of applying

canonical representatives’ properties in solving

pattern matching problems and designing data

interchange services.

The data interchange model we present provides

important practical advantages for increasing

organizational competitiveness, with a significant

societal impact on institutional and entities’

cooperation, efficient information access and

management for various stakeholders. A relevant

advantage of the solution is its flexibility and

efficiency in information exchange (only relevant

data is exchanged), with minimal resources involved

and significant security benefits.

Future work is related to further development and

implementation of the above described techniques.

REFERENCES

Andreica, A, Stuparu, D, and Miu, C. (2012). Applying

Mathematical Models in Software Design, 2012 IEEE

8th International Conference on Intelligent Computer

Communication and Processing, Cluj-Napoca,

Romania, Proceedings of ICCP 2012, IEEE, Ed: Ioan

Alfred Letia, p.87-90

Olmedilla, D., Saito, N., and Simon, B. eds. (2006).

“Educational Technology & Society”, Special Issue on

Interoperability of Educational Systems, vol 9

Andreica, A, Stuparu, D, and Miu, C., (2010). “Design

Techniques in Processing Hierarchical Structures at

Database Level”, Proceedings of Iadis Information

Systems 2010, Porto, 18-20 March 2010 , IADIS

Press, Ed: M Nunes, P Isaias, P Powell, p. 483-488

Ziemann, J. (2010). “Architecture of Interoperable

Information Systems - An Enterprise Model-Based for

Describing and Enacting Collaborative Business

Processes”, Logos Verlag, Berlin

Andreica, A, Covaci, F. and Küng, J. (2015). A Generic

Model for Cloud Data Interchange, Proceedings of

14h RoEduNet International Conference - IEEE,

Craiova, 24-26 September 2015, IEEE Computer

Society, p. 138-142

Adams, S, Hardas, D, Iossein, A. and Kaiman,. C. (2002)

BizTalk Unleashed. Indianapolis, Indiana: Sams

Publishing. p. 966

XML standard, http://www.w3.org/TR/xml11/#charsets ,

retreived Nov 2015

Rosetta – Rosettanet Overview: Clusters, Segments, and

PIPs (ver 02.13.00), 2011, retrieved December 2015

http://www.rosettanet.org/TheStandards/RosettaNetSt

andards/PIPOverview/tabid/3482/Default.aspx

OASIS – OASIS ebXML Messaging Services Version 3.0:

Part 1, Core Features, 2007. http://docs.oasis-

open.org/ebxml-msg/ebms/v3.0/core/ebms_core-3.0-

spec.pdf, retreived Dec 2015

Stumme, G., Wille, R. (2000). „Begriffliche

Wissensverarbeitung / Conceptual Knowledge

Processing“, Springer Verlag,

OpenIoT - Open Internet of Things architecture

https://github.com/OpenIotOrg/openiot/wiki/OpenIoT-

Architecture , retreived Dec 2015

Buchberger B. and Loos, R. (1982) Algebraic

Simplification, Computing, Suppl. 4, Springer Verlag,

p.11-43

Andreica, A (2016) Applying Equivalence Algorithms in

Solving Pattern Matching Problems. Case Study for

Expert System Design, Proceedings of International

Conference on Theory and Practice in Modern

Computing- TPMC, July 1-4, 2016, Portugal

Andreica, A, Stuparu, D. and Mantu, I. (2005). “Symbolic

Modelling of Database Representations”, International

Symposium on Symbolic and Numeric Algorithms for

Scientific Computing 2005, IEEE Press, p 59-62

Morris, E., Levine, L, Meyers, C., Place, P. and Plakosh.

D. (2004) “System of Systems Interoperability (SOSI):

Final Report”, Carnegie Mellon Univ., Software

Engineering Institute, http://www.sei.cmu.edu/reports/

04tr004.pdf , accessed May 2016

Weiss, G. E, (1999). “Multiagent Systems: A Modern

Approach to Distributed Artificial Intelligence”, MIT,

Faulkner, S, Kolp, M, Nguyen, Tai, Coyette, A, Do, T.

(2014). “Information Integration Architecture

Development: A Multi-Agent Approach”, retr Nov

2016

Wiederhold, G, (1992). “Mediators in the Architecture of

Future Information Systems”, IEEE Computer, 25:38-

Chawathe, S, Garcia-Molina, H, Hammer, J,.Ireland, K,

Papakonstatinou, Y, Ullman, J, Widom,. J, (1994).

“The TSIMMIS Project: Integration of Heterogeneous

Information Sources”, , retreived Jan 2015

http://ilpubs.stanford.edu:8090/66/1/1994-32.pdf

Designing Uniform Database Representations for Cloud Data Interchange Services

559