Nagano and Moraes

7/28/2019 Nagano and Moraes

1/12

African Journal of Business Management Vol. 7(4), pp. 273-284, 28 January, 2013Available online at http://www.academicjournals.org/AJBMDOI: 10.5897/AJBM11.2824ISSN 1993-8233 2013 Academic Journals

Full Length Research PaperAccounting information systems: An intelligent agents

approach

Marcelo Seido Nagano* and Marcelo Botelho da Costa Moraes

Department of Production Engineering, School of Engineering, So Carlos, University of So PauloAv. Trabalhador Socarlense, 400, So Carlos SP, 13566-590, Brazil.

Accepted 19 October, 2011

The purpose of this paper is to present new approaches in the development of accounting informationsystems to enable better data management and information creation. The objective is achieved byapplying an object-oriented modeling with the use of intelligent agents, according to the needs of usersof accounting information. In development of this work, it was observed that a structure based onobjects and using intelligent agents enables the development of reports for different users, with gainsin the quality of information developed. The major limitation of this work is that it was done on atheoretical basis; however, the practical aspect is yet to be carried out due to the extent of itsdevelopment. The great advantage of working is to use an object-oriented modeling with simultaneousapplication of intelligent agents, who carry on the development and analysis of accounting. Thus, theaccounting information system is able to meet fully the qualities needed to users, without loss ofcomprehensibility, relevance, reliability and comparability, even with changes in business model or inaccounting standards used. Furthermore, the development of new intelligent agents enables aretrospect on previous year's analysis.

Key words: Accounting information systems, DCA (debit-credit accounting) model, resource, events andagents (REA) model; intelligent agents, object-oriented.

INTRODUCTION

Accounting is a science which can be defined as aninformation and evaluation system. The aim is to provideto its users demonstrations and analyses of an economic,financial, physical and productivity nature in relation tothe entity.

Accounting information is quantitative and qualitativewhich meets the needs of internal and external users.

According to the Financial Accounting Standards Board(FASB) in the Statement of Financial AccountingConcepts (SFAC) No. 1, 1978, the information fromfinancial reporting is subject to limitations. Moreover,according to the SFAC No. 2, 1980, information should

*Corresponding author. E-mail: [email protected],[email protected]. Tel: +55 16 3373 9428. Fax: +55 16 33739425.

be comprehensible to those who have a reasonableamount of knowledge about business and economics(FASB, 1980), but it does not indicate to which level ofdepth this knowledge is.

These characteristics and needs intrinsic to accountingshould be observed in its information system. Thereforethese results in the importance of modeling an account-

ing information system which can meet all the ways andvisions needed to make decisions.

The most classic way of registering economic transact-tions was made formal by the friar Luca Pacioli, whoshowed merchants at that time how to registercommercial transactions, emphasizing the doubling oeach transaction in relation to cost vs. benefit, known asthe double entry system (Fisher, 1997).

Another way of retaining information is the REA mode(economic Resources, economic Events, economic

Agents), based on the modeling of databases. This


2/12

274 Afr. J. Bus. Manage.

approach is used in integrated system environmentswhereby each economic transaction is associated to(Entity-Relationship) a series of economic resources andeconomic agents (Fisher, 1997).

There are various modeling techniques in the literatureon information systems (e.g. entity-relationship diagramsand data flow charts), but the REA model is preferred dueto a specific technique in accounting information systems(Rom and Rohde, 2007).

Aims

The aim of this work is to propose a way of modelingaccounting information systems oriented towards theobjectives using support from intelligent agent tools tohelp take decisions according to various possibilities of

formatting among different types of users and decisions.The following specific aims are presented:

1. Present the accounting information system models;2. Check how they can be made suitable regarding theneeds of different types of information;3. Develop oriented modeling to objects using the supportof intelligent agents, which are flexible in terms of dealingwith information;4. Compare the proposed system with the current modelaccording to the qualitative characteristics of accountinginformation.

Motivation

Studying ways of information modeling has become arelevant subject in accounting as it includes the way thatdata and information are registered. It is worth men-tioning that data is a source which does not add value asthe information consists of the data in the structuring andrelationship used in decision making process.

Therefore, the work in this article can help improveaccounting information systems so that they can provideassistance to a wide range of information users who havedifferent levels of knowledge and distinct needsconcerning accounting.

Taking this into account, the way in which databasessave information, provide access, search for data andproceed in formalization are fundamental for the user touse them efficiently. Therefore, the REA model can behighlighted in this sense.

However, one of the characteristics of the REA modelis its limitation at a transactional level. Most of theliterature on the subject ignores the supply of informationto take decisions (Rom and Rohde, 2007), thus the needto use intelligent agents to overcome this limitation.

Artificial intelligence has been applied successfully in

structured, programmable and repetitive tasks wherebythe human knowledge obtained is not extremely difficul(Baldwin et al., 2006). Considering this, the use ospecialist systems is justified by problem structuring andthe possibility of obtaining current knowledge fromaccountants and auditors.

Research problem

Considering the aspects mentioned earlier, as well as theimportance of accounting information systems and theconstant need of improvement, this work describes andanalyses the following question: How can accountinginformation systems be developed in order to performbetter considering the users needs based on moderninformation technology?

METHODOLOGY

This work is of a theoretical nature and is strongly based onbibliographical research. It presents a methodology to developinformation systems which use object-oriented modeling togethewith intelligent agent applications to deal with information.

Therefore, from the existing theories of accounting informationsystems, as well as technological tools of information and intelligensystems, this work uses deductive methods to present how thesetheories can be applied by developing accounting informationsystems which are more suitable for organizational needs.

ACCOUNTING INFORMATION SYSTEMS

Progress in information technology and the increase inthe use of the Internet requires administrators, accountants, auditors and academics to become cleverer andmore knowledgeable in terms of design, operations andaccounting information systems control (Beard and Wen2007) having direct implications in the security andreliability of the system.

Considering this, it has become more important todefine how economic events can be classified:

1. Transactions, where something with measurable valueis passed voluntarily from one part to another, osimultaneously between both parts; and2. Inter-actions, whereby there is a measurable effect onthe value in the entity without there being the participationof another part (Birkett, 1968).

The transactions normally refer to the negotiationprocess, while the inter-actions are related to the processof aggregating value. In both cases, measuring isrelevant in terms of registering information regardless othe model and the way they are used. The measurementmethodology establishes criteria checked for accounting


3/12

meeting the needs of consistency and comparability. Inspite of this, there are some difficulties in devisingmeasurement techniques and the current systems canreflect incorrect metrics without having the capacity toalter them later (Chambers, 1998). Measuring cannot beconfused with quantification. The former derives from anestimate technique of evaluation, usually in monetaryterms, while the latter, quantification, is related tophysical counting, and is easily checked.

Therefore, it is important to observe the reduction ofinformation throughout the registration process inaccounting information systems. This synthesizing pro-cess (in terms of the economic transaction) is used tomake understanding easier for users. Thus, there is aneed to summarize information, grouping it in a smallnumber of account balances to increase the utility of thestatement in its understanding (Babich, 1975).Currently, accounting information systems can be found

isolated or inserted into ERP systems (EnterpriseResource Planning). In dedicated systems, as well asintegrated systems (ERP), performance is usually basedon databases. To understand the existing accountinginformation systems better, the DCA and REA modelsare described in detail as follows.

The DCA model

The double entry method consists of the most classicway of accounting. Part of the presupposition of eacheconomic event must be registered by debit and credit,

hence the name double entry. This happens due to theduality of each event having a resource origin and anapplication on the same date and for the same amount ofmoney.

To better understand the DCA model (debit-creditaccounting), the way in which accounting deals withthese origins and resource applications should beobserved. Each event, whether it be a transaction (nego-tiation) or inter-action (production) happens from anorigin, that is, a financial loss in terms of reducing assetsor generating liabilities.

After information technology was introduced andcomputers, as well as data processing systems became

more widespread, information systems of largecorporations began to be developed within availablecomputational language to use on a large scale.

The current ERP systems work with related databases,where each event automatically generates debits andcredits according to its configuration. This has made theaccounting process easier. Moreover, standardizedreports show the total amount of the accounts balancesanalytically or synthetically, producing the main financialstatements.

However, accounting information systems based on the

Nagano and Moraes 275

DCA model are not very malleable as any change in thestructure of a company that needs alterations in its charof accounts would need a total restructuring of configure-tions and system parameters. This would cause problemsof losing uniformity and consistency of accountinginformation hindering its comparability and quality oinformation which enables users to identify similaritiesand differences between two economic phenomena(Hendriksen and Van Breda, 1999).

This makes it impossible for users to compare theresults of the organization before and after changingaccounting decisions as these results reflect differenmeasurement methodologies. Therefore, changes inaccounting end up eliminating historical comparison datawhich help to check the efficiency and efficacy of theprocesses.

The REA model

The REA accounting model is based on the principle oeach economical event being in an entity carried out byinternal and external economical agents bringing abouchanges in economic resources. This is why it is calledREA: to relate aspects of economic Resources, economic Events and economic Agents.

Developed in the 1980s, this model came from applyingthe events theory proposed by Sorter, according to whomaccounting must be oriented to register the event whichtook place, and not just the financial amounts involved(Sorter, 1969).

Therefore, as well as the usual information of dates andamounts involved in the event, data that can describe theevent in detail and make its future forecasting possible isaggregated. The systems user can make his own modeonly decides which information is significant or not, giventhe function loss when only monetary values areobserved (Sorter, 1969).

The REA model can reduce the problems inherent tothe traditional model, which limits the monetary measurement, without multidimensional information and oftenclassified inappropriately, storing too much aggregatedinformation without integrating other areas of thecompany (McCarthy, 1982).

The model defines the economic resources as thepatrimony of the entity. The economic agents are responsible for patrimonial changes (Figure 1).

The economic events are transactions by negotiating(buying and selling) economic resources, or inter-actionsin aggregating value to the resources, carried out byinternal economic agents or together with external agentsin the entity (Figure 2).

As each event involves an origin and an application oresources (the main idea of double entry) there is adouble event in this model. This relationship is called


4/12


Hierarchy generalization

ECONOMICAGENT

STOCK HOLDER EMPLOYEE CUSTOMER VENDOR

CLERK SALES-PERSON PROFESSIONAL

Figure 1. Example of generalization (McCarthy, 1982).

Economic Resource

Stock-flow

Economic Event

Duality

inside

participation

Economic Agent

External

participation

Economic Agent

Economic Event

outside

participation

Economic Agent

inside

participation

Economic Agent

Economic Resource

Stock-flow

Give

Take

Figure 2. The standard REA (McCarthy, 2003).


5/12


MAPPING TO STORAGE

INTERNAL SCHEMA

CUSTOMER

SALE

INVENTORY

PURCHASE

FINISHED GOODS

WORK IN PROCESS

RAW MATERIALS

...

...

....

...

...

CONCEPTUAL SCHEMA

OPEN-ORDERS

No. NAME#

CUSTORMER-SERVICE

GENERAL-LEDGER

SALES CASH

XX XX

A/R

XX

WIP

XX

...

RAW- MATERIALS

No.

INVENTORY-MGT

ISSUE RECEIPT

...

EXTERNAL

SCHEMATA

Figure 3. Schema specification for databases (McCarthy, 1982).

duality and binds economic events because every incre-ment economic event must be related by a duality to adecrement economic event (Kasik and Hunka, 2011).

This shows two types of relationship of each entity. Atthis point, the term entity is interpreted according todatabase theories referring to events, resources andagents: The first is the association, which determineshow the different entities interact among themselves; thesecond is the generalization, in which each entity is ageneralization of various different factors, but having thesame characteristics (McCarthy, 1982).

Where the database program (language) refers tointernal schema, the conceptual schema which the REA

model is applied can be determined. Exiting the systemcan be presented in various ways using formatted reportsby the external schema (Figure 3).

Currently, the ERP systems which met the needs ofcompany resources attend to the REA model as eachprocess cycle needs to be modeled in the system bydatabases based on entity-relationship (McCarthy, 2003).

In practice, the ERP systems do not use this techniquecompletely. The aim of accounting in systems is usuallyto record debits and credits. This would not be necessarywhen the REA model is used.

When comparing the REA and SAP model, one of themain ERP systems on the market, significant similaritiesare found, although there are situations whereby the REAmodel is less detailed (OLeary, 2004).

Moreover, companies which implant ERP systemscannot use the whole capacity of the capturing systemprocessing and delivering financial and non-financiainformation to the decision makers in time (Wier et al.2007).

REA based system can capture a comprehensive set obusiness information more efficiently than traditionabusiness reporting systems, allowing organizations tomake more efficient and better business decisions, but

the REA model needs a technological language to helptake advantage of its benefits (Amrhein et al., 2009).

INTELLIGENT AGENTS

Intelligent agents are a field of artificial intelligence basedon computing, highlighting perception, reasoning andaction. Reasoning is particularly essential for higherintelligence to function (Wachsmuth, 2000).

The agent does some activity for another, known as the


6/12


Recognizeinput

Internal Representation

Apply MethodChange

Representation

Select

Method

MethodStore

GeneralKnowledge

Affects theWorld

Figure 4. Functional diagram of general intelligent agent (Wachsmuth, 2000).

main one, having autonomy when developing its func-tions. The idea of intelligent agents was introduced in themid-1950s, and they have been defined in many differentways (Liu, 2011).

Due to the inexistence of a universally accepteddefinition, concepts of weak-notion and strong-notionof the agency are used. The weak notion of the agencyconsists of the vision of an agent corresponding to anelement ofhardware and software based on the compu-tational system having characteristics of (1) autonomy,operating without human intervention; (2) social ability,interacting with other agents (human or not); (3)reactivity, noticing its environment and responding; and(4) pro-activity where initiatives are taken (Wooldridgeand Jennings, 1995).

The strong notion of the agency consists of an agentbeing a computational system having the characteristicsdescribed previously, as well as being developed and

implemented using concepts that are usually applied tohumans (Wooldridge and Jennings, 1995), as thereasoning, belief, intention and others.

The intelligent agent notices the input of the environ-ment (Figure 4) and acts to change the environment, butbefore it uses an internal representation to observe thepossible effects of alternative methods. These methodsare observed from an internal method store and itsexploration is guided by general world knowledge alsointernally (Wachsmuth, 2000).

Observing the characteristics given by the notions from

the agency, two aspects are extreme important indetermining an intelligent agent (Figure 5).

1. Agency/autonomy: The agency determines the level oautonomy invested in the agent being able to beevaluated from an asynchronous application, passing tothe representation of the user, to the interactivity withother agents in handling data, applications and services;2. Intelligence/capacity of reasoning: Intelligence is thelevel of learning and reasoning related to the agentsskills to incorporate the determined objectives by the useand fulfill the tasks which are delegated to it. Varyingfrom instructions on preferences, normally rules, progressing to reasoning from inference models to planningand learning (Vasarhelyi et al., 2005).

Therefore, the relationship between the agents autonomy and the capacity of reasoning (intelligence) deter-

mines an intelligent agent.Among the various existing models in the area o

artificial intelligence, expert systems are used in thiswork. Despite this, in accordance with the usefulness ofthe intelligent agent, other techniques can be used.

The expert systems are computational programs whichdo tasks for human specialists, although they do notsubstitute a human being as a decision maker (Barbera1987) because they do not have autonomy characte-ristics. The expert systems grew out of the data miningstudies in the 1990s because of increasing amount of


7/12


Preference Reasoning Planning Learning

Agent Interactivity

Service Interactivity

Application Interactivity

Data Interactivity

User Representation

Asynchrony

Intelligence

Agency

Artificial Intelligence

Fi

xedFunctionA

ents

Intelligent Agent Threshold

Figure 5. Intelligent agent scope (Adapted from Vasarhelyi et al., 2005).

data stored in relatively easily accessible databases(Kilpatrick, 2011).

Furthermore, the expert systems differ as far as thecomputational programs are concerned in terms offunction and structure. Regarding function, by doingactivities normally designated to humans and in itsstructure in the way in which these functions are carriedoutartificial intelligence tools and methodologies (Brownand Phillips, 1995).

Expert knowledge about the process can be expressedin linguistic rules by words of spoken or even artificiallanguage (Pamucar et al., 2011). To use intelligentagents in accounting, expert systems were selected asan artificial intelligence tool because the application inaccounting information systems is well-structured, having

specialists and documented knowledge on accountingoperations to obtain rules.

By using rules such as If., Then., thesesystems are developed with the help of specialists whoby themselves find ways of condensing their experienceto simple rules (Foltin and Smith, 1994). This system ofrules follows logic of how a human being takes decisions,called heuristics. In case the task is structured andadvanced calculations are not needed and heuristics areused, the system specialist will be ideal in helping withthe decision making (Shim and Rice, 1988).

OBJECT-ORIENTED INFORMATION SYSTEMS USINGINTELLLIGENT AGENTS

The presented model is based on applying the technology of intelligent agents to accounting informationsystems attempting to improve the way the systemsmake information available to the users, providingvariations of format and giving information in detail.

Even though, the REA model is the most accepted interms of structuring the accounting phenomena; it isefficient in storing data, but not in handling information(Verdaasdonk, 2003). Taking this into account, developing a system to deal with data is needed, which providesrelevant, reliable and comparable information. These areall characteristics that determine the usefulness of

accounting information (FASB, 1980).Developing this model is done by language databases

oriented to objects. Using the oriented modeling toobjects was introduced by Knaus (2001) in accountinginformation systems developing. However, his proposauses the DCA model as a base and each account is adistinct class (Knaus, 2001).

Modeling databases oriented to objects came from thelimitations of the relational databases. Their basicobjective is to manage large amounts of informationmaking handling data easier and meeting the curren


8/12


External agents:customer/supplier

Internal Agents

Inter-action

Transaction

Resource Allocation

Intelligent agents

Developing reports and

analyses

External agents: informationusers

Figure 6. Use case diagram general model for REA applying intelligent agents.

requirements of the systems with other types of data-bases, such as hypertexts, which are in constant use(Silberschatz et al., 1999).

Within this characteristic, the UML language (UnifiedModeling Language) can be applied, mainly as it is anappropriate tool for modeling intelligent agents (Heinze,2004) and is based on object-oriented databases. Thisapproach differs from the traditional accountinginformation systems because it separates data frominformation modeling operations the same time it uses anapproach oriented to objects.

Aiming to have a better performance in the proposedmodeling, the UML developing is used to make visualize-tion easier and integrate among the different objectives.

In traditional systems, reports are developed indatabase itself while in the REA model object-oriented(REAOO), the reports and their analyses are developedusing intelligent agents based on expert systems. Thebiggest difference is separating application from account-ing databases, which does not normally happen.

Therefore, in terms of this development, the use ofUML language to model the system is of great help. It is

important to mention that the model is a simplification ofreality presenting a general case of an informationsystem for any activity that needs to be done. Taking thisinto account, while Entity-Relationship modeling in theREA model has 3 entities (Resource, Events and Agentsto develop the modeling Object-Oriented, theseresources, events and agents are the classes to beimplemented.

Therefore, by using intelligent agents to developinformation for different types of users, the externascheme (according to the REA model) can be developed

by intelligent agents, where this agent is a new actor inthis diagram (Figure 6).

While the external agents can take part in transactionevents as clients and suppliers, the internal agents act inthe transaction and they are responsible for interactionand resource allocation; furthermore, inserting intelligenagents in the development and report analyses oinformation.

Normally, there are many similar objects in a databasethat is, they respond to the same messages, use thesame methods and have variables of the same name and


9/12


ECONOMIC

RESOURCES

CASH AND

EQIVALENTS

INVENTORY RECEIVABLES FIXED ASSESTS

FINISHED PRODUCTS WORK IN PROCESS RAW MATERIAL

Class: Economic Resources

Figure 7. Hierarchy of classes.

Economic ResourceEconomic EventEconomic agent

1..*1..* internal

1..*0..*

1..*1 cost

1..*1 benefit

duality

Client

Name

Credit Limit

Employee

Name

Function

Figure 8. REA object-oriented model.

type. Considering this , grouping these similar objects isdone by these classes (Silberschatz et al., 1999).

Taking this into account, the entities used in the REAmodel can be considered as classes. In object-orientedmodeling for databases, a class can have a specializationhierarchy, that is, the ISA relationship, which shows aclass as being specialized from another (Silberschatz et

al., 1999).Therefore, as in the generalization of the REA model,

specialization of Resource, Event and Agent classes canbe determined. An example is shown in Figure 7.

Determining the classes and their specializations, thegeneral model of the class diagram can be made in thepresented proposal of the REA model (Figure 8).

In the REA object-oriented model (REAOO), an econo-mic agent, which can be internal or external, carries outone or more economic events and the economic eventitself brings about the duality in relation to the resources,

providing consumption (cost) and generation (benefit) ofone or more resources simultaneously.

In the general case of commercializing, a producmanufactured in the entity itself, a client (externaeconomic agent) can acquire one or more products with aseller (internal economic agent), bringing about aneconomic event (sale) which will have an impact on the

economic resources, whether it be from the cost of one omore units or the benefit of a cash receipt or accountreceivable.

This model, similar to the original format of the REAmodel, can be applied to any situation within the varioustypes of organizations that carry out economic activities.

An intelligent agent is autonomous, that is, the intelligenagent is a software program separate from the databases, which accesses it remotely and makes searchesobtaining data which is needed to develop and/or analyzethe information.


10/12


DATABASE STORAGE

...

..

.

.

.

.

.

.

REAOO MODEL

INTELLIGENT AGENT SYSTEM

.

.

.

Client

(agent)

duality

Employee

(agent)

Sales

(event)

Inventory

(resource)

Acc. Receivable

(resource)

.

.

.

INTELLIGENTAGENTFinancial

Accounting

IASB

DATA STORAGE SYSTEM

INTERACTIVITY

INTELLIGENT

AGENT

Financial Accounting

USGAAP

INTELLIGENTAGENTBalance

Analysis

REPORTS AND ANALYSES

Figure 9. General model - REA object-oriented with intelligent agents.

Considering this, the accounting information systemconsists of two separate environments. The first one is anenvironment to store data, where the REA object-orientedmodel acts. The second environment consists of anintelligent agent system which accesses the database bysoftware interactivity and carries out its interpretationfunction (Figure 9).

As an example of the expert system technique, thedifferent ways of accounting a determined can becompared.

Observing the between international accounting stand-ards from the IASB (International Accounting StandardsBoard) and United States Generally Accepted AccountingPrinciples (USGAAP) by FASB, the accountant canincreasingly make inferences about the economic aspectof the events to determine the most suitable way ofaccounting.

Generating knowledge from the expert system whichwill make the inference within each intelligent agent(Figure 9) must be made by obtaining knowledge fromthe specialist, that is, an accountant and/or auditor who

can establish the knowledge in a formal way essentiallyusing rules and heuristics to take decisions.

Taking this into account, in accordance with the need othe information systems user, the intelligent agensystem will develop the financial statements according tothe desired standards, only altering its knowledge basethat is, the rules of the expert system.

After developing traditional accounting reports (BalanceSheet and Income Statement, for example), the intelligenagent can proceed to analyze traditional financia

indicators using an expert system developed for this. Thisagent will base its analysis on the other intelligent agentsresponses by the interactivity of agents.

Therefore, these intelligent agents can continuouslymonitor the situation and financial/operational performance of the entity in terms of returning recommendationsto the internal users of the system, helping or even takingdecisions.

It is worth mentioning that the preference to use expersystems as an artificial intelligent tool of intelligent agentsis due to the easiness of structuring knowledge from


11/12

human accounting specialists (accountants), but otherways of artificial intelligence could be used. An examplewould be to use neural networks to forecast results.

ANALYSIS OF THE MODEL

Based on the needs of the stakeholders, the proposedaccounting information system evaluation can be madefrom four qualities of accounting information determinedby FASB.

Comprehensibility

According to this analysis, the proposed system hasadvantages regarding the current systems because theREAOO model with intelligent agents provides the

development of intelligent agents which can developmore specific reports to different types of users accordingto their capacity of understanding and level of knowledgeabout accounting. Greater information is necessary to ahigher level of detail in making a specific decision (Cohenand Kaimenaki, 2011).

Relevance

The REAOO and intelligent agents are based onproviding relevant information to take decisions at theright time, that is, when necessary. However, the predict-tive value and feedback can be highlighted as important.

While the DCA and REA models essentially focus onstoring past information, the REAOO model can useintelligent agents to develop forecasts and future projects,adding the predictive value that current systems cannot.

Another factor is the value as feedback. In this case, itis worth mentioning that previous expectations arefulfilled in case they happen, or corrected otherwise.Essentially, this quality is applied in forecast and realizedcomparison. Once again, the current systems cannotmake these applications, but the REAOO model candevelop specific intelligent agents to develop and control,even for budgeting and cash flow development andanalysis.

Reliability

Reliability of information is closely linked to the definitionsof the rules of business. In the traditional systems,reliability is developed by the storing data system. In theREAOO model, reliability should at least be observed inthe intelligent agent.

Therefore, applying intelligent agents in financialaccounting eliminates the need of having computationalroutines to convert accounting standards. In the REAOO


model, accounting register according to differenaccounting standards is direct, without the need foconversion, avoiding problems of reliability in this processand making it easier to check auditory processes.

Comparability

This quality shows the capacity of comparing accountinginformation over time in the same organization orbetween different organizations at the same moment theREAOO model can be highlighted.

From the time in which the data analysis is dis-associated from its storage, the comparison onlydepends on developing a new intelligent agent for therequired objective. Therefore, from the time when thecompany starts using the REAOO model, it will be storingdata in the model. If the company needs to change the

way of accounting or even the way of analyzing someyears afterwards, they just need to change theresponsible intelligent agent and the past data will beavailable from the original REAOO system.

In the case where this same organization decides toimplant a budget system or even develop its accountingaccording to accounting standard from another countrythey just need to develop specific intelligent agents forthis work and the organization will have past informationand analyses since the REAOO system was implanted.

By doing this, possibilities of comparing become limitedonly from start storing the data in the REAOO system, theanalysis provided by the intelligent agents can be repliedto original data, without bias or adjustments. This is acharacteristic of the proposed system which makes ipossible not to lose past comparison parameters in caseof changes in accounting standards or calculus metho-dologies. In other words, theoretically it will never loseconsistency.

Normally, after having changes in standards, or evennew analysis needs (using new costing methodologiesfor example), entities end up being adopted to makeadjustments in previous periods (restricted to fewperiods) for comparison.

In the proposed model, REAOO, any alteration in thestandards implies that making a specific intelligent agentis suitable. Thus, when it is created, it can recalculate the

past, without having to be adjusted prone to errors.

Conclusion

As shown, accounting information systems are fundamental for taking decisions in entities. In spite of thismost of the modeling and development of the systems donot observe the users needs or a way of accountingwork.

Alles et al. (2008) discussed accounting separating ibetween research in accounting information systems


12/12


(AIS) and research which is not focused on accountinginformation systems. They show the need to aggregatevalue in AIS research as this area of research showsdifficulties in methodologies when presenting the resultsas they depend on application and rarely possiblepractical tests.

Therefore, it is common to see problems between theneeds shown by the accountants and the existingspecifications in the systems which make suitable repliesimpossible, as shown by Sayed (2006).

Analyzing the progress of information technology andits application in accounting information systems, it canbe observed that there were practically no improvementsin the way of working. The DCA model, which is still thestandard model, did not progress after the computer age.

Currently, most systems simply automatically replicatedouble entry in accounting general ledger. It was onlywhen the REA model came about that accounting began

to see computing as a way of improving processes. How-ever, these initiatives are still specifics, perhaps becauseaccountants do not know about the existing technologyand developers unknown the accounting needs.

The concept of new models, as in this work, hasbecome pertinent, but at the same time complex asmultidisciplinary knowledge is needed in distinct areas.There are also difficulties in applying it. Because of this,the present work has as the major limitation a nonpractical development of the model REAOO. Moreover,the definition of intelligent agent based on expert systemrequires a more complex modeling, identifying practicalproblems observed in companies.

Despite all this, by creating conditions to apply anobject-oriented model using intelligent agent support, thiswork makes it possible to study the development of newapplications of intelligent agents in more perspectivesareas such as costs, forecasting, budgets and simula-tions among others in practically all the objectives in-volved in accounting and using other artificial intelligencetechniques.

So, there are future perspectives as the practicalapplication of the model, with a further development ofartificial intelligence techniques in reporting and decisionmaking is associated.

REFERENCES

Alles MG, Kogan A, Vasarhelyi MA (2008). Exploiting comparativeadvantage: a paradigm for value added research in accountinginformation systems Int. J. Account. Inform. Syst. 9(4):202-215.

Amrhein DG, Farewell S, Pinsker, R (2009). REA and XBRL GL:Synergies for the 21

stCentury Business Reporting System Int. J. Dig.

Account. Res. 9:127-152.

Babich G (1975). The application of information theory to accountingreports: an appraisal ABACUS 11(2):172-181.

Baldwin AA, Brown CE, Trinkle BS (2006). Opportunities for artificial

intelligence development in accounting domain: the case for auditingIntell. Syst. Account. Finan. Manag. 14:77-86.

Barbera AT (1987). Artificial intelligence in accounting: the future has

arrived Rev. Bus. 9(2):17-21.

Beard D, Wen J (2007). Reducing the threat levels for accountinginformation systems CPA J. 77:34-42.

Birkett WP (1968). Accounting inputs ABACUS 4(2):164-173.

Brown CE, Phillips ME (1995). Expert systems for managemenaccounting tasks Montvale: The IMA Foundation for Applied

Research Inc. p.204.

Chambers RJ (1998). Wanted: foundations of accounting measuremenABACUS 34(1): 36-47.

Cohen S, Kaimenaki E (2011). Cost accounting systems structure and

information quality properties: an empirical analysis J. Appl. AccountRes. 12(1):5-25.

Financial Accounting Standards Board (1978). SFAC/01: objectives o

financial reporting by business enterprises Norwalk: FASB. p.8.Financial Accounting Standards Board (1980). SFAC/02: qualitative

characteristics of accounting information Norwalk: FASB. p.60.

Fisher SA (1997). In defense of double entry accounting Nat. PubAccount. 42(3): 33-34.

Foltin LC, Smith LM (1994). Accounting expert systems CPA J

64(11):46-53.Heinze C (2004). Modeling intention recognition for intelligent agen

systems Edinburgh: DSTO Systems Sciences Laboratory p.249.Hendriksen ES, Van Breda MF (1999). Accounting theory So Paulo

Atlas p.550.Kasik J, Hunka F (2011). Business process modelling using REA

Ontology. Econ. Manag. 16:1047-1063.

Kilpatrick J (2011). Expert systems and mass appraisal J. Prop. InvestFinan. 29(4/5):529-550.

Knaus M (2001). Object-oriented accounting a Framework for a

modern accounting information systems Zurich: PhD Thesis inEconomic Sciences, University of Zurich p.409.

Liu G (2011). The application of intelligent agents in libraries: A survey

Program 45(1):78-97.McCarthy WE (1982). The REA accounting model: a generalized

framework for accounting systems in a shared data environmen

Account. Rev. 57(3): 554-578.McCarthy WE (2003). The REA modeling approach to teaching

accounting information systems Issue Account. Educ. 18(4):427-441.

Oleary DE (2004). On the relationship between REA and SAP Int. JAccount. Inform. Syst. 5:65-81.

Pamucar D, Bozanic D, Dorovic B, Milic A (2011). Modelling of the fuzzylogical system for offering support in making decisions within theengineering units of the Serbian Army. Int. J. Phys. Sci. 6(3): 592609.

Rom A, Rohdes C (2007). Management accounting and integratedinformation systems: a literature review Int. J. Account. Inform. Syst8:40-68.

Sayed HE (2006). ERPs and accountant's expertise: the construction ofrelevance J. Enter. Inform. Manag. 19(1):83-96.

Shim JK, Rice JS (1988). Expert systems applications to manageria

accounting J. Syst. Manag. 39(6):6-13.Silberschatz A, Korth HF, Sudarshan S (1999). Database systems So

Paulo: Makron Books., p. 808.

Sorter GH (1969). An "events" approach to basic accounting theoryAccout. Rev. 44(1):12-19.

Vasarhelyi MA, Bonson E, Hoitash R (2005). Artificial intelligence in

accounting and auditing: international perspectives - Vol. 6 Princeton

Markus Wiener Publishers p.257.Verdaasdonk P (2003). An object-oriented model for ex ante accounting

information. J. Inf.Syst. 17(1):43-61.Wachsmuth I (2000). The concept of intelligence in AI In: Cruse H

Dean J and Ritter H, eds. Prerational IntelligenceAdaptive Behavio

and Intelligent Systems without Symbols and Logic (Vol. 1)Dordrecht: Kluwer Academic Publishers, pp.43-55.

Wier B, Hunton J, Hassabelnaby HR (2007). Enterprise resource

planning systems and non-financial performance incentives: the joinimpact on corporate performance Int. J. Account. Inform. Syst. 8:165190.

Wooldridge M, Jennings NR (1995). Intelligent agents: theory andpractice Knowl. Eng. Rev. 10(2):115-152.

Nagano and Moraes

Documents

Transcript of Nagano and Moraes