Microsoft Word - issue3.docx


 

150 

Intelligent Representation of Accounting Knowledge 
 

Bogdan Patrut 

Faculty of Computer Science, "Alexandru Ioan Cuza" University of Iași, Romania  
Bulevardul Carol I, 11, Iași 700506 

Phone: +40 234 542411 
bogdan@edusoft.ro 

 
Simona Elena Varlan 

"Alexandru Ioan Cuza" University of Iași, Romania  
Bulevardul Carol I, 11, Iași 700506 

Phone: +40 234 542411 
varlan_simona@yahoo.com 

 
Abstract 
In this paper we will describe some methods for representing different kind of accounting 

and financial knowledge. These methods are integrated in an original architecture modeling the 
complete accounting analysis, starting by describing an economic operation in natural language, and 
finishing by writing the financial reports. 

 
1. Introduction 
Representing accounting knowledge within an intelligent system implies, first of all, the 

establishing of the competences that an accountant should have. Then, we must represent the rules, 
reasoning, norms, value judgments, laws and principles that lie at the basis of accounting. At the end, 
we must represent the financial reports in an adequate model. 

Thus, hybrid intelligent systems can be conceived starting from the classic artificial 
intelligence systems such as the expert systems or intelligent agents. The final aim, hard to attain but 
not impossible, is realizing some complex programs that could assist and help the (human) 
accounting expert by taking over his/her tasks or even by replacing him/her. 

In this paper we will describe some methods for representing different kind of accounting and 
financial knowledge. These methods are integrated in an original architecture modeling the complete 
accounting analysis, starting by describing an economic operation in natural language, and finishing 
by writing the financial reports. 
 

2. The General Architecture 
The proposed architecture is presented in Figure 1. It consists of the following elements: 

 Documents – these refers to bills, invoices, contracts, and other input information 
 Economic operations – these represents all kind of economic operations that can be 

described in natural language sentences 
 Semantic interpreter – which will parse the natural language sentences and will 

automatically generate accounting entries 
 Norms – which represents all the rules, norms and financial and accounting lows, GAAP, 

IFRS 
 Accounting entries – are formal equations for representing economic operations 
 Rules – represents general accepted account functioning rules (i.e. technical rules) 
 Reports – which represents financial reports of an enterprise at the end of the financial 

period (month, semester , year) 
In the Figure we can observe different ways of representing accounting knowledge: regular 

expressions, first order predicate calculus, production rules, semantic nets, XBRL and SKOS. 



B. Patrut, S. E. Varlan - Intelligent Representation of Accounting Knowledge 
 

151 

 
Figure 1. The general architecture 

 
The bold arrows represent the flow of the process of complete accounting and financial 

operations, starting with reading or consulting the primary documents, and endind with the financial 
reports. It can be observed a special semantic interpreter, which can translate phrases describing 
economic operations into accounting entries. But the accounting entries are validated only if they 
respect the legal norms and the rules of functioning of the accounts. The regular expressions will be 
templates for accounting entries, as we will describe further. The rules of functioning of the accounts 
can be represented, as we will discuss further, by different ways: first order predicates calculus, 
production rules, semantic nets. 

The financial reports, which represents the final stage of the accounting analysis process,  will use 
the XBRL mark-up language for representing balance sheets or other kind of final reports. The SKOS 
language will be used to describe the taxonomy of the components of the financial reports. 
 

3. Semantic Analysis for Economic Operations 
This part of our paper deals with the “understanding” of the text afferent to an accounting 

entry. The procedure will parse the text, making the syntactical analysis, followed by an 
“interpretation” and “execution” of the analyzed text. Thus, we intend that by providing our system 
with the description of an accounting operation, to receive back the corresponding accounting entry.  

For example18, we would like that for phrases of the kind of those on the left to get answers 
like those on the right: 

a) a social capital worth 3000 RON is subscribed  => 456 = 1011, 3000 RON (RON = 
Romanian currency: sg. leu/pl. lei) 

b) a land worth 2000 RON is deposited under the form of  contribution in kind => 2111 = 
456, 2000 RON 

c) the social capital worth 3000 RON is entered as being paid => 1011 = 1012, 3000 RON 
 
How could we make this possible, using the PROLOG language? First of all the phrases will 

be turned from a row of characters into lists of rows of characters, knowing that in the PROLOG 
language efficient predicates for working with lists can be implemented (the predicate "create_list" 
will deal with this). Then, from the respective list the “word” representing the numerical sum will be 
extracted (using the predicate "extract_sum").  

In order to make things easier, words that are considered non-relevant for the phrases are 
taken out. Thus, neither prepositions nor other words that do not appear in a given dictionary, will 
not be taken into consideration. The predicate that simplifies a list goes through the list and analyses 
each word separately. This word is searched for in a dictionary (see the predicate "dictionary" and its 
clauses) and, if there is any, then it is replaced with its standard variant. The other variants are 
considered synonyms. In case the respective word does not exist in the dictionary, then it will no 
longer appear in the simplified list.  

                                                             
18 the numerical codes or symbols for the accounts are specific to Romanian economy; in Romania, like in France, there is a General Accounts Plan, 
in which a numerical symbol represents an account; for example, “1011” represents “subscribed capital”  



BRAIN – Broad Research in Artificial Intelligence and Neuroscience, Volume 9, Issue 3 (September, 2018) 
ISSN 2067-3957 
 

152 

After eliminating the “unimportant” words, the remaining words will be alphabetically 
ordered (by the predicate "sort" that implements a procedure of sorting through insertion). Thus, 
sentences of the type "capital is deposited" or "deposited capital" will have the same meaning. 

Finally, the predicate "translate" makes the desired translation is appealed to, thus reaching 
the required accounting entry. 

To illustrate this, let us consider the phrase "a social capital worth 3000 RON is subscribed". 
After the application of the first transformation ("make_list"), the following list is obtained 

["is","subscribed","capital", "social", "worth", "3000", "RON"]. After that, the predicate ”extract_sum” 
will get the value of 3000 (RON) which will be finally displayed next to the accounting entry. By 
applying the predicate”simplify” to the obtained list of words, we are only left with the list 
[”subscribed”,”capital”], because the inarticulate standard variant “capital” has been chosen instead of 
“the capital”. Then, the list is sorted, resulting in ["capital","subscribed"] which, according to the 
translation given by the predicate "translate" leads to the accounting entry 456=1011. 

 
4. The Representation of Accounting Knowledge  
To exemplify the accounting knowledge representation models in a knowledge-based system, 

we will consider the four account functioning rules that can be expressed in natural language as 
follows: 

1) If X is an assets account, X reflects Y and Y increases then X sells out. 
2) If X is an assets account, X reflects Y and Y decreases then X credits. 
3) If X is a liabilities account, X reflects Y and Y increases then X credits. 
4) If X is a liabilities account, X reflects Y and Y decreases then X sells out. 
 
For example, the first rule can apply to the account 512 that reflects the available in the 

current deposit on banks, in the case of payment of an invoice by a supplier, through payment order.  
The first rule represents accounting knowledge and we can notice its generality as an 

essential feature. This knowledge can be represented in a computer in more ways and the encoding 
done by the programmer should take into consideration the conceptual models of knowledge 
representation studied by Artificial Intelligence. We will present a review of some of these models 
(Sowa, Course Technology) and we will exemplify the respective representations on rule 1. 

 
a. First Order Predicates Calculus 
In first order predicates calculus, predicates represent attributes of entities or relations among 

these and can be regarded as functions with the codomain of the lots of true and false truth values. 
Besides predicates the existential quantifier  and the universal quantifier  are used, as well as the 
operators of mathematical logics: conjunction , disjunction , negation , implication  etc. 

Rule 1 above, expressed in natural language, can be represented in first order predicates 
calculus by the following formula: 

x assets_account (x)reflects(x,y)increase(y)sells_out(x) 
The names of the predicates are chosen by us, and their semantic will be encoded in the 

system according to the needs. 
 
b. Production Rules 
Similar to first order predicates calculus, production rules are the most frequently used model 

of knowledge representation in intelligent systems, especially in the data bases of expert systems. 
Within such a system, knowledge is procedural and the entire system is built around production rules 
whose structure is the following: 

<conditions>  <actions> 
This means: IF <conditions> are true, THEN the <actions> can be executed. 
Expert system generators like ExSys (http://www.exsys.com) or CLIPS 

(http://www.ghg.net/clips/CLIPS.html) are based on such production rules. 



B. Patrut, S. E. Varlan - Intelligent Representation of Accounting Knowledge 
 

153 

Researchers divide the production rules into two categories: deductive rules and inductive 
rules. Deductive rules are written under the form of the implication presented above, while the 
inductive ones under the form of a reverse implication. However, the difference between them is not 
syntactical. In the case of deductive rules we start from certain initial facts and, by applying the 
production rules we attempt to obtain new facts, finally reaching the intended aim, while in the case 
of inductive production rules we start with a purpose and, by applying the production rules 
“backwards” we attempt to demonstrate the purpose is attempted, on the basis of initial facts. Prolog 
is such an example of inductive system. 

Deductive rules 
In an expert deductive system generator the first functioning rule of accounts can be 

represented under the form 
IF [X] is asset account and [X] reflects [Y] and [Y] increases THEN [X] debits. 
Inductive - Prolog 
The Prolog language has been designed particularly for artificial intelligence problems. It is 

based on first order predicates calculus and a Prolog programme defines a series of predicates 
presenting for each of them its clauses that is, the facts and rules that render true the predicate. A rule 
is seen as an implication A1A2...AnB, conversely written. In order to represent rule 1 of 
account functioning in Prolog we will have to specify some predicates for the quality of being an 
account (assets or liabilities), for increasing (decreasing) the accountable matter that some account or 
other would represent, for the relation of reflecting the accounting matter through a certain account 
and to express the fact that an account sells out (credits). Then we will write the mentioned 
accounting rule under the form of a clause, as bellow: 

 
predicates 
 accont(integer,symbol)   increases(string)   
reflects(integer,string)    debits(integer) 
 ... 
clauses 
 ... 
 debits(X) if accont(X,a), reflects(X,Y), increases(Y). 
 
c. Semantic Networks 

 
Figure 2. The debiting rules representation of asset accounts by a semantic network 

 
Semantic networks constitute a much more abstract modality of knowledge representation, 

but a more general and efficient one also. Many researchers dealt with representing knowledge 
through semantic networks, but they are not so frequently used as the other models due to the 
difficulty of encoding them into a programme. We will not go into representation details but we can 
affirm that a semantic network is a graph whose knots contain sentence identifiers, logic connectives, 
entities etc., and the arc relations among these (class affiliation, subset, features, agent, predication 



BRAIN – Broad Research in Artificial Intelligence and Neuroscience, Volume 9, Issue 3 (September, 2018) 
ISSN 2067-3957 
 

154 

etc.). for example, the 1st functioning rule of accounts can be represented within a multi-sorted 
semantic network with the universal quantification of the variable x, as in figure 2. 
 

5. Regular Expressions as Templates for Accounting Entries 
Regular expressions represent a meta-language for describing words in a language (formal or 

natural), in a very simple and elegant formalised way. There are different variants for noting regular 
expressions, we will choose one that is sufficient for describing accounting formulas (Andone & Pătruț, 
2007). 

Regular expressions are rows of characters that represent patterns or models. They are built 
on the basis of a grammar, just like a programming language. These patterns are used to “recognise” 
and manipulate some rows of characters. Regular expressions are particularly used in text processing 
of any kind: programme writing, web pages generation. 

For some chapters in financial accounting, the accounting formulas are simple. However, 
there are some extremely complex cases for which there is a wide diversity of formulas. All 
accounting handbooks present the various chapters of financial accounting with more or less 
ambiguous explanations in Romanian, with more or less relevant practical examples. For an accuracy 
of presentation the use of a formal and clear representation mechanism is required, and we consider 
fit the mechanism of regular expressions. Also, the sums that are involved in an accounting formula 
should respect certain relations and these have to be accurately presented. 

To illustrate, will consider some accounting problems and we will present the representation 
modality of the adequate accounting recording by using regular expressions.      

    
1. The constitution of the share capital  
It is recorded the constitution of the share capital in a value s. Is also recorded the paid of 

share capital and its transformation in deposited subscribed and paid share capital. 
456 = 1011      s 
((5311|5121|5124|2xx|3xx|4xx) = 456)*        [s] 
 
2. The accountancy of the operations concerning financial assets 
For the acquisition of some shares from another entity in amount a, an entity will do the next 

accounting recording: 
((261|262|263)=( 5121|269))*     [a] 
then  
269 = 5121     [a]]   with a1 <= a. 
At the ending of the year there are recorded dividends afferent of the owned financial assets 

in amount d..    
(5121 = (7611|7612|7613|7614|7615|7616|7617))*     [d] 
There is recorded the sale of the bought shares at a price p 
461 = 7641     [p] 
There is recorded the issue of sold shares from the administration   
(6641 = (261|262|263))*     [a] 
 
3. An entity buys raw materials in amount [x] lei, the invoice including also TVA [x*0.19] 

RON. This is recorded in accounting, knowing that the entity uses the continuous inventory, like below: 
301 = 401    [x] 
4426 = 401    [x*0.19] 
 
4. An entity realizes an informatics program whose production cost is [x] RON, for circulation 

accounting of the good purchased for resale, and also it buys another one for “The wages calculus and 
accounting” whose cost of purchase is [y] RON, with TVA [y*0.19] RON. The amortization term is for 
3 years, for both the programs, so the monthly amortization is [(x+y)/36] RON. 



B. Patrut, S. E. Varlan - Intelligent Representation of Accounting Knowledge 
 

155 

a) It is recorded the realized informatics program by entity 
208 = 721      [x] 
b) It is recorded the invoice of bought informatics program.   
208 = 404      [y] 
4426 = 404      [x*0.19] 
c) It is recorded the amortization for the first month of both the programs.  
6811 = 2808 # [(x+y)/36] 
d) At the ending of the amortization time, is recorded the issue of both the programs. 
2808 = 208 # [x+y] 
667 = 4111 # [x*(1-y/100)*(1-z/100)*t/100] 
Even if our examples use numerical symbols from the Romanian General Accounts Plan, the 

bellow examples shows that the regular expressions are adecquate for writing templates for 
accounting entries. 

 
6. XBRL – A New International Standard for Representing Financial Reports 
The XBRL XBRL stands for eXtensible Business Reporting Language. It is one of a family 

of “XML” languages which is becoming a standard means of communicating information between 
businesses and on the internet. XBRL is being developed by an international non-profit consortium 
of approximately 450 major companies, organizations and government agencies. It is an open 
standard, free of license fees. It is already being put to practical use in a number of countries and 
implementations of XBRL are growing rapidly around the world (www.xbrl.org). 

In this section we will present an examples of business report created using XBRL. 
In this example we will use International Financial Reporting Standard-General Purpose 

(IFRS-GP) taxonomy adopted by the International Accounting Standards Board (IASB). The 
example expresses a balance sheet, statements about assets and equity and liabilities for two distinct 
periods for a virtual European company. The namespace prefix used here for the taxonomy is a 
recommended one: ifrs-gp. The human readable version of the financial report is the following: 

 
Table 1. Two Balance Sheets 

 
 
 
 
The XBRL document for this report is: 
 

<?xml version=”1.0” encoding=”utf-8”?> 
<xbrli:xbrl 



BRAIN – Broad Research in Artificial Intelligence and Neuroscience, Volume 9, Issue 3 (September, 2018) 
ISSN 2067-3957 
 

156 

xmlns:xbrli=”http://www.xbrl.org/2003/instance”xmlns:link=”http://www.xbrl.org/2
003/linkbase” 
    xmlns:xlink =  
“http://www.w3.org/1999/xlink” 
xmlns:xsi=”http://www.w3.org/2001/XMLSchema-instance” 
xmlns:iso4217=”http://www.xbrl.org/2003/iso4217” 
 xmlns:ifrs-gp=http://xbrl.iasb.org/int/fr/ 
ifrs/gp/2005-05-15” 
xsi:schemaLocation=”http://xbrl.iasb.org/int/fr/ifrs/gp/2005-05-15/sample ifrs-
gp-2005-05-15.xsd”> 
  <link:schemaRef xlink:type=”simple” 
xlink:href=”ifrs-gp-2005-05-15.xsd”/> 
  <xbrli:context id=”DATE_2006”> 
     <xbrli:entity> 
     <xbrli:identifier scheme= 
 “http://www.iasb.org/sample”>Samp  
 </xbrli:identifier> 
     </xbrli:entity> 
     <xbrli:period> 
     <xbrli:instant>2006-12-31</xbrli:instant> 
     </xbrli:period> 
  </xbrli:context> 
  <xbrli:context id=”DATE-2007”> 
     <xbrli:entity> 
     <xbrli:identifier scheme=“http://www.iasb.org/sample”>Samp  
 </xbrli:identifier> 
     </xbrli:entity> 
     <xbrli:period> 
     <xbrli:instant>2007-12-31</xbrli:instant> 
     </xbrli:period> 
  </xbrli:context> 
<xbrli:unit id=”U-Monetary”> 
     <xbrli:measure>iso4217:EUR</xbrli:measure> 
  </xbrli:unit> 
<!--Balance Sheet--> 
 <!-- For year 2006 --> 
  <ifrs-gp:InvestmentProperty contextRef=”DATE_2006” unitRef=”U-Monetary” 
decimals=”0”> 400,000</ifrs-gp:InvestmentProperty> 
   <ifrs-gp: Inventories contextRef=”DATE_2006” unitRef=”U-Monetary” 
decimals=”0”> 140,000 
</ifrs-gp:Inventories> 
 <ifrs-gp: ConstructionInProgress 
contextRef=”DATE_2006” unitRef=”U-Monetary” decimals=”0”> 100,000</ifrs-gp: 
ConstructionInProgress > 
<ifrs-gp: CurrentTaxReceivables 
contextRef=”DATE_2006” unitRef=”U-Monetary” decimals=”0”> 350,000</ifrs-gp: 
CurrentTaxReceivables> 
  <ifrs-gp:Prepayments contextRef=”DATE_2006” 
unitRef=”U-Monetary” decimals=”0”> 10,000</ifrs-gp: Prepayments> 
<ifrs-gp:CashAndCashEquivalents contextRef=”DATE_2006” unitRef=”U-Monetary” 
decimals=”0”> 650,000</ifrs-gp: 
CashAndCashEquivalents> 
… 
<ifrs-gp:EquityTotal contextRef=”DATE_2006” 
unitRef=”U-Monetary” decimals=”0”> 1,710,000</ifrs-gp: EquityTotal > 
<!-- For year 2007 --> 
  <ifrs-gp:InvestmentProperty contextRef=”DATE_2007” unitRef=”U-Monetary” 
decimals=”0”> 400,000</ifrs-gp:InvestmentProperty> 
 <ifrs-gp:Inventories contextRef=”DATE_2007” 
unitRef=”U-Monetary” decimals=”0”> 140,000</ifrs-gp:Inventories> 
 <ifrs-gp: ConstructionInProgress 
contextRef=”DATE_2007” unitRef=”U-Monetary” decimals=”0”> 100,000</ifrs-gp: 



B. Patrut, S. E. Varlan - Intelligent Representation of Accounting Knowledge 
 

157 

ConstructionInProgress > 
 <ifrs-gp: CurrentTaxReceivables 
contextRef=”DATE_2007” unitRef=”U-Monetary” 
decimals=”0”> 350,000</ifrs-gp: 
CurrentTaxReceivables> … 
<ifrs-gp:EquityTotal contextRef=”DATE_2007” 
unitRef=”U-Monetary” decimals=”0”> 1,710,000 
</ifrs-gp: EquityTotal > 
</xbrli:xbrl> 

 
The corresponding taxonomy for the XBRL document is an XSD (XML Schema Definition) 

file representing the XML Schema instance of the vocabulary needed to express assets, equitable and 
liabilities. 

 
<?xml version="1.0" encoding="utf-8"?> 
<schema  
   xmlns=”http://www.w3.org/2001/XMLSchema”  
   xmlns:xbrli=”http://www.xbrl.org/2003/instance” 
   xmlns:link=”http://www.xbrl.org/2003/linkbase” 
   xmlns:xlink=”http://www.w3.org/1999/xlink”  
   xmlns:ifrs-gp=”http://xbrl.iasb.org/int/fr/ifrs/gp/2005-05-15” 
   xmlns:sample =”http://xbrl.iasb.org/int/fr/ifrs/gp/2005-05-15/sample” 
   targetNamespace=”http://xbrl.iasb.org/int/fr/ifrs/gp/2005-05-15/sample”  
   elementFormDefault=”qualified” 
   attributeFormDefault=”unqualified”> 
     <annotation> <appinfo> 
 <link:linkbaseRef xlink:type='simple' 
 xlink:href='ifrs-gp-pre-bs-liquidity-2005-05-15.xml'     
 xlink:role='http://www.xbrl.org/2003/role/ 
presentationLinkbaseRef'   xlink:arcrole='http://www.w3.org/1999/xlink/ 
properties/linkbase'/> 
 <link:linkbaseRef xlink:type='simple' 
 xlink:href='ifrs-gp-pre-bs-netAssets-2005-05-15.xml'     
 xlink:role='http://www.xbrl.org/2003/role/ 
presentationLinkbaseRef'   xlink:arcrole='http://www.w3.org/1999/xlink/ 
properties/linkbase'/> 
 <link:linkbaseRef xlink:type='simple' 
 xlink:href='ifrs-gp-cal-bs-liquidity-2005-05-15.xml'     
 xlink:role='http://www.xbrl.org/2003/role/ 
calculationLinkbaseRef'    xlink:arcrole='http://www.w3.org/1999/xlink/ 
properties/linkbase'/> 
 <link:linkbaseRef xlink:type='simple' 
 xlink:href='ifrs-gp-cal-bs-netAssets-2005-05-15.xml'     
 xlink:role='http://www.xbrl.org/2003/role/ 
calculationLinkbaseRef'   xlink:arcrole='http://www.w3.org/1999/xlink/ 
properties/linkbase'/> 
   </appinfo> </annotation> 
<import namespace=”http://www.xbrl.org/2003/instance” 
schemaLocation=”http://www.xbrl.org/2003/xbrl-instance-2003-12-31.xsd” /> 
<importnamespace=”http://xbrl.iasb.org/int/fr/ifrs/gp/2005-05-15” 
schemaLocation=”ifrs-gp-2005-05-15.xsd” /> 

</schema> 
 
7. SKOS – Language for Describing the Taxonomy of the Accounts 
When we create a metadata for representing financial reports, we need to refer to shared 

group of accounts, reflecting some patrimonial elements that are common. This can be done by 
creating a selection of standardized terms with fixed meanings used to help metadata creators to 
describe a resource. In this way we can avoid ambiguity, control synonyms. This set of standard 
terms represent controlled vocabulary which can be a simple list of defined terms from which a 
metadata creator chooses a suitable one or might be a complex thesaurus made up of hierarchical 
relationships and synonyms. Controlled vocabularies are useful because they help metadata creators 



BRAIN – Broad Research in Artificial Intelligence and Neuroscience, Volume 9, Issue 3 (September, 2018) 
ISSN 2067-3957 
 

158 

and searchers to use common meanings. To create controlled vocabularies we need a simple 
language to describe concept and concept schemes. Semantic Web already offers languages like 
RDS and OWL. The Resource Description Framework (RDF) is a language for making statements 
about resources but provides only the low level semantics required for metadata statements (Klyne & 
Carroll, 2004). The OWL language by the other hand, provides the necessary semantic level to 
describe resources but it demands effort, expertise therefore costs because it is a class-oriented 
language and requires a precise logically modeling. Therefore there is a need for a language simple 
enough like RDF that does not required so much effort and expertise, bur power enough to define 
complex conceptual structures and to support semantically search like OWL does. For this purpose it 
was created a new language SKOS. 

SKOS is an extensible RDF language for describing concept and content of concept schemes 
(taxonomies, glossaries, classification schemes and thesauri) that include semantic relationships 
between these concepts. SKOS Core represents the core model for expressing the basic structure and 
content of a concept scheme (Miles & Brickley, 2005). SKOS Core Vocabulary is a set of RDF 
properties and RDFS classes that can be used to express the content and structure of a concept 
scheme as an RDF graph [4]. The SKOS Core Guide and the SKOS Core Vocabulary Specification 
are currently Working Drafts for W3C Working Group Notes. They present the basic metamodel 
consisting of an RDF/OWL schema, an explanation of the features that the properties and classes of 
the schema represent. Currently they are at the proposal stage within W3C.  

Taxonomies, or taxonomic schemes, are composed of taxonomic units or kinds of things that 
are arranged frequently in a hierarchical structure, typically related by subtype-supertype 
relationships, also called parent-child relationships.  

The example presented here is about some concepts from the balance sheet presented in 
Table 1. The properties used in this example are: skos:broader and skos:narrower used to define 
relationships of meaning between concepts. Broader is the inverse of narrower. 

 
<rdf:RDF  
  xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"  
  xmlns:skos="http://www.w3.org/2004/02/skos/core#" 
 
  <skos:Concept rdf:about="http://myTaxonomyScheme/concept/#1"> 
    <skos:prefLabel>Assets</skos:prefLabel> 
    <skos:narrower rdf:resource=" 
http://myTaxonomyScheme/concept/#2"/> 
    <skos:narrower rdf:resource=" 
http://myTaxonomyScheme/concept/#3"/> 
  </skos:Concept> 
 
  <skos:Concept rdf:about=" 
http://myTaxonomyScheme/concept/#2"> 
    <skos:prefLabel> Investment property 
</skos:prefLabel> 
    <skos:broader rdf:resource=" 
http://myTaxonomyScheme/concept/#1"/> 
    <skos:narrower rdf:resource=" 
http://myTaxonomyScheme/concept/#3"/> 
  </skos:Concept> 
 
  <skos:Concept rdf:about="  
http://myTaxonomyScheme/concept/#3"> 
    <skos:prefLabel>Inventories</skos:prefLabel> 
    <skos:broader rdf:resource=" 
 http://myTaxonomyScheme/concept/#1"/> 
    <skos:narrower rdf:resource=" 
 http://myTaxonomyScheme/concept/#4"/> 
  </skos:Concept> 
   </rdf:RDF> 



B. Patrut, S. E. Varlan - Intelligent Representation of Accounting Knowledge 
 

159 

 
8. Conclusions 
Our research is to identify a complex architecture for the complete financial and accounting 

analysis, using the new information technologies and knoledge representation model. This 
architecture will be used to develop an intelligent information system, able to realize the accounting 
analyze and to make financial reports, augmented with semantic. 

 
REFERENCES 

Andone, I., Pătruţ, B., ContTest – a Multi-agent System for accounting education, SIGEF 2007, 
Poiana Brasov XIV Congress of Fuzzy Systems in Economy and Management 

Klyne, G., Carroll, J. J., Resource Description Framework (RDF): Concepts and Abstract Syntax, 
W3C Recommendation 10 February 2004. World Wide Web Consortium, 2004. 

Miles, A., Brickley, D.,  SKOS Core Guide, W3C Working Draft, February 2005. World Wide Web 
Consortium, 2005.  

Miles, A., Brickley, D., SKOS Core Vocabulary Specification, W3C Working Draft, World Wide 
Web Consortium, 2005. 

Sowa, J., Knowledge Representation: Logical, Philosophical, and Computational Foundations, 
Course Technology; 1 edition ISBN 978-0534949655 

* * * Exsys Ltd., http://www.exsys.com 
* * * CLIPS, http://www.ghg.net/clips/CLIPS.html  
* * * www.iasb.com – the official site of the IASB organization 
* * * www.xbrl.org – the official site of the International XBRL community 

 
Bogdan Pătruţ, PhD in Accounting and Computer Science, is a senior lecturer 
at the Faculty of Computer Science, ”Alexandru Ioan Cuza” University of Iași, 
Romania. His domains of interest/research are computer science applied to 
social and political sciences and multi-agent systems applied to accounting 
education. He is also interested in social media challenges in the new academic 
environment. He published or edited more than 25 books on programming, 
algorithms, artificial intelligence, interactive education, and social media. 

 
Simona-Elena Vârlan is an Assistant Professor at "Alexandru Ioan Cuza" 
University of Iași, Romania. Education: PhD in Cybernetics and Statistics at the 
Faculty of Economics and Business Administration at “Alexandru Ioan Cuza” 
University of Iași; BSc. in Computer Science from “Alexandru Ioan Cuza” 
University of Iași; she has the ability to establish software requirements, to design, 
implement, and test using different programming languages and paradigms, data 

bases modelling, knowledge management and representation, to process and to interpret 
experimental data by statistical methods proved in 2 research & development grants where she was 
IT expert. Currently, her subjects of research are Recommender Systems, GIS programming, 
Semantic Web and Internet of Things. Publication activity: She published over 30 research papers in 
journals indexed in international data bases from which 10 are indexed WOS. The international 
visibility of the scientific work is revealed by 19 citations (GS), h index 3.