-The present paper describes the approach and two technical solutions for interoperation between business rules represented in various formats. The Semantic Web techniques are used to enable this interoperation. One of the interoperation methods uses the Java Interoperation Object (JIO) described in the context of Positional-Slotted Language (POSL), which a human-friendly variant of the Rule Markup Language (RuleML), and Notation 3 (N3) representations. Details of the connections between these document representations are demonstrated with the use of query-based interoperation between POSL and N3. Another solution described in the present paper is conversion of business rules stored in Microsoft Excel as decision tables into POSL using OpenL tablets. Although the current business rules interoperation framework involves three formats (Excel, POSL, and N3), it can be extended to other document representations through appropriate conversions of data in rule bases and queries.
Keywords— Knowledge Representation, XML, RDF, Notation 3, Positional-Slotted Language, Rule Markup Language, Semantic Web, Query
I.
INTRODUCTION
Business rules are becoming ubiquitous in modern industry
and are usually created, stored, and maintained by business
analysts, knowledge engineers and software engineers in
various formats. Some formats are technical, and some formats
are non-technical and more user-friendly. Classically, business
rules are logic constructs (e.g. “IF-THEN” type), and they are
often represented using decision tables or decision trees.
Technically, business rules can also be implemented using a
programming language like C or Cobol, or by the use of a
controlled English. There are many specific solutions for
creation and maintenance of business rules. For instance,
Microsoft Excel tables can be deployed as a user-friendly way
to build documents representing decision tables. Systems like
Drools provide excellent platforms to build and maintain more
complex business rules. There are some standards for business
rules representations. The most known is the Semantics of
Business Vocabulary and Business Rules (SBVR) [
II.
The resented approach to interoperation is based on semantic interoperability using a mediator, which can convert business rules among various knowledge representations. The software mediator can process and interpret business rules stored in various formats, as well as convert a query formulated in any of these formats to search an answer in all rule bases connected. This will assist clerks, brokers, managers, and other specialists in finding better business solutions and decision-making.
Since business rules become Web-based, the modern solutions for interoperation can be deployed. The solutions presented in this work use the Sematic Web infrastructure and related tools. The Semantic Web offers solutions allowing to semantically enriching business rules using a background ontology, which serves as a knowledge base (or vocabulary). On the other hand, the disadvantage of creating and maintaining business rules in a Semantic Web language is that rules are difficult for human understanding. Even POSL, which is more human-oriented than RuleML, is difficult for a non-specialist to understand. The focus of the present work is query-based interoperation between two Semantic Web based languages: POSL and N3, and conversion of business rules in MS Excel format into POSL. The focused interoperation framework is presented in Fig. 1.
The framework presented in Fig. 1 consists of the following main components: 1. POSL rule base, which consists of business rules and facts in the POSL format. rule base, which consists of rules and facts in the N3 format. 2. MS Excel database, which contains rules and facts in the user-friendly format.
Excel Rule Base
POSL Query
N3 Query Excel-POSL Converter
Parsers
JIO Reverse Parsers
OOjDREW
Ontology Euler Conversion Rules
N3 Answer
POSL Answer
N3 Rule Base
3. JIO (Java Interoperation Object), which The Java
Interoperation Object is the basis of interoperation
methods developed for the Knowledgebase Representation
Interoperation Tool (KRLIT) described technically in [
Euler for N3. 6. Parsers – to syntactically analyze business rules in POSL and N3, and convert them into the JIO format. 7. Reverse parses – to convert business rules from JIO format into a format required by the user.
In the present framework, the user can query the knowledge bases using POSL or N3 formats, and receive the answer in the same format as query.
The technical details of JIO (Java Interoperation Object)
architecture and the use of JIO within the KRLIT are
explained in [
POSL provides object-centered instance descriptions via
binary properties, taxonomies over classes and properties,
class-forming operations and class/property axioms, and
derivation, integrity, transformation, and reaction rules [
Notation 3 is a compact, rule-extended version of RDF's
XML syntax [
In order to deal with any input and fetch the answers from the available Knowledge, the system should re-present this input in order to convert it to intermediate JIO representation (RuleML building blocks) which from-and-to the system can be interoperated to the target language. The interoperation process using the JIO representation can be done by implementing Parser and ReverseParser.
The main goal of Parser is to take a query or answer of a
query of a language from a File, URL or as a String in the form
of InputObjectCollection, and then parses it (breaks down) to
RuleML building blocks, which will compose a single
AtomCollection as JIO representation in order to provide it to
ReverseParser as input. Similarly, ReverseParser takes this
JIO AtomCollection as input and reverse parses it (translates) to
the target language as output with option of returning the result
as an answer or query. Details of parser’s implementation are
presented in [
IV.
INTEROPERATION BETWEEN POSL AND N3 BUSINESS
RULES
The present work describes POSL-N3 interoperation using an example of an insurance company Farm Insurance and two on-line insurance brokers, which are insurance companies Mainland Insurance and Healthy Life. The companies use different knowledge representation languages, but use the same schema for their facts and rules. Assume the Farm Insurance has a business rules set describing automobile insurance Age-Class discounts as follows:
Age From
Age To
Customer Class 16 21 26 16 21 26 16 21 26 20 25 30 20 25 30 20 25 30
Economic Economic Economic
Gold Gold
Gold Platinum Platinum Platinum
Discount
Value 0.0 0.1 0.2 0.2 0.3 0.4 0.5 0.6 0.7 This rule base is not accessible by individual users because it is not Web-based, but can be accessed by insurance brokers through some internal communications.
Mainland Insurance focuses primarily on Economicclass customers, and prefers to use an N3 knowledge base as follows (policy for providing a discount of 10% to an Economic customer who is between the age of 21 and 25): { ?Client :type :client; :clientID ?ClientID; :age ?Age; :name [:type :fullname; :first ?FName; :last ?LName]; :class ?b. ?Age math:notLessThan 21 . ?Age math:notGreaterThan 25 . ?b log:equalTo :Economic. } => {?resultDiscount :type :Discount; :company
:MainlandInsurance; :clientID ?ClientID; :age ?Age; :class ?b; :discount 0.1.}.
Healthy Life focuses on Gold and Platinum-class customers, and prefers to use a POSL knowledge base as follows (policy for providing a discount of 20% to a Gold customer who is between the age of 16 and 20): Discount(company->HealthyLife; clientID->?ClientID; age-> ?a:Integer; class-> ?b;discount-> 0.2:Real) :client(clientID->?ClientID;age->?a:Integer; name->fullname[ first->?FName; last->?LName]; class-> ?b), greaterThanOrEqual(?a, 16 : Integer), lessThanOrEqual(?a, 20 : Integer),
equal(?b, Gold).
Suppose the customer is familiar with POSL only, but wants to find discount policies of both insurance brokers. The query is: Discount(company->?All; clientID->?clientID; age->?age;
class-> ?b;discount->?discount).
If business rules interoperation is not enabled, the only Healthy Life database is accessible. During query processing time, this query is transformed by POSL parser into JIO, and the N3 reverse parser class accepts the transformed query as input. This provides the N3 representation of this POSL query as follows: ?subject :type :Discount; :company:MainlandInsurance; :clientID ?ClientID; :age ?Age; :class ?Class; This query in POSL is given to OOjDREW, whose answers are returned in POSL. Since now the equivalent N3 query is available, it can be given to Euler as input, whose answers are given in N3. The answer is used by N3 parser and stored in JIO. It is then sent to the POSL reverse parser to generate the POSL representation of the N3 answers. This answer is combined with the OOjDREW answer resulting in the following combined POSL answer: Discount(company->MainlandInsurance;
clientID->1:Real;age->19:Real;class->Economic;discount->0.0:Real).
Discount(company->MainlandInsurance;
clientID->6:Real;age->17:Real;class->Economic;discount->0.0:Real).
Discount(company->MainlandInsurance;
clientID->2:Real;age->22:Real;class->Economic;discount->0.1:Real).
Discount(company->HealthyLife;
clientID->3:Real;age->19:Real;class->Gold;discount->0.2:Real).
Discount(company->HealthyLife; clientID->5:Real;age->30:Real;class->Gold;discount->0.4:Real).
The answer is consistent with the business rules maintained by the Farm Insurance.
INTEROPERATION BETWEEN MS EXCEL AND POSL
USINESS RULES
Assume Healthy Life would like to update its rule base
automatically using data from Excel sheets created by Farm
Insurance. The business rules below need to be converted
from the user-friendly Excel format into POSL:
Using rule transformation templates, the table was
automatically converted to POSL syntax, parsed, and loaded
into the OO jDREW reasoning engine [
The query asks “what is the discount value for a customer with age 25 and type Gold?” The results of query, issued using the OO jDREW Top-Down reasoning engine, are as follows: ?discount = 0.3 of type Real.
The solution presented allows automatically updating the rule base in POSL using rules created in Excel. A similar solution can be developed for N3. This allows business analysts to work with user friendly formats rather than to use heavily human readable Semantic Web languages.
CONCLUSION AND FUTURE WORK
The present paper has described the business rules interoperation framework as a solution to the Web-based interoperation gap issue. The work has focused on interoperation between business rules created in two different Semantic Web languages. The usage examples have been presented. The second focus of the paper is a methodology to partially automate the process of converting human-readable business rules stored in the form of MS Excel tables to machine-processible POSL, with the goal of combining the ease of use of Excel-based rule tables with the semanticallyrich queries supported by reasoning engines. Although the work in current state covers Excel, POSL and N3 formats only, it can extend to other business rules representations.
ACKNOWLEDGMENT
The present work includes business rules interoperation techniques which had been implemented as part of students’ projects conducted at the University of New Brunswick under supervision of the author. The students were Girish R Ranganathan (Dr. Ranganathan now), J Anthony Brown, Taylor Osmun, and Patrick Thébeau. The main funding sources for this work are NSERC Discovery Grant and ACOA AIF.