A Semantic Web Application for Matching a Citizen’s Profile to Entitled Public Services Sotirios K. Goudos, Vassilios Peristeras, Konstantinos Tarabanis Center for Research and Technology Hellas, Thessaloniki, Greece
[email protected] [email protected] [email protected] Abstract: This paper contributes to the more general problem of supplying the citizen with the correct information from the public administration in a quick and easy way. Specifically, getting the information about all the benefits that someone is entitled of is a very complex task in many countries. To address this problem, a domain ontology is first created in OWL (Web Ontology Language). The ontology is used as the knowledge base for a semantic application that uses a reasoner. This latter queries the knowledge base through a web interface by receiving the user’s profile as input , and providing the benefits that match to the specified profile as output. A list of the required documents and the procedure for receiving the various benefits from the appropriate service provider is also returned. In this way, a complex procedure is simplified and a serious overload on the client’s side (being business or citizen).is relaxed The application is highly reconfigurable in order to e.g. easily accommodate legislation changes. It accommodates law changes by ontology modifications. Moreover, the application is transferable to a large number of similar service provision cases in the public administration domain. In light of these, the advantages of semantic web technology and their application to the e-government domain are discussed. Keywords: Public Administration, Semantic Web, OWL, Matching Profiles
1. Introduction The former eEurope Action Plan has established quantitative and qualitative indicators for benchmarking the progress of e-Government in member countries. The new i2010 policy adopts this same approach as well. In establishing indicators for e-Government, the approach taken is to focus on the demand side, i.e. the everyday practice and reality of citizens' and businesses’ contacts with government. For e-Government indicators, the following maturity model is used:
• • • •
Stage 1 Information: online info about public services, Stage 2 Interaction: downloading of forms, Stage 3 Two-way interaction: processing of forms, incl. authentication, Stage 4 Transaction: case handling; decision and delivery (payment).
Information services available on-line from the public administration is the first stage in the eEurope maturity model for e-government. This stage has drawn a great amount of effort from many EU countries in order to satisfy the demand for faster information available. The implementation of an information system that will serve that demand is not always an easy task. This is due to the inherited difficulties that exist in public administration domain. In public administration there are usually many complicated services with numerous executional paths, with multiple participating organizations. In many cases in the EU countries diverged and conflicting legislation may exist for the same case something that complicates the search effort for the average citizen. As a result in many cases, it may be difficult for a citizen to find through the legislation all the correct information about the formal documents and the procedure that is required to follow for a certain task. In the general case initially the citizen has a need. He/she may not know which public services are currently available by public administration to address this need. The citizen is needs-aware, but not services-aware (being aware of the actual services he/she really needs). On the contrary, PA is services-aware, but not needs-aware. The need may arise for several reasons, e.g. due to an external event (life-event or business episode) or due to the profile of the customer (e.g. unemployed with five children).
1
The particular case that attracts our interest in this paper is that of finding the entitled benefits for a specific profile of a public servant in Greece. Legislation about entitled benefits is usually subject to changes quite often. Our motivation was to find a flexible and scalable framework to use it in an information system that supports such cases. To solve this problem, we have chosen to use semantic web technologies, and here we propose an infrastructure that could substantially simplify the overall process. This paper is organized as follows. Section 2 presents a technology overview. Section 3 describes the business case. Section 4, which is divided in the ontology and reasoner section, presents the proposed system architecture. Section 5 has a use case example. Finally section 6 contains the discussion about the advantages of such an approach and the conclusions.
2. State of the art Several information technologies exist for the creation of web-based e-government applications. Data representation in a relational model and/or object-oriented approaches are very popular and they are widely used. Creating dynamic web pages linked to an external database is a very common task. A more recent technology is the semantic web. The state-of-the art technology in a web environment is adding semantic meaning to web recourses. Currently these resources are usually only human understandable: the mark-up (HTML) only provides information for textual and graphical information intended for human consumption. Semantic Web aims for machine understandable information that can be processed and shared by both computers and humans. Tim Berners-Lee provides (Berners-Lee 2001) the definition of the Semantic Web as “an extension of the current one (Web), in which information is given well-defined meaning, better enabling computers and people to work in cooperation.” The semantic technology is not a competitor of relational model and object-oriented technologies since they apply to different types of applications for different reasons but there are some areas of overlapping. It is not an easy task to design the relational data model to represent a complex scenario. It is even more complicated to create the appropriate SQL queries to extract the correct information. In simpler cases the relational model can be used. But in complicated cases an ontology model provides more flexibility and robustness in design and implementation. The advantage of semantic technology over the relational model lies on the fact of creating machine-readable data capable of modelling complex cases. The same information can be shared not only among humans but also among clever agents in the web. Another major advantage is the fact that information using semantic technologies can be distributed anywhere in the web. Ontologies can be imported, merged with others, populated and expanded in a distributed way. This kind of scalability is perhaps the most important advantage of the semantic technology. More details about the question answering capabilities of the semantic web can be found in (McGuinness 2004). Data representation in a semantic web environment is given in four layers as shown in Figure 1 (Schwartz 2003). These layers are XML, RDF (Resource Description Framework), Ontology (OWL), and Logic. OWL is an ontology language for the Semantic Web, developed by the World Wide Web Consortium (W3C) Web Ontology Working Group (Dean 2004), (McGuinness, Harmelen 2004). In OWL, an ontology is a set of definitions of classes and properties. OWL has the ability of applying constraints on the way those classes and properties can be employed. Logic Layer Ontology Layer RDF Layer XML Layer Figure 1: The Layered architecture of the semantic web
Goudos S.K.,Peristeras V.,Tarabanis K.
There are three sublanguages of OWL; OWL Lite, OWL DL and OWL Full. OWL Lite supports those users primarily needing a classification hierarchy and simple constraints. OWL DL (Description Logic) supports those users who want the maximum expressiveness while retaining computational completeness. OWL DL is so named due to its correspondence with description logics, a field of research that has studied the logics that form the formal foundation of OWL (Dean 2004). Description Logics (DLs) is a family of knowledge representation (KR) formalisms that represent the knowledge of an application domain by first defining the relevant concepts of the domain (its terminology), and then using these concepts to specify properties of objects and individuals occurring in the domain. OWL Full extends OWL DL by adding the syntactic freedom of RDF with no computational guarantees. It is unlikely that any reasoning software will be able to support complete reasoning for every feature of OWL Full. Currently there are no reasoners available for OWL Full. Therefore OWL DL was the sublanguage selected for the creation of our ontology. Semantic technologies have been used in the literature in order to address similar problems. In (Goudos 2006) an approach based on an ontology implementation of the GEA object model (Peristeras 2006) is used. This generic approach can be used in any case of public service. In (Sacco 2006) dynamic taxonomies are used for an automated service discovery.
3. Description of the business case Following the above-mentioned line of work, we have used a quite complex informational public administration service as provided in Greek, namely the case of a citizen that needs information about all the benefits that he/she is entitled of, as a test-bed for applying semantic web technologies. This paper discusses this issue by proving a solution using this family of technologies. This case presented here can be also seen as a one-stop e-government implementation. In Greece, there is a complex legislation about the benefits that any public servant is entitled to due to his/hers education, his/hers family status, his/hers working location or even some special skills such as capability of working with PCs. For example a married public servant is entitled to a certain benefit. Furthermore, the exact amount of this benefit depends on the number of her/his children. People that have many children are entitled to an additional amount of that benefit depending on the children number. Another level of distinction is that of postgraduate studies. People that have a postgraduate degree receive a postgraduate benefit. The amount of this benefit depends again on the type the degree, higher for a PhD, lower for an MSc. Public servants who work in border areas also are entitled to certain types of benefits. The benefit here depends on the specific area (e.g. islands, mountain) according to Greek legislation. Public servants with computer skills receive an additional benefit depending on the skills they have. A large number of such benefits (more than 100) exists and legislation describes in detail the documents required to be provided by the applicant in order to obtain these benefits. An additional complexity, is that the amounts for these benefits may change quite frequently. Thus the task of quickly providing to anyone interested relevant information for the amount he/she is entitled is really challenging. Therefore anyone who wants to be adequately informed about the existence and the exact provisions of such benefits must search in various and complex legislation. The above scenario has many possible combinations. To implement that kind of scenario electronically one would have to build an application that uses a simple form that the user fills in with his/hers personal details. For example his/hers family status, his/hers work area, his/hers education level etc. Then the system should be able to return different pieces of information. First the right benefits the user is entitled to receive and then the correct documents that is required to provide to the PA agency. The information about the amount of every benefit should also be communicated to the applicant.
4. System Description Figure 2 shows the system architecture. It consists of a web server, a reasoner and an OWL file which is used as the knowledge base. The users access the application though a common Internet browser. The advantage of every web-based front end is that it requires only an Internet browser in order to execute and it can be accessed from anywhere in the Internet. The system architecture used is server-side therefore the client only shows the form and the results page.
Figure 2:The application architecture The server uses the data given to invoke the reasoner. The reasoner sends various queries to the knowledge base. The answers returned are parsed to web server that creates the results web page. The extracted results contain a list of the benefits (benefit name and amount) entitled to, plus the list of documents required in order to obtain them. More specifically the web server used was Apache Tomcat, the knowledge base was an OWL file with the ontology definition and the reasoner was JTP (Java Theorem Prover) (Fikes 2003). The system components are explained in detail in the next sections.
4.1 The ontology The business case presented in the previous section can be modelled with a set of OWL classes having properties and individuals. Individuals in OWL DL represent objects in the domain that we are interested in. Individuals can also be referred as instances of classes. In fact OWL classes are thought as sets of individuals. In OWL properties represent relations between two individuals or between an individual and a data value. Two types of properties exist, Object properties and Datatype properties. The Object properties link an individual to an individual and the datatype properties link individuals to data values such as an integer, a string or a Boolean. For example, if we assume the class Person and two properties of it: the datatype property hasAge and the object property hasParent. Individuals of this class may be John and Peter. The datatype property hasAge for John may have the value 30 and the object property hasParent may have the value John. Properties may be stated to have a unique value. If a property is a FunctionalProperty, then it has only one value for each individual (it may have no values for an individual). The property hasParent declared above is not a FunctionalProperty but the property hasFather is. Properties have a domain and a range specified. Properties link individuals from the domain class to individuals from the range class. For example, the property hasParent may be stated to have the domain of Person class. The range of a property limits the individuals that the property may have as its value. The property hasParent may be stated to have the range of Parent class. In OWL, restrictions are used to restrict the individuals that belong to a class. There are three types of restrictions; quantifier restrictions, cardinality restrictions and hasValue restrictions. The restrictions are used in class definitions. In fact a restriction is a description of an anonymous class. This anonymous class contains all of the individuals that satisfy the restriction. OWL restrictions can also be divided to necessary and necessary and sufficient. A necessary restriction means that if an individual is a member of a certain class then it is necessary to fulfil this restriction (a Primitive class in OWL terminology). A necessary and sufficient restriction (a Defined class) has the meaning that if an individual fulfils this restriction then it must be a member of this class. A reasoner can deduce the above conclusions based on these class descriptions.
Goudos S.K.,Peristeras V.,Tarabanis K.
The ontology for this case has been created using the Protégé tool with the OWL plug-in (Knublauch 2004). The class tree is given schematically in Figure 3. The ontology consists of seven major classes; Document, Benefit, Beneficiary, FamilyStatus, EducationLevel, AreaType, and Degree. More specifically:
•
Class Document is the abstract representation of the documents required. The documents required for every case are modelled as individuals of this class.
•
The Benefit class defines the concept of the benefit. The various types of benefits like family benefit or border benefit are modelled as subclasses of this class.
•
Class Beneficiary is the abstract representation of the Person (public servant) entitled to the benefits. Its subclasses are the different types of persons entitled to different benefits.
•
The Degree class represents the concept of a university degree like a Bsc or an Msc. Individuals of this class are: Msc, Bsc and Phd.
•
Class FamilyStatus is the abstract representation of the family status. Examples of individuals that belong to this class are Marrried, NotMarried and Divorced.
•
Class EducationLevel represents the four different types of education levels valid in Greece. These are modelled as four distinct individuals of the class. These are: PE (university level), TE (technical education), DE (secondary education) and YE (mandatory education).
•
AreaType class represents the distinct areas that a public servant in Greece may live in like border areas. There are three individuals of this class: CategoryA area, CategoryB area and Border area. These areas are defined in Greek legislation for certain geographical areas of Greece.
Class Benefit has four major subclasses which represent the different types of Benefits. These are:
•
Border_Benefit subclass represents the different types of benefits that a public servant in Greece is entitled to if he/she lives in a certain area. There are three types of this kind of Benefits modeled as individuals of this class.
•
Family_Benefit subclass represents the different types of benefits that a public servant in Greece is entitled to if he/she is married and has one or more children.
•
PostGraduate_Benefit subclass represents the different types of benefits that a public servant in Greece is entitled to if he/she has a postgraduate degree like a Msc or a Phd.The amounts are different for different degrees.
•
Info_Benefit subclass represents the different types of benefits that a public servant in Greece is entitled to if he/she has a degree in informatics or he/she works in the IT infrastructure of a public organization. Class Benefit has one property the datatype property amount which represents the amount in euros for each benefit. This property is defined functional with property domain the class Benefit. The definition of this property is given in Figure 4 below. The various types of public servants that are entitled to different benefits are given as subclasses of the Beneficiary Class. The four major subclasses are Person_for_FamilyBenefit, PostGraduate_Beneficiary, Person_for_BorderBenefit and Person_for_InformaticsBenefit. These classes have also other subclasses as it is shown in Figure 3. For example PostGraduate_Beneficiary class has two subclasses MasterPerson and PhdPerson representing public servants holding an Msc or a Phd degree respectively. A number of object properties are also created for the above class (Figure 5). The object properties for describing the different attributes of an Beneficiary are entitledTo, livesIn, hasDegree, hasDegreeIn, hasDocument, hasEducationLevel and ChildrenNum. The latter is a datatype property representing the children number and has single values as expected. For all the above properties domain is the Beneficiaty class, while their range corresponds to the respective self explicable class. For example property entitledTo has range the Benefit class.
Figure 3:The class hierarchy
Figure 4: The definition of amount property
Figure 5:Properties of the Beneficiary Class In a similar way object property hasDegree has range defined in the Degree class. Three object properties are declared functional which implies single values. These are hasEducationLevel, hasFamilyStatus and livesIn. It is obvious why these properties can only have a single value. The property definition for hasEducationLevel is given in Figure 6. Another object property is hasDocument, which is defined with domain the Beneficiary class and range the Document class. This property represents the documents required in order to obtain a benefit of each category.
Goudos S.K.,Peristeras V.,Tarabanis K.
Figure 6:The definition of the property hasEducatioLevel in OWL The MasterPerson class definition is given below in Figure 7. Class MasterPerson represents the person that has an Msc degree and is entitled to the Master Benefit. One may notice that this class is defined using several hasValue restrictions. The hasDegree property must have the Msc value. This restriction is declared as necessary and sufficient. Therefore if someone has a Msc degree a DL reasoner can deduce that he/she is an individual belonging to MasterPerson class. Other restrictions are declared as necessary. For example the object property entitledTo has the hasValue restriction MscBenefit. In a similar way the hasDocument property has the necessary hasValue restriction MasterCopy. The above restrictions mean that any individual that belongs to the MasterPerson class is entitled to Msc benefit and the document named MasterCopy is required. The other subclasses of the Beneficiary class are modelled in a similar way. OWL individuals may belong to more than one class.
Figure 7:The definition of the Class MasterPerson in OWL
4.2 The Reasoner The reasoner used was JTP (Java Theorem Prover). JTP supports OWL DL knowledge bases, and uses KIF format for queries and answers. When the user finally presses the submit button, the web server invokes JTP by sending KIF formatted queries and assertions. First a new individual of the Beneficiary class is created on the fly. The properties set by the user with the user interface are entered as assertions to the reasoner. The application works in a simple way using the fact that a reasoner can at first identify all the subclasses of the Beneficiary class that an individual belongs to based on the profile data. In a second step the reasoner can deduce the benefits entitled for every Beneficiary subclass found and the documents required in order to obtain these benefits. In a final step the reasoner can find the amounts for every entitled Benefit. JTP uses the OWL knowledge base and finally returns the extracted results. These include the entitled Benefits, the monthly amounts for every Benefit and the documents required in order to obtain them. That way the user gets all the information The system uses JSP (Java Server Page) and Java servlets in order to create dynamically the results in HTML format.
5. Use Case When the user enters the first page of the application a form appears and the user fills in his/hers personal details. For example, the user selects family status (married or not), number of children, education level, postgraduate degree type if any, the degree domain (as informatics, engineering etc.) and the type of area he/she lives in. When the selection process is over then the user presses the submit button and sends the selected data to the server. The results obtained are a complete list of all the benefits entitled to the user, the amount of each in euros and the list of documents needed in order to get the benefits. In case of a new benefit is added by a new law, the application can be easily expanded to accommodate the new provisions. The changes required can be made by directly editing the OWL file. Minor changes to the front end may be also required.
6. Discussion and Conclusions A semantic web application to support a common informational PA service in Greece has been presented. The proposed framework and application can be used in a wider sense to help in simplifying the searching procedure for anyone interested in obtaining benefits that he/she is entitled to. For example, disabled people usually want to be informed about the benefits thery are entitled. The recently approved by W3C web ontology language OWL has the potential to be applied for modelling complex e-Government cases. In these cases, OWL can increase performance and decrease development time due to its flexibility and scalability. Another advantage of the presented system is the ability for quick and easy updates. For example lets assume that a new law is issued that gives a new benefit to single mothers. In a traditional relational database implementation this would require a database schema modification. This type of change would probably require a major rewrite in tables and queries. An ontology modification is also required in this case but this can be done in a more flexible way. For example the main ontology may remain unchanged and a new ontology can be created that imports the old. An additional class as a subclass of the Beneficiary class (e.g. SingleMotherBeneficiary) would have been created plus a subclass of the Benefit class (e.g. SingleMotherBenefit) to that ontology. The physical location of the new ontology can be in any site in the web. In any case, ontology modification requires simple text file editing in contrast to a database schema modification, which is usually a more complicated procedure. Our future work will include the development of the provision of a complete semantically enabled web service for a common public administration case in Greece. In particular in our SemanticGov
Goudos S.K.,Peristeras V.,Tarabanis K.
project we will use the Web Service Modeling Ontology WSMO (Roman 2005) Framework for modelling similar cases. In a future version the above-presented application will be updated. The presented case will be the first processing step of a larger application that will finally execute a semantic web service based on the data found in the first step.
Acknowledgements This work has been partially funded by the “SemanticGov” EU project (FP6-2004-IST-4-027517).
References Goudos, S.K., Peristeras, V., Tarabanis, K. (2006) “Mapping Citizen Profiles to Public Administration Services Using Ontology Implementations of the Governance Enterprise Architecture (GEA) models”, 3rd Annual European Semantic Web Conference, Budva, Montenegro. Peristeras, V. Tarabanis K (2006) “Reengineering the public administration modus operandi through the use of reference domain models and Semantic Web Service technologies. in 2006 AAAI Spring Symposium”, The Semantic Web meets eGovernment (SWEG), Stanford University, California, USA: AAAI. Sacco, G. (2006), “User-centric access to e-government information: e-citizen discovery of eservices”, AAAI Spring Symposium Series, Stanford University, USA Berners-Lee, T., Hendler, J., Lassila, O. (2001) “The Semantic Web”, Scientific American. Dean M. et al. (2004) “OWL web ontology language reference”, W3C Recommendation, [online] http://www.w3.org/TR/owl-ref/ McGuinness, D. L., Harmelen F. (2004) “OWL web ontology language overview”, W3C Recommendation, [online] http://www.w3.org/TR/owl-features/ Schwartz, D. G. (2003) “From Open IS Semantics to the Semantic Web:The Road Ahead ”, IEEE Intelligent Systems, Vol. 18, No 3, pp. 52-58. Fikes, R., Jenkins, J., Gleb, F. (2003) “JTP: A System Architecture and Component Library for Hybrid Reasoning.” Proceedings of the Seventh World Multiconference on Systemics, Cybernetics, and Informatics. Orlando, Florida, USA. Wine agent demo [online] http://onto.stanford.edu:8080/wino/index.jsp McGuinness, D. L. (2004) “Question Answering on the Semantic Web ”, IEEE Intelligent Systems, Vol. 19, No 1, pp.82-85. Knublauch, H., Musen, M. A., Rector A. L. (2004) “Editing Description Logic Ontologies with the Protégé OWL Plugin”, International Workshop on Description Logics - DL2004, Whistler, BC, Canada. Roman D. Lausen H., Keller U. (2005) Web Service Modeling Ontology (WSMO), Applied Ontology, 1(1), pp 77-106.
