Internet , portal to justice ? Tom

In the Netherlands people with low income are entitled to receive legal aid. In almost every big city in the Netherlands centres for delivering this legal aid are present. Currently these centres are reformed into a Legal Services Counter (Het Juridisch Loket). This reform process has several aims, including improving accessibility, improving the quality and standardisation of legal aid throughout the country, and improving efficiency and consequently diminishing operational costs. To achieve this, the organisation will be turned into a service oriented one and new technologies, like call centre technology and Internet will be used. The Leibniz Center for Law has developed an Internet based knowledge system that should support both future employees of the legal services counters as well as their clients. In this paper we present our experiences in designing and implementing this system in a changing organisation and the impact of the organisational change on the design team. 1 Internet access to the law versus internet access to justice. The internet has become an unavoidable means of communication and together with mobile phones established an enormous market growth. The internet hype that started in the nineties of the last century was caused by the opportunities that were seen to use this communications facility to create new forms of services (so-called e-services). Successful applications can for instance be found in postal order business where different products are offered, varying from books to computers systems. Governments have also discovered the internet, a development called e-Government. In the Netherlands as in other European countries citizens can send their tax returns via the Web1, while a number of Dutch cities support electronic requests for a housing permit2. International studies however show that both citizens and businesses are not satisfied with the level of support offered by their governments (see e.g. [1]). The same study shows that the Dutch government is not playing a leading role in using internet for providing e-services. High volumes of electronic documents might have become available via the Web due to measurements of the Dutch Government, but citizens don’t consider this to be a service to them. Most citizens are relative laymen in the legal domain and therefore don’t want access to legal sources. They rather want answers to questions like “I want to rebuild my house by extending it with a garage, is that allowed?” or “I want to close a contract for this pension arrangement, is this pension arrangement acceptable according to the tax law” (which will make the premium paid deductible from the Dutch wage tax) [2]. These type of questions can be answered by so-called knowledge-based systems, sometimes made available on the Web. An example of such application is the life insurance application of the Dutch Tax and Customs Administration. That application calculates the amount of money that can be deducted from the wage tax. The ‘knowledge’ that is put in those 1See: http://www.belastingdienst.nl 2See e.g.: http://www.cromstrijen.nl/Roccade/GPC/Formulieren.nsf/ Woonvergunning?OpenForm&Woonvergunning; 132 Internet, Portal to Justice? applications springs from the regulations that are related to the legal problem to be solved, completed with expert knowledge (especially needed to resolve legal interpretations). 2 Legal Services Counter (Het Juridisch Loket) When citizens are involved in legal procedures they similarly want to be able to present their case and subsequently receive advice or a judgement about it. Typical cases citizens might want to be advised about are divorce, coping with rental debts and being fired. Citizens are likely to rely on trade unions, consumer organisations and law firms for legal assistance. In the Netherlands, citizens belonging to the lowest income categories are entitled to legal aid at Legal Aid and Advice Centres (Bureaus voor Rechtshulp), governed by the Legal Aid Boards (Raden voor de Rechtsbijstand). These Legal Aid and Advice Centres are currently being replaced by the so-called “Legal Services Counter” (LSC) (Het Juridische Loket). While the old centres could handle client cases all the way up to the court room, the new organisation is meant to focus on first legal aid. Cases that are too complex or not within the aims of the LSC, should be forwarded to other organisations as soon as possible. It was foreseen that the employees of the new organisation will gradually have a different background than the present ones. The level of legal expertise will probably diminish in the future, at least in depth. The typical LSC employee will be a generalist more than a legal specialist. Therefore the Legal Aid Boards were thinking of a knowledge management solution to support (future) employees and clients of the LSC’s in solving legal problems. The Leibniz Center for Law developed such a knowledge management application called “clarification of legal questions” (vraagverheldering in Dutch). This knowledge-based application is part of a virtual front office that is aimed to automatically support legal aid as much as possible. The prospective users are both clients and employees of the LSC’s. The information stored within that application and the way this information can be accessed is designed for reuse. Consequently the application’s functionality supports the physical and the virtual office equally. Basically the functionality supports: • Clarification of a legal assistance request by the client; • Legal advice to the client. One of the aims of the LSC is to enhance accessibility. Furthermore the first virtual client contacts should form a basis for creating an electronic client file which will be used for more elaborate forms of assistance. An equally important goal is the reduction of the operational costs of legal aid by reducing the number of physical contacts with clients (and consequently by reducing the number of staff as well as their educational level and the related labour costs). The use of internet for establishing contacts with the LSC’s clients could have big impact on the future level of easy accessible legal aid for the target group of the current Legal Aid and Advice Centres. In this paper we will describe the development of our knowledge management application from different perspectives, a functional, technological and an organisational perspective. Especially the impact of the organisational change was quite dramatic for the employees and one can imagine that this had an enormous impact on the design teams. The design of neither the new organisation nor its supporting systems has been completed yet and at least three succeeding development plateaus are foreseen. Therefore we can only draw preliminary conclusions based upon our experience in the first plateau. Nevertheless we believe our experiences might be useful for every knowledge engineering practitioner working in large legal organisations. Tom M. van Engers et al. 133 3 Clarification of legal questions at the Legal Service Counter The application for clarification of legal questions supports employees of the LSC’s to determine the legal question(s) behind the posed problem. Based upon the legal questions identified this way a standardised series of products is offered. The application leads the user to an increasingly better defined legal issue by letting him choose between a number of possible answers to a posed question or by asking for specific answers to a question (e.g. “At what date did the contract expire”). This way the user can work his way to questions that have a one-to-one relationship with a set of products of the LSC. These products could be information products, for example an electronic brochure on alimony payments, but also a referral to a lawyer or another organisation than the LSC, or an appointment for an extended (face-to-face) consultation with an LSC employee. The LSC’s products can consequently be categorised in three types: information, referrals or appointments. The design of the “clarification of legal questions” module was guided by the following considerations: 1. It should serve as a kind of check list for the LSC employee or client; 2. The user is not obliged to answer questions or to be forced to answer them in a specific order; 3. Easy problems can be handled quickly, more complex problems should allow for a more detailed exchange of information (cf. JURICAS advice systems that allowed ‘deepening’ at certain moments [6]). Figure 1: Sub-categories of legal domains in the application Clarification of Legal Questions (in Dutch). The user first makes a choice for a certain legal domain and next refines his/her question by selecting one of its sub-categories. Here the user has selected ‘Rent/Let’ as main category and can now choose between ‘House exchange’, ‘Termination of tenancy by owner’, ‘Is the tenant obliged to evict the apartment?’ and ‘termination of tenancy by tenant’. Typically, the user of the application, especially the LSC employee, already has a general idea of the legal domain to which the question belongs. Therefore we present an overview of 134 Internet, Portal to Justice? the legal areas the user can choose from (left column in Figure 1, in Dutch). Some legal areas however are rather broad and therefore a further division in sub-areas is necessary. After choosing a legal area, the user can select a sub-category by picking it from a list. In some cases a problem can be related to more then one legal area. In that case the user can select more than one area and switch between them as he or she likes. Answers to questions are globally available, so users never have to answers questions more than once; these answers will be reused when relevant for other legal areas.3 While the user provides answers to the questions posed by the system, potentially relevant information and other products appear (see Figure 2 at the right side of the screen). As long as the system has not concluded these products are certainly relevant, a question mark is presented in front of them; otherwise a red exclamation mark will be shown. If the user is not able to provide an answer to a question the user may choose ‘unknown’ (in Dutch onbekend). The system then responds by presenting a list of questions that together, when answered, will provide an answer to the original question (see Figure 2). This is a more elaborate and pervasive version of the ‘optional deepening’ the JURICAS systems offered, as mentioned earlier [6]. The user may change the answers to questions at any time, as long as this does not lead to inconsistencies in the data given. Should an inconsistency arise, the user is told he cannot change the answer because of this reason, and all questions that are involved in the inconsistency are indicated with a red triangle in front of them. These inconsistencies may also arise from answers given in other legal areas. Figure 2: If the user is not able to provide an answer to a question he/she chooses ‘unknown’ (in Dutch onbekend). The system then responds by presenting a list of more specific questions that together may provide the top-level answer. In the example here the top question has been answered with ‘unknown’ and so has the first sub-question. This figure presents the interface of the internet or client version of the system. At any time the user may decide to end the question-answer dialogue and continue with either one of the presented products, another legal area, or to submit the gathered data to 3Assuming the questions and answers have the seem meaning in the other area. Tom M. van Engers et al. 135 the rest of the system and e.g. plan an appointment with a lawyer. The user can also get an overview of all answer provided during the consultation session. These answers can be stored on request and reused in succeeding consultations (this might be useful if the user wants to temporarily terminate the session and restart it again, e.g. after collecting some relevant data). If the client is not an anonymous contact, and further action is taken, the client data will be stored in a small electronic file. This e-dossier can later on be used during subsequent consultations or by the lawyer who is going to handle the case. Figure 3 presents a schematic overview of the flow of data through the system. 4 System architecture and adaptability One of the aims of the application is to streamline and increase uniformity of the client handling processes. Clients are to receive legal advice as efficiently as possible. Consequently, the application should play a central role in the client handling processes of the organisation. The creation, maintenance and control of the knowledge (models) is in an approach as depicted of utmost importance. It will be necessary to adapt handling strategies from time to time, e.g. as a result of changed legal procedures or changes in the law. The system should therefore be rather adaptive and both the time needed for adapting the system as well as the costs of maintenance (in fact one of the biggest parts of the costs of ownership, or TCO) play an important role. Besides this it would be desirable if small changes could be implemented by the domain experts themselves rather then involving knowledge engineers to do the implementation for them. 4.1 From rules to decision trees As can be seen from the system description above, the dialogue is presented as question or decision trees. Decision trees can be useful (cf. e.g. [7]), but they are also difficult to maintain and not storable in an efficient way. Questions may occur in more then one decision tree and when changing them, inconsistencies are likely to appear. Furthermore, knowledge engineers find it hard to map knowledge sources directly to decision trees. We’ve consequently chosen to represent the legal knowledge in rules rather then in decision trees. These rules have the well-known format: IF condition_1 AND condition_2 AND AND condition_n THEN conclusion_1 AND conclusion_2 AND AND conclusie_m Using a server-side production rule engine with a java based client interface like JESS may seem the obvious solution, but external requirements for the server (not too powerful but still fast performance), clients (very thin and no use of applets) and bandwidth of the connection (not too broad), forced us to look for a different solution. We developed our own rule language, based on propositional logic. The rules are stored in XML and then automatically converted into decision trees (also stored in XML). The application runs those decision trees. The decision trees are also expressed in XML and an obvious candidate for the translation language was XSLT (eXtended Stylesheet Language Translator). The program is quite complex, the algorithm exponential in nature, but since the separate rule sets are not too big it works in practice.4 An example of such a rule belonging to the legal area of tenant’s rights is as follows: If the landlord terminates a tenancy timely, by certified mail and a valid reason, then the termination procedure is valid. 4The algorithm was specifically developed for this project. We knew of transformations of decision trees to rules, but not vice versa. To transform all the rules in the current version to trees takes about 8 hours on a Pentium 800MHz. The algorithm could probably be improved, but the problem is hard to solve in XSLT. 136 Internet, Portal to Justice? Figure 3: Data flow through the clarification of legal questions. In XML format it looks like this:5