In this paper an automated solution for finding cases for analysing the impact of legal change is proposed and the results are analysed with the help of a legal expert. It focuses on the automatic classification of 15,000 judgments within civil law. We investigated to what extent several machine learning algorithms were able to classify cases ‘correctly’. This was done with accuracies around 0.85. However, the data were scarce and the initial labelling not perfect, so further research should focus on these aspects to improve the analysis of the impact of legal change.

In this paper we present the results of a study to see whether automatically generated concept maps help users of legal information systems in understanding the topics of documents they retrieve. A small formative evaluation with novice users is presented. We did not find a significant difference between the ability to connect the correct visualisation to a document between a topic cloud and concept map approach. Topic clouds are probably a little easier to understand quickly in a superficial way.

Since the OpenLaws portal is envisioned as an open environment for collaboration between legal professionals, recommendation will eventually become a collaborative filtering problem. This paper addresses the cold start problem for such a portal, where initial recommendations will have to be given, while collaborative filtering data is initially too sparse to produce recommendations. We implemented a hybrid recommendation approach, starting with a latent dirichlet allocation topic model, and progressing to collaborative filtering, and critically evaluated it. Tentative conclusion is that giving recommendations, even bad ones, will influence user selections.

A growing number of Dutch court judgments is openly distributed on Rechtspraak.nl. Currently, many documents are not marked up or marked up only very sparsely, hampering our ability to process these documents automatically. In this paper, we explore the problem of automatic assignment of a section structure to these texts. We experiment with Linear-Chain Conditional Random Fields to label text elements with their roles in the document (text, title or numbering). In this subtask, we report F1scores of around 0.91 for tagging section titles, and around 1.0 for the other types. Given a list of labels, we experiment with Probabilistic Context-Free Grammars to generate a parse tree which represents the section hierarchy of a document. In this task, we report an F1score of 0.92.

In this paper we present the results of ongoing research aimed at a legal recommender system where users of a legislative portal receive suggestions of other relevant sources of law, given a focus document. We describe how we make references in case law to legislation explicit and machine readable, and how we use this information to adapt the suggestions of other relevant sources of law. We also describe an experiment in categorizing the references in case law, both by human experts and unsupervised machine learning. Results are tested in a prototype for Immigration Law.

This paper describes the outcomes of experiments in automated support for argument reconstruction from natural language texts. We investigated several possibilities to support a manual process by using natural language processing, from classifying pieces of text as either argumentative or non-argumentative to clustering text fragments in the hope that these clusters would contain similar arguments. Results are diverse, but also show that we cannot come a long way without an extensive pre-tagged corpus.

This paper describes ongoing research on automatically determining relevant context to display to a user of a legislative portal given the article they are retrieving, purely based on ‘objective’ criteria inferred from the network of sources of law. A first prototype is presented and a formative evaluation of it by legal expert users. Results are promising, but there is room for improvement.

In this paper we present the results of an experiment in automatic concept and definition extraction from written sources of law using relatively simple natural language and standard semantic web technology. The software was tested on six laws from the tax domain.

In this paper we present the results of two studies to see whether the analysis of the network of citations between cases can be used as an indication of the relevance and authority in the Dutch legal system. Fowler e.a. have shown such results for the US common law system, but given the different status of case law in continental tradition it is not clear whether this will hold in the Netherlands. Moreover, we introduce a way to validate the results using selections made by human experts for legal education. We discuss the results and conclude that network analysis of cases is a useful tool for legal research.

This paper presents results of an experiment in which we used machine learning (ML) techniques to classify sentences in Dutch legislation. These results are compared to the results of a pattern-based classifier. Overall, the ML classifier performs as accurate (>90%) as the pattern based one, but seems to generalize worse to new laws. Given these results, the pattern based approach is to be preferred since its reasons for classification are clear and can be used for further modelling of the content of the sentences.

There are several reasons why citizens, businesses and civil servants need access to regulations. Unfortunately, traditional approaches that aim to provide this access fall short, especially in the area of spatial planning. Fairly straight-forward questions such as "where will I be able to perform this kind of activity" or "is this activity allowed here" are not answered automatically by current systems. There are many attempts to create one-stop-shop front-ends to eGovernment, but these are seldom built from the perspective of the user.

The ultimate goal of the research line described here is support for automated modelling of sources of law. One of the first steps is the automatic recognition of norms. In earlier work we presented a categorization of norms or provisions in legislation. We claimed that the categories are characterized by the use of typical sentence structures and that this would enable automatic detection and classification. In this paper we present the results of experiments in such automatic classification of provisions. We have defined fourteen different categories of provisions, and compiled a list of 88 sentence structures for those categories from twenty Dutch laws. Based on these structures, a parser was used to classify the sentences in fifteen different Dutch laws, classifying 91% of 592 sentences correctly. It compares well with other, statistical approaches. An important improvement of our classifier will be the distinction of principal and auxiliary sentences.

This paper describes extensions to our native XML legislative editor MetaVex, dealing with specific requests of legislative drafters: The automatic generation of amending documents based on the editing of an existing law or proposal, and the automatic consolidation of (proposed) legislation based on the original and its amending documents. Moreover, we try to automatically detect (potential) clashes of amendments and amendments of amendments.

The work described here builds on [1], where we presented a categorisation of norms or provisions in legislation. We claimed that the categories are characterized by the use of typical sentence structures and that this would enable automatic detection and classification. In this paper we present the results of experiments in such automatic classification of provisions. We have defined fourteen different categories of provisions, and compiled a list of 81 sentence structures for those categories from twenty Dutch laws. Based on these structures, a parser was used to classify the sentences in fifteen different Dutch laws, classifying 94% of 476 sentences correctly. It compares well with other, statistical approaches. An important improvement of our classifier will be the distinction of principal and auxiliary sentences.

The Leibniz Center for Law is involved in the project DURP (Digital exchange of spatial plans) which develops an XML-based digital exchange format for spatial regulations. Involvement in the DURP project offers new possibilities to test the feasibility of a machine readable regulative framework related to an object oriented representation of the real world such as the zoning plan. This marriage enables access to legal source through maps and vice versa. The combination supports increased transparency and delivery. We describe the combined architecture of the MetaLex XML schema for 'regular' legal sources that we developed in the past with geospatial regulatory information. We have developed a prototype viewer application and demonstrate how the spatial planning information in GML can be combined with MetaLex XML using the Web Ontology Language OWL.

The Dutch Tax and Customs Administration (DTCA) is one of many organizations that deal with a multitude of electronic legal data, from various sources and in different formats. In this paper, we describe the results of a study aimed at better access to these sources by having a supplier and format independent knowledge store that describes the sources and their interrelations in a semantic network. Furthermore we developed parsers to automatically detect the identity of sources and typed references within the sources to other legal documents. These parsers can be used to fill and update the semantic network as new documents are added.

In the last decades the interst in the problem of comparing and harmonizing legislation has been steadily increasing. One reason is the increasing legal convergence between governments in the European Union, and the increasing traffic of people over borders of jurisdictions. Another reason is the increasing globalization of companies; Products and services are offered in many jurisdictions at the same time, and the product or service has to meet the provisions of all jurisdictions in which it is offered. In the E-POWER project relevant tax legislation and business processes are modeled in UML to improve the speed and efficiency with which the Dutch Tax and Customs Administration can implement decision support systems for internal use and for its clients. These conceptual models have also proven their usefulness for efficient and effective analysis of draft legislation. We are currently researching whether conceptual modeling can also be used to compare 'similar' legislation from different jurisdictions to improve the capacity of the Dutch Tax and Customs Administration to react to future consequences of increased movement of people, products, and money between EU member states and increased harmonization between tax authorities in Europe. In addition, addressing the problem of comparing models is also expected to improve our methodology for modeling legislation. This paper discusses problems and requirements of comparing legislation as we understand them now, and attempts to relate them to relevant research.

No abstract available.

Traditionally Legal Knowledge Based System (LKBS) reinformation retrieval environment.A successful prototype has been developed, and we are now in the phase of building the actual system.

In this paper, we describe which roles deep structures of