Jusletter IT

Article 23 of the Treaty on the functioning of the European Union gives every citizen of the Union the right to consular and diplomatic protection if his or her Member State is not represented in a specific third country. Whichever mission (of another EU member state) the EU citizen ends up asking for support, the mission has to provide support on the same conditions as for their own nationals.

Article 46 of the Charta on Human Rights lays down the same right. The Green Paper «Diplomatic and consular protection of Union citizens in third countries», presented by the Commission in 2006, focuses on strengthening this right: In it, the European Commission points out that European citizens are not fully aware of this right, and that the legal consequences of it are far from being fully implemented by the Member States. After the consultation phase of the Green Paper, the Action Plan 2007-2009 «Effective consular protection in third countries: the contribution of the European Union» was adopted. One important measure is the examination of Member States' legislations and practices on consular protection and the assessment of the extent and nature of the observed discrepancies between Member States.

The CARE (Citizens Consular Assistance Regulation in Europe) project (http://www.careproject.eu) aims at offering tools to the Commission which support the European Commission in performing this examination. A report will analyze the legal framework in all the EU member States, and a database will contain all relevant legal documentation from all EU countries. The database will also be accessible by all European citizens via the Internet. The project is due to be finished by the end of July 2010. At present (February 2010), the project partners have started to present and discuss the project’s interim findings in various workshops which are all open to the public: «EU Member State Practice on Consular Protection» (Vienna, January 2010), «Catastrophes and Civil Protection» (Bologna, April 2010), «European External Action Service and Consular Protection» (Brussels, May 2010).

In this paper, we would like to develop some ideas following a constant observation that we have been making during the course of the project: The consular staff of all Member States have significant difficulties in coping with requests for consular assistance. The growing quantity of requests, as well as the expected quality of consular services, causes difficulties. This is not at all surprising when we look at the political and social background of this situation: More and more EU citizens are travelling abroad. The number of EU expats has constantly been increasing. Catastrophes like the recent earthquake in Haiti require instant and efficient support for distressed EU citizens. All this, however, happens in a time when all Member States have to cut back on expenses for consular services. Member States have to consider decreasing consular staff, and even closing down consulates. It is obvious why this situation is problematic. Two main solutions have been proposed: increased co-operation between EU Member States, and further automation of consular services. For the rest of this paper, we will focus on the second option and present our ideas for further automation of consular services.

A close reading of Article 23 of the TFEU results in the following conclusion: Every consulate abroad should be able to handle every consular case (see for the long list of functions Article 5 of the Vienna Convention on Consular Relations), taking into account the 27 different consular protection laws and policies and – not a rule but «best practice» – in all the 23 Community languages. It is obvious that this puts almost impossible demands on the people who have to provide consular assistance. Even when only dealing with one national legal framework, it has to be noted that a good administration of one consular law requires also a proficient knowledge of related consular practice (bulky handbooks that are often not disclosed to the public), based on extensive training. Especially small consulates certainly cannot cope with these demands.

On a practical level, it is interesting to note that despite this seemingly desperate legal framework, Member States successfully handle such questions, opting for very pragmatic solutions, often on a regional level.

Handbooks in paper have long ceased to constitute best practice. Websites and information systems are able to very nicely present the complex knowledge while coping very efficiently with often daily updates (just think of travel recommendations). Thus, we propose – based on the CARE database – the implementation of an advanced electronic consular law and practice information system.

A standard lexical ontology on consular law describes the content of the document collection, providing conceptual access to the stored texts. The indexing of the document collection is automatically performed by the information retrieval system. This lexical ontology is available in all Community languages. By making use of it, conceptual access to the documents is available in all the various languages. Using the results of the EU LOIS project (which we were a partner of from 2004 to 2006), multi-lingual information retrieval is offered. On top of that, we propose the formalization of the contained information using the LKIF ontology. The use of such a concise description of the legal domain holds significant potential in terms of further automation.

Most of the dialog with the citizen could in fact be handled by a computer in the form of a dialog system. EU citizen may employ text, speech, graphics, haptics, gestures or other communication modes for input. A list of questions would support the citizens in order to establish a coherent documentation of his or her request and all related facts. The dialog would be offered in the languages of the State concerned, taking into account its often very specific consular law.

The consulates on site will support the EU citizen by providing the input, but also by taking the role of a notary, if this turns out to be necessary. Such functions can be asked for via the dialog system but also by direct communication lines between the consular services of the Foreign Ministries and the consular agent.

Only in well-defined standard cases, the consul will still have to make the decision him- or herself. This rule is required as the consul may not know very well the applicable consular law but also the language. Normally, the request will be transmitted to the consular services of the country concerned. Very shortly, the consular services will give orders to the consular agent on site. The citizen will be informed via the dialog system of the decision and possible support. If necessary, the EU citizen can speak via video or telephone link to the responsible officer.

Since the 1990ies, ontologies as a conceptualisation of a domain are considered as a tool for organising legal knowledge. Later, the idea of a semantic web [Berners-Lee 2001] with a mark-up that makes the text intelligent and active energized the concept of legal ontologies. For a long time, the University of Amsterdam has set the standards of legal ontologies with LRI-Core and now LKIF [Hoekstra/Breuker/De Bello/Boer 2007]. Legal ontologies were implemented for tasks of conceptual information retrieval, knowledge representation, multilingual information retrieval or exchange of information and knowledge (see [Casanovas et al. 2007] and [Casellas et al. 2009]).

In our case, we consider using two ontologies: a lexical ontology like in the LOIS project [Dini et al. 2005] and a much more developed Dynamic Electronic Legal Commentary Ontology [Schweighofer 2006] (see below).

A thesaurus for indexing contains a list of every important term in a given domain of knowledge and a set of related terms for each of these terms. A lexical ontology builds up from this basis with works on glossaries and dictionaries, extends the relations and makes this knowledge computer-usable in order to allow intelligent applications. Lexical ontologies provide this formalised description of a domain that can be understood and re-used by a knowledge system.

Based on already existing indices and sketchy conceptual structures, a lexical ontology with about 500 legal and factual descriptors with definitions and relations with be established. Content relations will be taken from standard WordNet relations (especially hyperonymy and hyponymy). For all concepts, an ILI will be created in order to support multilingual use but also multilingual retrieval.

Legal retrieval remains the best solution for determining the similarity between the records and the queries [Manning et al. 2008, Turtle 1995, Schweighofer 1999]. For smaller domains like consular law with a complex structure, hypertext systems are a powerful tool. The flexible way of access with a non-linear representation of knowledge allows a user-friendly access to this body of knowledge.

Our document retrieval system is going to be built using Apache Lucene [Apache Lucene 2010, Gospodnetic & Hatcher 2006], which offers state-of-the-art text retrieval capabilities. Like even commercial providers in the field of legal text retrieval (e.g. Wolters Kluwer Spain or Weblaw, Bern) we will opt for open-source technology rather than commercially available search solutions. Standard commercial information retrieval packages certainly work well for normal search scenarios over diverse document collections, but we would like to fine-tune our information retrieval system according to the document properties of our text collection.

Apache Lucene fulfils also the requirement of easy maintenance of the text corpus but also an efficient handling of the various versions of changing applicability. We are considering using the well-developed system of the Austrian legal information system RIS (http://www.ris.bka.gv.at).

The Dynamic Electronic Legal Commentary (DynELCom) [Schweighofer 2006] CONSUL consists of a legal information system, a legal ontology, the world ontology, semi-automated indexing tools for text summarisation and text categorisation, and a sketchy inference machine. A document typology with segmentation rules will support retrieval and text analysis. The goal is an ontological index like that in legal commentaries, however, without the textual components. Even if readability of such ontological structures may require some learning, in the very end the capture of the underlining conceptual und logical structure of the legal system may be improved. Further, adaptable text extraction and summarization tools support browsing and faster reading of legal documents. Further, the DynELCom should support the automated generation of dialogues with user. A knowledge base with an ontology of 500 thesaurus entries, text extraction rules, document categorization, summarization tools and conceptual analysis of text corpora has to be established. The DLEC is based on a co-operative model of work between the man and the computer. The DLEC offers real time indexing and analysis of the legal system.

Like in information retrieval, we will use standard technology for building the dialogue system (see for an overview of options [Wikipedia: Dialogue Systems 2010]. In the beginning, the dialogue will be text-based with a graphical user interface. A spoken dialog system is strongly in consideration; first thoughts are focused on a free spoken dialogue system called Olympus (http://www.ravenclaw-olympus.org/ ). The dialogue will be between the local computer and the human being with the option of interference of a third party, e.g. the responsible foreign ministry or the consular agent. The dialogue management will be handled by RavenClaw, a task-independent dialogue engine. Olympus supports engines form the CMU Sphinx family for speech recognition but provide an interface for other engines as well. Natural language understanding is done by a robust parser called Phoenix.

It should be noted that the dialogue system is considered only for the first part of the consular case. Later, a human interaction by telephone or video will take place if necessary.

In a later stage of the system, the applicant may be guided to a fully-developed e-government system.

The main part of the dialogue system consists of a well-structured list of questions in order to establish the facts for the case. Questions are mostly automatically generated using the ontology. A sequence of dialogue boxes leads the user through a series of. This component is a decisive one as ministries should easily maintain the knowledge base in order to react very quickly to ongoing changes in consular law.

The dialogue will be offered in the relevant national languages. However, we consider having a multilingual part for selecting the best dialogue and available languages.

In this paper, we have given an outline of a system for semi-automatic application of consular law in a multilingual and multinational environment. Only already existing standards tools were considered. For the moment, we are working on a more sophisticated outline.

The CARE Project is funded by the European Commission, Grant No. JLS/2007/FRC-1/50-30-CE-0226854/00-31.

Gospodnetic, O., Hatcher, E.: Lucene in Action. Manning Publications (2005).
Ferrer, A.S., Rivero, J.M.M., García, A.M., Improvements in Recall and Precision in Wolters Kluwer Spain Legal Search Engine. Computable Models of the Law, Languages, Dialogues, Games, Ontologies 2008: 130–145 (2008).
Apache Lucene, High-performance, full-featured text search engine library. http://lucene.apache.org/java/docs/ (14. Februar 2010) (2010).
Berners-Lee, T. et al., The Semantic Web. In Scientific American Vol. 284 (05/2001). Scientific American Inc., New York, NY, 34-43 (2001).
Casanovas, P., Biasiotti, M.A., Francesconi, E., Sagri, M.T. (eds.), Proceedings of LOAIT 07, II. Workshop on Legal Ontologies and Artificial Intelligence Techniques, pp. 43–64. www.ittig.cnr.it/loait/LOAIT07-Proceedings.pdf (2007)
Casellas, N., Francesconi, E., Hoekstra, R., Montemagni, S. (Eds.), Proceedings of LOAIT 2009, 3rd Workshop on Legal Ontologies and Artificial Intelligence Techniques joint with 2nd Workshop on Semantic Processing of Legal Text. Barcelona: IOT Series
Ferrer, A.S., Rivero, J.M.M., García, A.M. Improvements in Recall and Precision in Wolters Kluwer Spain Legal Search Engine. Computable Models of the Law, Languages, Dialogues, Games, Ontologies 2008: 130–145 (2008).
Gospodnetic, O., Hatcher, E.: Lucene in Action. Manning Publications (2005).
Hoekstra, R., Breuker, J., De Bello, M., Boer, A., The LKIF Core Ontology of Basic Legal Concepts. In: Casanovas, P., Biasiotti, M.A., Francesconi, E., Sagri, M.T. (eds.) Proceedings of LOAIT 07, II. Workshop on Legal Ontologies and Artificial Intelligence Techniques, pp. 43-64. www.ittig.cnr.it/loait/LOAIT07-Proceedings.pdf (2007)
Manning, C.D., Raghavan, P. & Schütze, H. (Eds.), Introduction to Information Retrieval. Cambridge: Cambridge University Press (2008)
Olympus, architecture for spoken dialog system. Carnegie Mellon University (CMU). www.ravenclaw-olympus.org/ (14. Februar 2010) (2010).
Schweighofer, E. (1999), Legal Knowledge Representation, Automatic Text Analysis in Public International and European Law. The Hague: Kluwer Law International (1999).
Schweighofer, E. Computing Law: From Legal Information Systems to Dynamic Legal Electronic Commentaries, in: Cecilia Magnusson Sjöberg and Peter Wahlgren (eds), Festskrift till Peter Seipel.: Norsteds Juridik AB, Stockholm pp. 569–588 (2006).
Turtle, H., Text Retrieval in the Legal World. In: Artificial Intelligence and Law, Vol. 3, Nos. 1–2, 5–54 (1995).

Erich Schweighofer, Associate Professor; Anton Geist, Project Assistant
University of Vienna, Faculty of Law, Centre for Computers and Law, Departement of European, International and Comparative Law, Schottenbastei 10-16/2/5, 1010 Wien AT
erich.schweighofer@univie.ac.at; anton.geist@univie.ac.att; http://rechtsinformatik.univie.ac.at