Formal Models for a Legislative Grammar. Explicit Text Amendment

In this paper we will present a method for mining information within legal texts, in particular in regards to corpora of statutes. Text mining, or more in general Information Extraction, can provide a valuable help to people involved in research about the linguistic structure of statutes, and, as a side effect can be the seed for a new generation of applications for the validation and conversion in the legislative domain. 1 Scope and Assumptions For the communication of legislative sources through the Internet, the parliamentary and governmental institutions of many countries1 have begun a process of converting their “deposits” of these into a standard format for facilitating the retrieval and display of texts. The XML mark-up language seems to be the tool deputised for reaching this scope2. In fact, this language combining its dual nature as a mark-up language and a Web standard, is able to form the common ground for action both “at the source”, namely, legislative drafting, and action “downstream” relating to the publication of the texts and the identification of tools for accessing legislative information [2] [10] In Italy, the introduction of the XLM language for processing legislative instruments was proposed and experimented in the “Norme in rete” [Law on the Net] Project, solicited by the 1 See, for example: the data bank of legislative instruments of the Australian State of Tasmania at http://www.thelaw.tas.gov.au and the DTD relating to the parliamentary acts in the United States on the site http://xml.house.gov. 2 Although standard languages are available, it is nevertheless necessary to reconcile the use of these languages for the man-machine dialogue with the citizen’s access to and fruition of information. Access to and fruition of the mass of public data, although transmitted by machines, cannot be achieved with artificial languages, but only with non specialised communicative formats like natural language. Therefore , it appears that the automated processing and recognition of natural language, has a fundamental role to play in this man-machine interaction. Ministry of Justice, financed by AIPA (Agenzia per l’Informatica nella Pubblica Amministrazione) [Agency for Informatics in the Public Administration] and developed under the guidance of the Istituto di Teoria e Tecniche dell’Informazione Giuridica (ITTIG) of the Italian National Research Council. The project has produced, amongst other things, the DTD rules adopted as a standard by AIPA3 for the on-line publication of Italian legislative instruments. In order to adopt this language as a standard and, above all, for the conversion of the legislative instruments in force into the format provided for by the DTD rules, two factors, in our opinion, must interact. Definition and promotion of a “controlled” legislative language Rules for law-making or techniques for legislative drafting have introduced unambiguous and recurrent elements into legislative instruments, whereby it is possible to identify a more controlled language in legislative language compared to natural language. In fact, specific rules of orthography, lexicon, syntax, style and structure for the drafting of legislative instruments have been adopted. A collection of these rules is to be found in a Circular4 issued, in 2001, by the President of the Council of Ministers, and the Speakers of the Chamber of Deputies and the Senate and adopted by the Government and Parliament. The Circular updates an earlier one of 19865. For drafting their legislative measures, almost all the Regions in Italy have adopted the “Rules and Suggestions for Drafting Legislative Texts” Manual, a set of rules that are almost the same as the state rules, compiled in 1991 and updated in 2002. These rules have been applied and complied with in the drafting of legislative instruments enacted by the State and Regions since the end of the 1980’s6. Leafing through the legislative documents, it cannot be said that these rules have, up until now, been strictly and uniformly applied by all law-makers. However, some analyses of sample texts have shown that the use of the legislative drafting rules is spreading. The drafting of other legislative documents (such as the regulations of local authorities, collective contracts, etc.) is not bound by these rules. It can, however, be said that it is widespread, in practice, to make reference to these drafting rules, even if their application depends on the sensitivity and knowledge of the draftsman. On the other hand, many initiatives are underway for the formal and binding adoption of the State/Regional drafting rules by all those persons who produce legislative documents. The “Norme in rete” [Law on the Net] Project has contributed to accelerating the process for spreading and receiving the 3 AIPA Circular, 22 April 2002, “Formato per la rappresentazione elettronica dei provvedimenti normativi tramite il linguaggio di marcatura XML” [Format for the Electronic Representation of Legislative Provisions by Means of the XLM Mark-up Language]. The text can be consulted at : http://www.normeinrete.it/standard/Circular-xml.htm. 4 Circular 20 April 2001, no. 10888 of the Presidency of the Council of Ministers, “Regole e raccomandazioni per la formulazione tecnica dei testi legislativi” [Rules and Recommendations the Technical Formulation of Legislative instruments], published in the Gazette Ufficiale No. 97 of 27 April 2001. The same rules have also be adopted by the Chamber of Deputies and the Senate with identical circulars by their relative Speakers. 5 Circulars of the Speaker of the Senate, the Speaker of the Chamber of Deputies and the Presidency of the Council of Ministers of 24 February 1986 (G.U. No. 123 of 29 May 1986, Supplemento ordinario No. 40). For an in-depth illustration of the rules for legislative drafting in Italy and in Europe see: [16] [17]. 6 For the historical framework of Italian legislative drafting, see [25]. legislative drafting standards, drawing attention to their utility for electronic processing, whilst still maintaining that the main purpose of these rules is to guarantee greater clarity in and ability to understand legislative instruments. At the same time, research into legal theory, legal language, and legal artificial intelligence have contributed to the definition of the syntactical and semantic structures and to morphological and lexical behaviour peculiar to the legal discourse. Use of tools for natural language recognition It is evident that the presence of common rules consolidate the definition of text models, which the interdisciplinary studies we have just mentioned describe with ever increasing exactitude. It is also evident that this modelling assists in the automated recognition of the structures of legislative instruments and their tagging according to the XLM standard. In fact, this tagging will be difficult to obtain from the law-maker as it is extraneous to the tasks and objectives involved in his/her normal activities. If other professionals do it later, it may provoke an often unsustainable increase in the time needed and the costs involved in building and managing the legislative knowledge base, structured according to XLM standards. It is within the perspective of the implementation of a parsing system efficient for the automated recognition of the structures of legislative instruments and the subsequent tagging and conversion of these texts in XML format that we shall now begin the description of the research presented here.