Document Assembly and evidence Analysis: Two Approaches to Hypertext

• Autore: Vijay Mital/Anthony D. Elliman
• Pagine: 149-175

Page 149

@1. Introduction

Although the term Hypertext was coined as long ago as 1967 and Hypertext like systems has been widely studied in more recent years we still have much to learn about its appropriate use in information systems development. Rada (1991) describes it as a new technology for producing text and draws the analogy with the introduction of the printing press to explain the necessarily slow pace or development of hypertext based systems. Unfortunately it could be mis very analogy with books (and libraries) which is restricting our thinking.

In discussing navigational problems Nielsen (1990) uses the words ´loosely structured informationª as if this is an implicit feature of all hypertexts. This may be true in some situations, but we believe this to be the exception rather than the rule. Among the most successful and widely used hypertexts are the telephone information systems such as British Telecom's Prestel and on-line documentation, systems such as Microsoft Windows Help. There is also a significant interest in the technology as a media for educational material. In all these cases readers are quite separate from the authors who take positive steps to produce a well defined structure. As with the printing press many users see hypertext as a publshing media.

Central to this issue is the extent to which a link is an unexplained association between two pieces of text. In Bush's Memex (1945) links are just subjective associations left within the corpus of text by its readers. The hypertext software cannot ´knowª why the association is significant, and as Nielsen (1991) says, there is ´no easy way to specify general actions or computations on the dataª. This clearly inhibits us in exploiting the information processing powers of the system being used to store and retrieve the text.

In his discussion of the need for different link types Murray (1993) identifies two types of knowledge or text base:

Page 150

* ´Everythingª In a subject area;

* The knowledge associated with a complex tool.

The second of these Identifies information provided to support a ´mission criticalª task and Murray argues that the process of producing highly structured links within the hypertext Is particularly Important In this latter case. Help like hypertexts extend to other complex activities besides software tools and It can be argued [Huang & Elliman 1991] that there are some common linking structures across this whole range of applications.

However, these views are still restricted by the ´publishingª analogy and we would like to add at least one more type of text base:

* The local knowledge associated with the execution of a business process.

This is a hypertext acquired and structured by an organisation or work group In the course of some business activity. It is local in the sense that, although the business process may be common, the text represents specific Information accumulated by a particular group. Examples of this type are the Issue based discussion models [Rada 1991].

Since we begin with a good knowledge of the activity our hypertext can Include a range of link types to represent meaningful semantics from within the application domain. Further depending on the- strength (or objectivity) of the semantics we may construct the software In such a way as to react to the different semantics to support tasks In an ´intelligentª way. Another aspect of these applications Is that the roles of author and reader are not so clearly distinguished and since users add or modify certain types of links according to their role within the business process.

We will explore the power of this view of hypertexts by looking at two rather different legal tasks which are supported by domain specific hypertext systems:

* Authoring of a new document by assembling Information from previously prepared documents;

* Organisation of ´deadª, evidential documents Involved in a litigation.

The first of these deals with a corpus of text which represents the accumulated knowledge of a work group spanning several cases. It demonstrates that hypertext offers a powerful alternative to other Implementation strategies when the links are given relatively strong semantics so that the system can take simple actions to advise the user during the task. In the second example we deal with a case specific hypertext representing the current perception ofPage 151 an action in progress. This illustrates a system where the strength of the semantics assigned to groups of link types varies. The links in the strongest group actually constrain user actions to maintain the integrity of the text base. In contrast we shall see that those in the weakest group have a more significant semantic value to the user but cannot be interpreted by the system.

@2. Document Assembly

Much of the work of lawyers involves creating complex legal documents - wills, leases, contracts, pleadings, etc. There are in fact three kinds of activities which fall under the umbrella term 'document creation'. These are form-filling, template-based drafting and document assembly.

Form-filling is the process of creating simple and common documents by reworking pro formas. Template-based drafting describes the process of creating more complex documents- which require relatively minor variations on a theme using a template and boilerplate text. When varied situations have to be cateced for and complex documents are required, the construction involves the drawing upon and merging of fragments from previously created documents. The fragments are altered or added to suit the current situation. This- is the process of document assembly.

Document-assembly is qualitatively different from the first two types of document creation, requiring considerable application of legal knowledge throughout the process. In form-filling the assessment of the situation at the very start yields the appropriate pro forma document which can be instantiated in a routine manner. In template-based drafting, many of the decisions en route to the final document require assessment of the import of the various alternatives or options. However, the assessment is not open-ended and is confined to choosing from a small set of predefined elements.

Document assembly is time-consuming, involving significant effort and the exercise of considerable experience and discretion. This activity often requires discrete tasks by a number of workers:

* Lawyers specify the outline form of the document;

* Para-legals and legal secretaries take the outline, insert relevant clauses from sources indicated by the lawyer and make the document structurally consistent;

* Finally the lawyer rectifies omissions and verifies the finished product

The feature of document assembly that is relevant here is the reliance on precedents. Precedents are previously created documents which dealPage 152 with similar or related situations, Clearly, when there Is a choice between constructing a document de novo and adapting one or more existing documents to do the job, there is often little doubt that the latter will be preferred. This Is not plagiarisation, nor even an unwillingness to expend the requisite effort. Reuse of time honoured language Is simply good legal sense.

Sole practitioners and those working In small groups usually have some sort of filing system which holds documents that they have personally created or received In the course of their work. Consequently, they have at least a Yague recollection of what to look for, and where. This allows them to cope with the complex task of using the valuable Information contained In precedents, in a law firm with more than a handful of fee earners, however, this ad hoc approach is not viable, since the volume of information processed is large and users may never have previously read material which they are to, or should, use and adapt. Consequently, law firms are Increasingly turning to computerisation of the task of organising, maintaining and making Information available to multiple users.

There have been a number of approaches to the task of providing computer based support for legal document creation. We will first describe an approach that supports the first two of the activities we have described above - form-filling and template drafting. Following that, we will Introduce hypertext-based work that Is meant for the third kind of activity -assembly of a complex document by drawing upon multiple precedents.

@@2.1. Knowledge-based template systems

This essentially involves the creation of a template into which al the related precedents can be held In an amalgamated form and the relationships between clauses are captured as 'knowledge'. There Is some fixed text which must be included, and Interspersed with or embedded within the fixed text are markers or delimiters which Indicate the places where some action, such as the following, is required:

* An optional piece of text must either be retained or omitted;

* One or more of some given text alternatives have to be selected;

* Some numerical value has to be calculated based on the state of facts and Inserted;

* Some Information In text form has to be Inferred from supplied facts by the system and Inserted;

* The user has to add some text,

Page 153

Each of the actions implied by the delimiters is essentially an act of derivmg or acquiring information. This may be done by executing a procedure which causes some calculations to be done or some rules to be invoked in order to derive the requisite information on the basis of the facts supplied to the system by the user. Alternatively, the information may be acquired directly from the user.

The seminal work in knowledge-based document drafting systems was done by Sprawl (1979) under the aegis of the American Bar Foundation, The system, and the representational language underlying it, are entitled abf. Most commercially available document assembly systems which claim to employ knowledge-based techniques are...