Introduction
In 2002 Sweet & Maxwell began a long term project to apply a controlled vocabulary to their products, both in print and online. This vocabulary is essentially a thesaurus, structured on the principles of facet analysis, called the Legal Taxonomy. Sweet & Maxwell produce a wide range of legal publications that attempt to cover the gamut of a lawyer's information sources. As everybody working in legal information knows, very few of a lawyer's information needs will be met by reference to just one source. A typical query may start with reference to a textbook, followed by a check on recent case law or legislation, and finally the enquirer may have recourse to up-to-date commentary in a journal. Indexing all of these sources using a controlled vocabulary allows a single query to be formulated, thereby reducing the work that has to be done by the researcher. At its simplest, if Contract of employment is the term always used to describe that concept in all the publications the researcher looks at, then no time need be wasted in seeking alternative descriptions such as Employment contract. Although the advantages to be gained from this approach become more obscure when viewing individual sources in isolation, particularly traditional paper sources, the greatest benefits arise when the entire collection has been indexed consistently. All of the content indexed in accordance with the controlled vocabulary becomes accessible by means of each and every concept described in the vocabulary.
Origins
The foundations of the Legal Taxonomy were laid in 1986 when Legal Information Resources was founded by Christine Miskin and Nigel Smith. Both law librarians, their intention in founding the company was to use their library and information skills to fill some of the perceived gaps in legal information provision. One of the gaps that was identified was the lack of a comprehensive index to articles in legal journals published in the UK, the only source in 1986 being very partial coverage in Current Law. Legal Journals Index was designed as a rotated keyword index so that the abstract and bibliographic information of each article appeared under every keyword assigned to it. Although it was held on a database, it was produced in print as at the time there was no suitable means of delivering it electronically.
One of the initial problems in creating LJI was the lack of a suitable detailed controlled vocabulary that could be used to index journal literature. Ian Sainsbury at the University of Reading had produced a list of legal subject headings for use in libraries in 1974 but this was relatively high level and had rapidly become dated.Footnote 1 The Library of Congress Subject Headings for the Literature of Law was too US biased, variable in its depth of vocabulary and unwieldy to use.Footnote 2 In the same year that the third edition of the Library of Congress list appeared, Elizabeth Moys produced the second edition of her classification scheme which now contained an Index–Thesaurus.Footnote 3 This was a conversion of the previous subject index to a “partial thesaurus” which included some related terms as well as Use and Use for terms (see and see also terms), but did not include broader and narrower terms. All of these schemes were intended for use as subject descriptors for books and were generally not detailed enough to describe journal articles.
Initial development of a more detailed subject vocabulary for indexing journal articles took place over the Summer of 1986. This base vocabulary was added to extensively during both the ongoing indexing process for LJI and a retrospective exercise to index the first six months of 1986. Although it was not initially sold as a separate product, this Legal Thesaurus that was created by Legal information Resources was soon in demand by law libraries, particularly in law firms. It was consequently first published in print in 1987 and subsequently appeared in print in various editions and revisions until 2001.Footnote 4 It was also sold in electronic form to an increasing number of subscription customers during that period.
The decision to create a proper thesaurus structure, rather than a simple list of subject headings, was driven by the need to maintain control over a vocabulary that would continue to grow in order to reflect the literature it was being used to describe. Unless the relationships between terms are recorded structurally, it becomes very difficult to determine whether individual new terms are required because the relationships between existing terms are not immediately visible. In a simple alphabetical sequence the compiler cannot see that Capital punishment already exists when considering whether to add Death penalty. Or when looking at Nationality that Dual nationality is also a term that may be more appropriate because it is more precise.
As noted in the introduction to the first edition, the thesaurus was “based on a detailed classificatory structure of English law” using the standard thesaurus relationships of BT (broader term), NT (narrower term), RT (related term) and PT (preferred term). However it was a polyhierarchical rather than hierarchical structure “because law does not have a terminology that is susceptible to clearcut single relationships”.Footnote 5 This meant that terms could appear within more than one hierarchy in order to reflect the different meanings that could be associated with the term. House of Lords appeared in both the Courts hierarchy and the Parliament hierarchy; Capacity was a narrower term of Children, Contract of marriage, Contracts, Mental health and Testators.
The way in which the thesaurus was created and used for Legal Journals Index in the first few years has been described in more detail elsewhere.Footnote 6 As Legal Journals Index became more and more embedded as a key law library resource, so more law librarians began to use the thesaurus to index their own collections. Paradoxically this popularity had detrimental effects on the thesaurus structure. Built as a detailed vocabulary for journal indexing, it now began to be used for traditional classification in law libraries. This demanded the addition of a notational structure so that each hierarchical level had its own unique number just like a standard library classification scheme. For example, Dispute resolution became 1.4.3 and its NT of Alternative dispute resolution became 1.4.3.1.
Unfortunately this did not sit well with a polyhierarchical structure because the same term now had a different number depending on which hierarchy it was in. House of Lords for example might have been 1.1.7 in the Courts sub-hierarchy of Administration of Justice hierarchy as well as 9.8.6.1.5.2 in the Legislatures sub-hierarchy of Constitutional law. The decision was therefore taken to abandon the polyhierarchical structure and the thesaurus had become mono-hierarchical by the third edition published in 1997. An additional factor in this decision was the difficulties caused by a polyhierarchical structure in electronic searching.Footnote 7
From thesaurus to taxonomy
In 1996 Legal Information Resources was acquired by Sweet & Maxwell. Products such as Current Law that would benefit from the technology and indexing expertise of LIR were moved to Mytholmroyd. This resulted in the development of Current Legal Information on CD which in turn led very quickly to the huge investment programme to create Westlaw UK. By the time the dust settled after the launch of Westlaw UK in 2000, it was clear that a new paradigm had come into being. Very large databases of case law and legislation had been created in a few short years and for the first time, only part of the data held in Mytholmroyd had been indexed using the thesaurus. Recently acquired data such as most of the older case law and all the legislation remained unindexed. Organisational change had also occurred. Legal Information Resources had officially disappeared into Sweet & Maxwell, itself owned by Thomson Corporation, and the editorial work at Mytholmroyd was now closely tied to the Westlaw operation in Eagan, Minnesota.
This huge change of scale prompted new thinking about the purpose and capabilities of the LIR thesaurus. The LIR philosophy had always been to apply keywords from the thesaurus to all of its content. Although the original LIR content was now only a small segment of the data held by Sweet & Maxwell, why not expand the original keywording philosophy to all of the Sweet & Maxwell content? This content included commentary works, journals, law reports and loose-leafs as well as the new Westlaw case and legislation databases. If all of that extensive content could be linked using a common keywording system then the result would be equivalent to creating an encyclopaedia of English law. Sweet & Maxwell had prepared just such an encyclopaedia to challenge Halsbury's Laws of England in the early 1940 s but the only proof copy had been destroyed in the Blitz and had never been recreated.
There were a number of reasons why the thesaurus was not up to that challenge in its existing format:
1. The terms within the thesaurus only reflected a limited range of legal literature – journals and case law. There were not enough detailed terms to accurately reflect paragraphs in Chitty on Contracts for example.
2. Sweet & Maxwell used over 100 subject headings in products like Current Law, yet the thesaurus was structured around only 44 subject headings.
3. There was limited information available within the thesaurus on the scope and meaning of individual terms, making it harder for editors to apply terms consistently.
4. The structure of the thesaurus was one dimensional:
a. Keywords could only appear in one hierarchy. For example Duress only appeared under Criminal law.
b. A hierarchy of terms could only appear under one subject top term. For example the hierarchies for Waste and Environmental health only appeared under Environment, not under Torts or Local Government respectively.
c. The hierarchies tended to reflect associated terms, and were not always true hierarchies. For example:
Families
Children
Birth
Reproduction
In order to address these issues the thesaurus underwent a radical restructure and changed its name to Legal Taxonomy to reflect the break with the past. The terms were now arranged under 111 subject headings to reflect usage in other Sweet & Maxwell products. This entailed considerable consultation as to what those subject headings should be and the result was the adoption of an agreed standard list of such “top terms”. More terms were added based on the material that was going to be keyworded. The vocabulary for Contracts for example was revised based on the text of Chitty. There was much discussion on scope notes that explained the various meanings that terms might have and how they should be used.
However the biggest change was a move to a three dimensional structure that returned the vocabulary back to its polyhierarchical origins to reflect the true nature of legal terms:
1. Hierarchies were completely restructured based on true facet analysis. It was the characteristic of the term that denoted its hierarchy, not its association with other terms. Terms for different types of documents or judgments would always be placed in the hierarchy of Documents or Judgments for example.
2. Terms were repeated in every hierarchy in which they were appropriate. For example Liens appeared under both Remedies and Security.
3. Each subject area contained all the hierarchies that were appropriate to that area. The effect was that the hierarchy of Contract types appeared under several subjects such as Construction law, Contracts, Insurance, Sale of goods, Shipping etc.
This restructuring meant that all relevant terms could be collected under each top term so that the user searching for a term would find everything under the one subject heading rather than being scattered through various hierarchies under different subject areas. The terms were now organised in a consistent fashion in facets. This meant not only that browsing the vocabulary was easier but that indexing was easier because less time was spent looking for the right term to use, and ultimately indexing was more consistent because there was less risk of failing to find the right term.
Structural aspects
The core of the taxonomy structure is the 111 broad, top level terms or subject areas. While it is recognised that the length of this list is well beyond what is usually suggested as the upper limit for any single level of a taxonomy, it is important in compiling such a list to ensure completeness and not to stray too far from some of the “natural” divisions of subject matter which are instantly assumed by anyone familiar with the law. It is also worth noting that not every implementation of the taxonomy has to present the list in its totality. It is possible to manipulate the taxonomy to provide a more focused list so that, for example, a service providing information on “crime” might use only seven of these headings.
A minor anomaly here is that a few of the top terms relate to Scots law. In order to ensure that the whole of Sweet & Maxwell's output can be covered by the taxonomy it does contain terms relating purely to Scots and Irish law. There are a few areas where these concepts do not fit in to the English law structure of the taxonomy and separate top level headings are required. The Scots law of Unjustified enrichment is one example.
The primary relationship within the top terms is that of broader/narrower term thus providing the core hierarchical structure for the taxonomy. A narrower term will generally represent a type or an aspect of its broader term. For example, Costs certificates is a narrower term of Certificates which is a narrower term of Documents.
Facets
The decision to make the taxonomy polyhierarchical, repeating concepts in each relevant subject area, necessarily introduces additional complexity to the hierarchical structure. In order to minimise that complexity, commonly occurring terms are arranged in standard facets such as time, place, persons, documents etc. These facets can then be repeated under each top term as needed.
In determining the standard facets appropriate to the legal taxonomy, there are several that are intuitively familiar to anyone analysing legal materials. Lists of concepts under such facets as Courts or Liabilities are self-explanatory. Other facets end up having rather more abstract labels. Time is a facet which is usually part of any faceted classification but is perhaps not a heading that a lawyer would immediately recognise. However terms within that facet, such as Delay or Limitation periods, will be instantly recognisable. Another standard facet, Vitiating factors, may not be immediately obvious but the terms within it, such as Equitable defences or Estoppel, are familiar enough to clarify the nature and purpose of that facet.
As well as 19 standard facets which occur throughout the taxonomy, the hierarchical arrangement of terms is based on more subject specific facets. For example Charity types occurs only once in the taxonomy under the top term Charities. Descriptive terms such as Charity types or Vitiating factors are, like most of the standard facet terms, not made available for use in indexing but are there simply to assist navigation of the taxonomy by clarifying the hierarchical structure.
Related terms
No hierarchical arrangement of concepts can capture every possible analysis of a subject and additional relationships are required to reach across the standard facets that form the core of the hierarchical arrangement. The standard related term or associative relationship fulfils this requirement. For example, a link is made between Vexatious litigants, a category of person, and Civil restraint orders, a type of court order, or between Schools and Teachers, terms where despite there being an obvious relationship there is clearly not an hierarchical association.
Preferred terms
Underpinning the whole of the taxonomy is the preferred term relationship. The selection of terms for inclusion is on the basis of the language used in the material indexed. As a result, some areas of law will tend to use legislation as the primary source of terminology while other areas will be informed by the language used in secondary sources. The taxonomy is about the identification of discrete concepts, rather than simply the identification of language. The indexing task for which the taxonomy is used is concerned with ensuring that all the material on any given concept is brought together. Assigning a useful term to label that concept and identifying its synonyms follows on from that initial identification.
Scope notes
The choice of vocabulary is supported by the use of scope notes on the definition and usage of terms within the taxonomy. Many aspects of the practice of law rely on the precise usage of language. Incorporating guidance on meaning helps to guide the indexer to a correct understanding of the indexed subject matter and ensure that the researcher is provided with more accurate results. For example the term Cautions has a number of distinct meanings relevant in different areas of legal practice and the following note seeks to ensure that the indexer is aware of the various ways in which the term can be used:
(1) Real property: An entry on the Land Register protecting an interest in registered land by requiring the cautioner to be notified of any dealings in that land. (2) Criminal procedure: A warning given by a police officer to an accused person instead of charging the person. Replaced by “Reprimands and warnings” for juvenile offenders as from September 30, 1998 by s.65 of the Crime and Disorder Act 1998. (3) Criminal evidence: Warning required under Code C of PACE before a suspect is questioned.
Vocabulary size
In establishing the structure of the taxonomy it is important to note that the taxonomy is not, and can never be, fixed and unchanging. As the Sweet & Maxwell taxonomy has developed, the expansion of the material to which the taxonomy has been applied has required an equally large expansion in the number of terms. The level of detail required to index books is much greater than that required to index legal journals. It is also axiomatic that as new materials are indexed, the areas of law covered expand. For example, very little is written in journals on the subject of Temporary road restrictions and it is not until statutory instruments are indexed that such a term will be used very often. The result of these pressures is that the number of terms within the taxonomy has more than trebled, from 5,160 useable terms in 1999 to 17,904 at the time of writing.
Even without such an extension in the scope of its application the taxonomy must change and adapt along with the law. Large scale changes to the language used in legal literature, such as the introduction of the Civil Procedure Rules or the Tax Law Rewrite project, require widespread changes to the language and the structure of the taxonomy. Less extensive changes happen constantly, whether they be the ever changing names and roles of public bodies or the development of new analyses of legal topics in the courts. In keeping abreast of such changes, the scale of Sweet & Maxwell's published output becomes a great advantage. The taxonomy is kept up to date with the involvement of editors across the product range whom, as a group, have access to an impressive array of law and commentary.
Application of the Taxonomy
Within Sweet & Maxwell the taxonomy is used across the whole gamut of its legal publishing activities. This encompasses a range of different sources of information published using different media. In the same way as a publisher applies a house style to its content, much the same reasoning applies to the adoption of a house style for the description of subject matter. As new titles or editions are produced, the indexing is done, as far as is possible, in accordance with the standard vocabulary set out in the taxonomy. Whereas the application of the taxonomy to online material takes the form of a number of keywords chosen from the taxonomy, print indexes compiled using the taxonomy look no different to a traditional back-of-book index. It is solely through the consistent selection of taxonomy terms that the taxonomy provides a subject matter link between all the content, whether print or online.
There are exceptions to the taxonomy's control of back-of-book indexes. Many print indexes will reference aspects of a text which do not fall within the taxonomy's analysis of law, such as the names of particular people or references to the titles of legislation, and these are of course controlled by other means. It can also be the case that an individual text analyses its subject matter in a unique way. Where this represents a development in the law and the way in which it is described, the taxonomy must change to support that. Where it merely reflects the whims of a particular author, an index may need to diverge from the taxonomy's control.
Beyond mere considerations of style the utility of a common scheme for the description of subject matter becomes most visible in the context of online information services. A comprehensive legal research service, such as Westlaw UK, which draws on the whole of Sweet & Maxwell's published output, is made significantly more useful, if it has the ability to perform a single search for all the information on a particular topic. Ensuring that the value of indexes is maintained in an online environment is a challenge. One way in which that challenge may be met is in ensuring that as much consistency as possible is maintained throughout the indexing of all those products which are available online.
Before the benefits of comprehensive searching can be achieved there is a lot of indexing to be done. As already stated this process began with the Legal Journals Index. This continues to be produced, although now only electronically rather than in print. The indexing of Sweet & Maxwell's own journal output now also uses the taxonomy and indeed takes place, in part, in conjunction with LJI. Articles within those journals now display the keywords assigned to them by LJI.
A similar link exists with case law. A large archive of taxonomically indexed case law exists that was initially based on the contents of Current Law but expanded with Sweet & Maxwell's acquisition of Lawtel. New case law is indexed using the taxonomy whether for publication as part of the cases component of online services or as printed law report series. Each law report in a Sweet & Maxwell series also carries the taxonomic keywords assigned to it.
One area for which there was very little background in indexing is the legislation archive. Sweet & Maxwell has recently undertaken the indexing of all UK law in force using the taxonomy. The level of detail at which this has been done is unparalleled, every section of every Act in force having been indexed. A less detailed, but still comprehensive, approach was adopted for secondary legislation.
Global applications
Sweet & Maxwell is part of Thomson Reuters, a worldwide publisher, and opportunities arise to make content available to customers of other companies within that global business. The taxonomy provides one means of enabling that process.
United Kingdom legal material is of interest in other jurisdictions, particularly in Commonwealth countries. The internationalisation of regulation and finance mean that there are few areas of legal practice that can afford to ignore developments out of jurisdiction. Providing access to information in terms with which researchers in diverse jurisdictions are familiar is vital to making existing material as widely available as possible. Knowing what shape the existing indexing takes and having it all described by the taxonomy allows certainty in that process. Currently the taxonomy has been implemented in Australia and Hong Kong and its use is being developed in India and Malaysia. The Australian implementation is the most extensive, with the UK taxonomy being enhanced for the use of the Australian business but still managed from the UK.
Although the taxonomy adopts a structure based on English law it provides a useful means by which access to information in other jurisdictions can be made easier. As it essentially deals in concepts, differences in language do not pose insurmountable problems to its portability. Where changes are required it may need little more than an overlay of more appropriate headings or the inclusion of additional synonyms. The fact that Australia refers to Goods and Sales Tax (GST) rather than Value Added Tax (VAT) provides few difficulties, as the nature of the subject, the concepts involved and their relationships to each other are practically unchanged. Where areas of law are unique to particular jurisdictions, modules can be added on to the taxonomy. For example, the Australian version has an additional subject heading for law relating to Aborigines.
Availability of the Legal Taxonomy
The utility of the taxonomy as a publishing tool has been central to its development since its inception in 1986. It is also available free to anyone else seeking to manage legal information. The taxonomy can be downloaded as separate files for loading to another system or as part of an editorial tool and is updated quarterly.Footnote 8 It has already been noted that from very early days the LIR thesaurus began to be used as a classification scheme for law libraries. The absence of a notational structure as a result of the increased complexity of the taxonomy due to the introduction of a polyhierarchical structure now means that it would no longer be recommended for use as a traditional classification scheme. This does not present a barrier to the usefulness of the taxonomy in an electronic environment however.
As mentioned above, it is not every collection of legal information which will require the full range of subject matter dealt with by the taxonomy nor the detail it contains. It is of course possible to restrict its application and to tailor it to the needs of a particular situation. The ease with which this can be done is determined by the information management system within which it is used. The advantage of using a taxonomy that is utilised across a high percentage of published legal material is that unpublished material in information systems elsewhere can be linked to the published material through use of a common vocabulary. This is exactly the same concept that drives application of the Legal Taxonomy across Sweet and Maxwell products – a virtual subject linking system between otherwise unconnected information.
