Crystal Reference launches Global Data Model

Crystal Reference launches the Global Data Model, a unique advanced semantic network dataset, based on concepts which derive from linguistics, knowledge management, and language technology. It offers organizations the ability to solve indexing and classification issues through the combination of classified datasets with advanced language engineering techniques.

Crystal Reference launches Global Data Model

West Sussex, UK., April 5, 2002 -
Crystal Reference, chaired by leading language expert, Professor David Crystal, has launched a Global Data Model. The launch follows over five years of development and investment in excess of 5 million.

The patent-pending Global Data Model seeks to answer a number of the problems faced in this information driven age by providing individuals and computers with a robust model through which they may standardize and exchange information.

The Global Data Model (GDM) is a unique dataset and is best described as an advanced semantic network dataset, based on concepts which derive from linguistics, knowledge management, and language technology. It offers organizations the ability to solve indexing and classification issues through the combination of classified datasets with advanced language engineering techniques.

Since pure software solutions cannot currently enable computers to handle complex patterns of meaning automatically, or for that matter to reproduce human methods of thinking, the combination of software and human-generated machine readable datasets such as the GDM provides an immediate pragmatic solution.

The development of the GDM has been undertaken by a team of lexicographers with editorial intervention at every step of the classificatory process: the assignment of a category to a word or concept is always made by a member of the editorial team, who takes full account of context and frequency of use, and the decision is then checked by the general editor.

Commenting on the development, Crystal Reference chairman, Professor David Crystal said, "The model brings together Crystal Reference extensive resources in lexical and encyclopedic products and it is for this reason that we describe the GDM as a 'language and knowledge' framework. And it is the focus on the specific meaning relationships between words (strictly, lexical items) which justifies our use of the term 'semantic network' ".

The chief feature of the GDM is the breadth of its coverage. Because it originated in a general-purpose encyclopedia, its classification categories are capable of handling all domains of human knowledge. And because it has incorporated the coverage of a desk dictionary, it can handle any content-related word from general English vocabulary.

The combination of encyclopedia and dictionary perspectives makes the GDM a unique and powerful classificatory tool, especially in the way it can remove ambiguity and relate words to other words belonging to the same semantic field. Because each word in the dictionary, and each sense of each word, has been assigned an individual encyclopedic classification, ambiguity in searching and retrieval can be eliminated.

The GDM is not just a conceptual classification: each topic domain has been cross-classified according to the kind (genus) of information it contains. Five genera are recognized: general notions, people, places, organizations, and products.

Each classification category is characterized by a carefully compiled set of semantically related keywords. The different senses of 'bridge' are thereby related to other words which belong in the same category - words like 'card' and 'bid' in the domain of card games, words like 'arch' and 'suspension' in the domain of transport technology.

The GDM currently recognizes 1500 topic domain categories with an associated 60,000 keywords. The cross-classification in terms of genera produces a potential further 6000 categories. Over 300,000 English names, words, and phrases have now had categories assigned (the 'classified vocabulary'), enabling the approach to provide solutions to enquiries of a general nature, as well as from several specialized fields. Further development and extension of GDM categorization is anticipated.

Commenting on the opportunities for the GDM, Managing Director Ian Saunders said " The GDM is a tool of considerable power, and has a wide range of potential applications. We are adopting a pragmatic approach to its development, being prepared to develop first those applications which turn out to be of immediate practical use to other organizations. We welcome the opportunity to work in partnership with organizations where our advances can be effectively utilized."

Background to Crystal Reference and origins of the Global Data Model

Crystal Reference develops and manages an extensive range of content databases that originated in the mid-1980s as part of a project undertaken by Cambridge University Press. In time, these databases have grown significantly to form the basis of a number of print, online, and electronic publishing projects, including all four editions of the Cambridge Encyclopedia and associated volumes, and the first edition of the New Penguin Encyclopedia.

In developing the core content database, editor Professor David Crystal needed to classify the individual entries in order to provide a sophisticated method of indexing and searching. He therefore devised a scheme which assigned the distinctive features of each entry to a hierarchical framework of categories. The biographical entry on Winston Churchill, for example, had a dozen categories assigned, recognizing his many roles (such as politician, painter, writer, and soldier). The categories were chosen after discussion with the various specialist contributors who were writing the database entries - for example, the large domain of natural history categories came from a team based at the Natural History Museum in London.

The scheme was tested by applying it to all the entries included in the range of enyclopedias belonging to the Cambridge 'family'. It was then applied to the content database as a whole, thereby greatly enhancing the ability of editors and researchers to interrogate the database with questions ranging from the most general (e.g. all places, all people) to the most particular (e.g. all 19th-century male French novelists). The system was also much used by encyclopedia readers, who were encouraged to send in 'datasearch' questions about the coverage of a topic throughout the encyclopedias.

The commercial application of this concept was recognized by AND International Publishers, who in 1997 acquired the database from Cambridge University Press. In conjunction with Professor Crystal, AND invested in the development of the approach, acquiring a wide range of data to move towards a global coverage of subject-matter. Between 1997 and 2001, some 5 million was invested by AND into the wider development of the approach, now referred to as the Global Data Model (GDM, patent-pending), to include organizations and products as well as further locations and people. The classification of general terminology was also increased, through the addition of all content-related entries from a large desk dictionary.

The initial approach to classification was extremely successful, chiefly because of the painstaking way in which the classification categories had been refined and tested using individual encyclopedia entries over a 10-year period. Other possible uses for the scheme were therefore sought, and several prototype applications were developed, including a search-engine assistant known as ALFIE (A Lexical Filter Internet Enquirer) and a method of Automatic Document Classification. Following the acquisition of the classification scheme by Crystal Reference in 2001, the scheme has begun to evolve into a language and knowledge framework, with the immediate aim of providing rapid solutions for people who need to organize large quantities of topically diverse data in a coherent, accessible, and efficient way. A long-term goal is to make a contribution towards a classification standard for data within computer systems on a global level.

Background to Professor David Crystal

Professor David Crystal OBE, FBA is one of the world's foremost authorities on reference publishing and on language.

He was a lecturer in linguistics, first at Bangor, then at Reading. He held a chair at the University of Reading for 10 years, and is now Honorary Professor of Linguistics at the University of Wales, Bangor. These days he divides his time between work on language and work on general reference publishing. He has written over 100 books, and travels throughout the world lecturing on linguistic and reference subjects.

David Crystal is editor of the first four editions of the Cambridge Encyclopedia, the Biographical Encyclopedia, and Cambridge Factfinder for Cambridge University Press. Under AND's ownership of the database, his role was as Editorial Consultant with responsibility for all content.

David Crystal is currently chair of the UK National Literacy Association (NLA), patron of the International Association of Teachers of English as a Foreign Language (IATEFL) and of the National Association of Professionals concerned with Language Impaired Children (NAPLIC). He is past president of the Society of Indexers. He has been a member of the Board of the British Council and is currently a governor of the English-Speaking Union. He received an OBE for services to the English language in 1995.

David's role within Crystal Reference is as Chairman with a lead role in the development of new content, and as overseer of the long-term maintenance of the database.

Organizations looking for more information about Crystal Reference, should call +44 (0) 1798 817207, email info@crystalreference.com, or visit the Web site at www.crystalreference.com