Cascade Systems Ltd has launched MediaSphere 1.0 a database engine aimed at the publishing industry, designed to make searching for information a simple, even instinctive, task. Normally, when searching a database keywords are used. For example, if you want to search for green parrots, you type in green AND parrots and you get the intersection of the two data sets. According to Richard Patterson, chairman of Needham Market, Suffolk-based Cascade Systems, the systems integrator and software manufacturer for the printing industry, the problem with this method is that data is being accessed through a narrow gateway comprising the keywords entered, and results are merely entries that conform those keywords. You can also search for green ‘OR’ parrots for the contents of both data sets. This approach relies on the typing in of keywords and their presence in the article. This form of search was developed in the US and relies on Boolean algebra. More advanced data search engines offer a thesaurus so synonyms of the keyword are suggested which enables the search to be broadened and data entries are seeded according to the number of times keywords occur. While the US academics were working on this type of search in the 1970s, Cambridge University in the UK, academics were working on a method to broaden the search gate by removing the need for keywords and enabling a natural language database search. With the probablistic mathematic method they developed a user can find everything about green parrots and not just get the intersection of these words. In this search each word becomes a mathematical variable, weighted according to its distribution throughout the database. Therefore, common words like ‘everything’ and ‘about’ will be discounted, while the less common word ‘green’ will have a higher value and ‘parrots’ the highest. The search finds not only those items that fulfil all the criteria, the intersection, but also those with the high value words. When the search is completed, items will be rated according to their scores. This approach is at the core of MediaSphere 1.0. Cascade claims the MediaSphere is the first database engine for text, image and video. MediaSphere’s initial target is magazine and other publishing houses with huge data stores that are not digitally managed. Filing cabinets and librarians Until now the filing of images, text, and video footage has been primarily hard copy, with vast arrays of filing cabinets and librarians to manage the data. Despite full digital pagination, no effective database engine for managing and searching pictures and text existed. But manual filing and single copies means data is easily lost. If an image is removed from a folder and not replaced or the folder is not returned to the right place the data is effectively lost. The result is that the same photograph has to be commissioned over and over. These problems in handling data have erected too high a barrier to republish it, according to Patterson. The only digital libraries have been text only or text and picture data separately. Cascade’s MediaSphere 1.0 promises a full multimedia archive containing all existing media objects and able to absorb any future formats. It comprises a text search engine, a Relational Database Management System, database and Adobe Acrobat 2.0. Acrobat converts existing PostScript files into ASCII-based Portable Document Format files, transferable over modem, electronic mail, local and wide area Networks. PDF files also enable the user to access documents regardless of file format, operating system or the software used to create the document. MediaSphere presently uses a Sybase relational database, but is set to run on any SQL based database and costs around UKP65,000 for eight users including a Sparcserver and the necessary networking software, shipping January 1995. Formed in 1993, Cascade had turnover of around $10m from systems integration, Open Pre-press Interface graphic management products ImageFlow and ReproFlow, and product management tracking products DataFlow and ViewFlow in its first
year. The company recently secured around UKP1.4m from Adobe Venture and Jarrold & Sons Ltd (CI No 2,487) and will use venture funding for the foreseeable future. However, Cascade plans to float on the American Stock Exchange. Its holding company Cascade Systems International comprises Cascade Systems Ltd and Cascade Systems Inc, is registered in the US which was chosen over the UK as it is easier to secure a listing there and gives the company a higher profile in its biggest potential market.
