| Striving to disentangle the World Wide Web |
|
The World Wide Web is, at best, a mixed blessing. We all know that it contains a huge amount of information, but it is so poorly signposted that few are able to get at the details they require.
A group of computer scientists at Hitachi's advanced-research laboratory in Hatoyama-machi, Saitama prefecture, may have the solution. They hope to develop a system to filter, extract and translate information according to individual preferences. Steven Myers reports in Computing Japan that the system would be able to search any huge collection of documents, extract pertinent articles, highlight relevant sections, translate the information into the user's language and adjust future search criteria according to the documents the user selects. In addition, the user should be able to search for documents by graphics as well as by keyword. A search on one picture would then return documents containing similar or related pictures. Hitachi's advanced-research laboratory was founded in 1985, to commemorate the company's 75th anniversary. The facility's budget is around 4 billion yen a year - 1 percent of the company's 400 billion yen total. The laboratory emphasizes long-term, basic multidisciplinary research. This is reflected in the wide variety of projects under way. In addition to the work by computer scientists, projects range from radiation and electron-beam physics to biotechnology and material science. The laboratory hosts up to 15 foreign visiting researchers at any given time, and sends its own scientists to overseas universities and research institutions. The computer team combines research in artificial intelligence, probability theory, linguistics and computer architecture. Team members have already produced a document-retrieval system that can search for key documents rather than key words. The system can scan a document, assign it to one or more categories, then search through other documents from the same sategories in order to find related information. By using this type of classification scheme as a "preprocessing" mechanism, the search space is greatly reduced. This leads to much faster retrieval. Searches on articles related to popular topics falling into clearly defined categories produce better results than those on topics about which little has been written. Documents in the database are continuously recategorized according to what the user has previously chosen as being relevant to his or her interests. Team members have also achieved much in the area of using an initial command to predict subsequent commands for carrying out a certain task. This type of prediction method will be useful in developing the capability automatically to make suggestions to a user for accomplishing various tasks, based on that user's previous actions. The team has developed numerical techniques for examining English words with multitude meanings, and choosing the correct translation based on the context in which the word appears. When the system encounters the word "suit," for example, it first examines the surrounding terms and stores information about each work based on its position within the sentence. The appearance of words such as "claims" or "court" within the same sentence would signal to the system that the suit in question is a lawsuit rather than an item of clothing. Obviously, the whole system demands a large-memory personal computer with the power to carry out such computation-intensive tasks. The team is attempting to create a personal computer with 2Gb of main memory. The work is far from complete. With so many different areas of research integrated into a single project, it is difficult for team members to outline a detailed plan for the future. But their work is expected to last for another ten years. The message to people battling with the World Wide Web today is clear: don't hold your breath!
|
| See also Wiley's Corner for monthly reviews of the latest international business books |
asia's only dedicated daily ezine for the asian business, management, strategy & marketing professional
research articles news independent columnists business strategy market & street intelligence