TRL
TOP PAGETokyo Research LaboratoryEmploymentProjectsRelated InformationIBM Research
Japanese page is here.

Site Outlining



Overview

The aim of the Site Outlining project is to research and develop new systems and technologies which allow users to analyze, retrieve, and visualize information from enormous collections of heterogeneous and time-varying data from multi-dimensional aspects. We can significantly enhance the usability of data on the internet by means of this technology.



Research items
  • Metadata extraction ... study of extraction of various types of metadata to help users understand the characteristics of and singularities in a collection of data.
  • Dynamic information ... study of time-varying information sources and their metadata.
  • Knowledge Visualization ... study of metadata visualization and metadata-dependent searching and navigation methods.
  • Mathematical models and algorithms ... development and implementation of new and efficient algorithms for ranking and retrieval of documents in very large, dynamic databases.
Systems/Products
  • Business Model Construction (BMC)

    BMC system collects various information from the internet, and supports construction of business models.

  • Digital News Library (DNL)

    DNL is a system for retrieving/managing an enormous collection of newspaper articles, which is updated on a daily basis. The collection grows by at least several hundreds of gigabytes annually. DNL is being used as a core technology in one of the biggest solutions in the Japanese newspaper industry.

  • Information Navigation System (INS)

    INS is a system for retrieving digital video contents through the use of information outlining technology developed by our group. We believe it will be adopted as one of the key technologies for digital broadcasting in the twenty-first century. A wide range of users can benefit from the versatile set of views which INS can provide for outlining video data.

  • Let'sSVD

    Let'sSVD is a program to compute the singular value decomposition (SVD) of very large, sparse matrices. The program uses dynamic data structures to minimize use of memory space.

Publications

Research home IBM home Order Privacy Legal Contact IBM
Last modified 11 Sep. 2000