Browse Prior Art Database

Method of Japanese Address Entry

IP.com Disclosure Number: IPCOM000114896D
Original Publication Date: 1995-Feb-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 2 page(s) / 55K

Publishing Venue

IBM

Related People

Itoh, N: AUTHOR

Abstract

This article describes a method for entering Japanese addresses by using on-line character recognition and a Japanese address dictionary.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 65% of the total text.

Method of Japanese Address Entry

      This article describes a method for entering Japanese addresses
by using on-line character recognition and a Japanese address
dictionary.

      A full string to be entered is predicted from a partial input
by consulting a address dictionary, and listed in menu.  A user can
enter an address by writing just small part of all characters and
selecting one of the listed candidates.

      A Japanese address is identified by unique compound of address
words and block numbers.  Each word has a certain position in the
hierarchy of an address.  In addition lower-level address word
follows upper-level word, which is an important context for selecting
address words at the next level.

      Address words that match the current context (upper-level word
or zip code) are selected from a hierarchically structured dictionary
and each word is compared with the recognition result of partial
input based on the following likelihood L (Figure).

      Let an address word be W = t[1]t[2], ...,t[n], and let the jth
candidate character in the ith column be C[i,j].  L is expressed by
the following formula:
  L = summation of d(i,j) *  (S-i+j),
  where d(i,j) = 1 (t[i]= C[i,j]), or 0 (otherwise).

      S designates a constant value, which is given to the word whose
character coincides with the first recognition candidate in the
corresponding position.  Candidate words are sorted in the decreasing
order of L.

      Selecti...