Query transformation system and method enabling retrieval of multilingual web documents

• US 6,064,951 A
• Filed: 01/12/1998
• Issued: 05/16/2000
• Est. Priority Date: 12/16/1997
• Status: Expired due to Term

• Current Assignee: Knapp Investment Company Limited
• Original Assignee: Electronics and Telecommunications Research Institute
• Inventors: Choi, Sung-Kwon, Yuh, Sang-Hwa, Chae, Young-Soog, Kim, Young-Kil, Jung, Han-Min, Park, Dong-In, Kim, Tae-Wan, Sim, Chul-Min, Chang, Won
• Primary Examiner(s): Thomas, Joseph
• Application Number:
  US09/006,838
• Time in Patent Office: 855 Days
• Field of Search: 704/2, 704/3, 704/4, 704/5, 704/6, 704/7, 704/8, 704/9, 704/10, 704/1, 707/1, 707/2, 707/3, 707/4, 707/5, 707/6, 707/7, 707/104, 707/530, 707/531, 707/536, 707/501, 707/513, 379/88.05, 379/88.06, 379/88.17
• US Class Current: 704/8
• CPC Class Codes: G06F 16/3337 Translation of the query language, e.g. Chinese to English
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

A query transformation system and method capable of not only solving an ambiguousness of words involved in the transformation of queries from one language to another language, but also executing its processing independently of the processing of an information retrieval system used, so that it can be applied to a variety of information retrieval systems, thereby enabling the information retrieval system used to function as a multilingual information retrieval system. The system includes a translation generator for generating all possible translations of an input query consisting of a source language by reference to a translation dictionary, a semantic category verifier for receiving the generated translations from the translation generator, and eliminating translations having a low semantic similarity from the received translations, based on a semantic category tree, and a collocation information verifier for receiving the translations, which includes no translation having a low semantic similarity, from the semantic category verifier, and eliminating translations having no collocation from the received translations, based on word collocation information.

Claims

1. A query transformation system enabling retrieval of multilingual web documents comprising:

a query input unit for imputing a query consisting of a source language, the query input unit comprising a keyboard;

an input query memory unit for storing the query, consisting of the source language, received from the query input unit;

a transformation control unit for controlling the entire query transformation operation of the system;

a translation generating/filtering unit for generating translations of the input query and filtering unnecessary ones of the generated translations;

a translation knowledge unit stored with at least one electronic dictionary to be used for a transformation of queries and a variety of information;

a transformed query memory unit for storing the query transformed from the source language into a target language; and

a result output unit for outputting the result of the transformation in the form of the target language on a screen;

wherein the translation generating/filtering unit comprises;

a translation generator for generating all possible translations of the source language input query by reference to a translation dictionary;

a semantic category verifier for receiving the generated translations from the translation generator, a eliminating translations having a low semantic similarity from the received translations, based on a semantic category tree; and

a collocation information verifier for receiving the translations, which includes no translation having a low semantic similarity, from the semantic category verifier, and eliminating translations having no collocation from the received translations, based on word collocation information.

2. The query transformation system in accordance with claim 1, wherein the translation knowledge unit comprises:

a plurality of electronic translation dictionaries consulted when a word consisting of the source language is transformed into the target language;

a semantic category tree to which the semantic category verifier is referred; and

a collocation information dictionary to which the collocation information verifier is referred.

3. The query transformation system in accordance with claim 1, wherein the semantic category verifier checks a semantic similarity of each of the translations generated from the translation generator, using a special word dictionary.

4. The query transformation system in accordance with claim 1, wherein the source language is Korean language, and the target language is English language.

5. The query transformation system in accordance with claim 1, wherein the source language is English language, and the target language is Korean language.

6. The query transformation system in accordance with claim 1, wherein the source language is Korean language, and the target language is a language other than Korean and English languages.

7. A query transformation method enabling retrieval of multilingual web documents, comprising the step of:

generating a transformation start signal when a tool button on a screen associated with the start of a transformation is clicked;

starting a query transformation in response to the transformation start signal;

generating all possible translations of an input query from a user by reference to a translation dictionary;

determining whether or not translations are generated;

if there is no translation generated, informing the user of the fact that there is no translation generated, while if there are translations generated, executing a comparison processing for the generated translations, based on a semantic category tree, eliminating translations having a low semantic similarity, thereby eliminating unnecessary ones of the translations;

analyzing a collocation of the resultant translations by reference to a collocation information dictionary, thereby eliminating unnecessary ones of the analyzed translations;

determining whether or not there are translations left; and

if there are translations left, outputting the translations left as a transformed query on the screen, while if there is no translation left, recovering the generated translations, and outputting the recovered translations as a transformed query on the screen.