DICTIONARY CREATION DEVICE, WORD GATHERING METHOD AND RECORDING MEDIUM

• US 20120303359A1
• Filed: 12/03/2010
• Published: 11/29/2012
• Est. Priority Date: 12/11/2009
• Status: Abandoned Application

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Mizuguchi, Hironori, Kusui, Dai, Kusumura, Yukitaka
• Application Number:
  US13/515,135
• Publication Number:
  US 20120303359A1
• Time in Patent Office: Days
• US Class Current: 704/10
• CPC Class Codes: G06F 16/353 into predefined classes
• CPC Class Codes: G06F 40/242 Dictionaries

Abstract

When gathering words through a dictionary growth process, a dictionary growth unit (102) stores information indicating through what process of input and output a word has been gathered in a gathering process memory unit (107). Then, a clustering unit (103) classifies the word that has been gathered by the dictionary growth process into clusters on the basis of information recorded in the gathering process memory unit (107). Next, a type determination unit (104) determines whether a word comprising a cluster is of the same type as a seed word or of a different type, for each cluster into which the word has been classified, on the basis of information recorded in the gather process memory unit (107). In addition, an output unit (105) associates information indicating the gathered word, the cluster to which the word belongs and whether the cluster is of the same type as the seed word or of a different type, and displays such.

Claims

1. A dictionary creation device, comprising:

an input/output process recording means for recording information indicating the process of inputting and outputting input words and output words output by the input words, in a dictionary growth process for gathering words by repeatedly accepting input of words, outputting words related to the input words from document data, adding to the input words words output until a prescribed condition is satisfied, and outputting words related to the input words from document data;

a cluster classifying means for classifying words that input word or output word becomes the same into same cluster among words gathered by the dictionary growth process based on information recorded in the input/output process recording means;

a similarity determining means for determining whether or not words in a cluster are words of the same type as input words for which input was initially received, for each cluster classified by the cluster classifying means, based on the number of turns required to output each word in the cluster from the input word, by referencing information recorded in the input/output process recording means; and

a gathered word output means for linking together and outputting words gathered by the dictionary growth process, clusters to which the words belong and information indicating whether or not the words comprising the cluster are words of the same type of the input words for which input was initially received.

2. The dictionary creation device of claim 1, further comprising a dictionary growth means for gathering words by repeatedly accepting input of words, outputting words related to the input words from document data, adding to the input words words output until a prescribed condition is satisfied, and outputting words related to the input words from document data.

3. The dictionary creation device of claim 1, wherein the input/output process recording means records information indicating the input/output process of input words and output words output by the input words, with input and output repeated multiple times.

4. The dictionary creation device of claim 1, wherein the cluster classifying means calculates a degree of unity between words indicating a value that becomes larger between words which have common words as inputs or between words that output common words in the above-described dictionary growth process, out of the words gathered in the dictionary growth process, from information recorded in the input/output process registration means, and classifies words into clusters based on the calculated degree of unity.

5. The dictionary creation device of claim 1, wherein the similarity determining means calculates an average value for words in a cluster of the minimum number of inputs/outputs for a word in the cluster to input/output an input word the input of which is initially received, for each cluster, based on information recorded in the input/output process recording means, and determines that words are of the same type when the calculated average value is not greater than a prescribed threshold value.

6. The dictionary creation device of any of claim 1, further comprising:

a word group memory means for classifying words gathered by the dictionary growth process into multiple word groups, for each type, and storing such; and

a word selecting means for selecting a prescribed number of words from one word group meeting prescribed conditions;

wherein the dictionary growth process is executed using the words selected by the word selecting means as input words; and

the similarity determining means determines whether or not words in a cluster are the same type of words as the input words selected by the word selecting means, for each cluster classified by the cluster classifying means, based on information recorded in the input/output process recording means.

7. The dictionary creation device of claim 6, further comprising a re-execution means for recording words gathered by the dictionary growth process in the word group memory means, based on results determined by the similarity determining means, and instructing the word selecting means to select words when there is a word group satisfying prescribed conditions out of the recorded word groups;

wherein when recording the gathered words in the word group memory means, when the cluster to which the gathered words belong is the same type of word as the word selected by the word selecting means, the re-execution means records the gathered words in the same word group as the selected word, and when the word is a different type and has already been stored in the word group memory means, records the gathered words in the same word group as the stored words, and when the word is of a different type and has not yet been stored in the word group memory means, records the gathered word in a new word group.

8. The dictionary creation device of claim 6, further comprising a degree-of-unity memory means for storing a degree of unity between words indicating a value that becomes larger between words which have common words as inputs or between words that output common words in the above-described dictionary growth process, and that is computed from information recorded in the input/output process recording means;

wherein the word selecting means selects a prescribed number of words based on the degree of unity between words in the one word group.

9. The dictionary creation device of claim 8, wherein the word selecting means selects a prescribed number of words based at least on condition information in which at least the ratio for selecting words in decreasing order of degree or unity or the ratio of selecting words in increasing order of degree of unity is preset.

10. A word gathering method, comprising:

an input/output process recording step for recording information indicating the process of inputting and outputting input words and output words output by the input words, in a dictionary growth process for gathering words by repeatedly accepting input of words, outputting words related to the input words from document data, adding to the input words words output until a prescribed condition is satisfied, and outputting words related to the input words from document data;

a cluster classifying step for classifying words that input word or output word becomes the same into same cluster among words gathered by the dictionary growth process based on information recorded in the input/output process recording step;

a similarity determining step for determining whether or not words in a cluster are words of the same type as input words for which input was initially received, for each cluster classified by the cluster classifying step, based on the number of turns required to output each word in the cluster from the input word, by referencing information recorded in the input/output process recording step; and

a gathered word output step for linking together and outputting words gathered by the dictionary growth process, clusters to which the words belong and information indicating whether or not the words comprising the cluster are words of the same type of the input words for which input was initially received.

11. A computer-readable recording medium on which is recorded a program that causes a computer to function as:

an input/output process recording means for recording information indicating the process of inputting and outputting input words and output words output by the input words, in a dictionary growth process for gathering words by repeatedly accepting input of words, outputting words related to the input words from document data, adding to the input words words output until a prescribed condition is satisfied, and outputting words related to the input words from document data;

a cluster classifying means for classifying words that input word or output word becomes the same into same cluster among words gathered by the dictionary growth process based on information recorded in the input/output process recording means;

a similarity determining means for determining whether or not words in a cluster are words of the same type as input words for which input was initially received, for each cluster classified by the cluster classifying means, based on the number of turns required to output each word in the cluster from the input word, by referencing information recorded in the input/output process recording means; and

a gathered word output means for linking together and outputting words gathered by the dictionary growth process, clusters to which the words belong and information indicating whether or not the words comprising the cluster are words of the same type of the input words for which input was initially received.