Knowledge based information retrieval system

• US 6,182,062 B1
• Filed: 08/21/1996
• Issued: 01/30/2001
• Est. Priority Date: 03/26/1986
• Status: Expired due to Fees

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Hatakeyama, Atsushi, Cohn, David, Kiuchi, Itsuko, Fujisawa, Hiromichi
• Primary Examiner(s): Hafiz, Tariq R.
• Assistant Examiner(s): CHAVIS, JOHN Q
• Application Number:
  US08/663,840
• Time in Patent Office: 1,623 Days
• Field of Search: 707/1, 707/3
• US Class Current: 1/1
• CPC Class Codes: G06F 16/243 Natural language query formulation
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/232 Orthographic correction, e.g. spell checking or vowelisation
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching

Abstract

A multi-lingual information retrieval system which includes an information storage section for storing information, a data storage section for storing data for language conversion and for converting the stored information to strings of characters in natural language expressions in order to enable a person'"'"'s interpretation, an interface section for user'"'"'s interaction operations for information retrieval, a retrieval section for retrieving information from the information storage section in response to a request for a retrieval from the interface section, and an output section for outputting a retrieved result of the retrieval section. The data storage section coexistently stores the information therein in a plurality of different languages, thereby making possible to retrieve identical information by use of the plurality of different languages.

Claims

1. A multi-lingual information retrieval system comprising:

information storage means for storing a plurality of concepts each defining relations between a plurality of items of information;

language-expression converting data storage means for storing data to covert a retrieval condition expressed in any one of a plurality of different natural languages to an expression of internal data which is an interpretation of a natural language represented by a formula and storing data to convert information retrieved from said information storage means based on an expression of internal data to any one of the different natural languages, thereby making it possible to retrieve identical information for the same request for retrieval using any one of the different natural languages;

interface means for inputting a request for retrieval of information in one of the different natural languages and outputting a result of the retrieval;

retrieval condition converting means for receiving the request for retrieval from said interface means and converting the received request for retrieval to the expression of the internal data by using said language-expression converting data storage means; and

retrieval means for retrieving information stored on said information storage means according to the converted expression of internal data, converting the retrieved information into the one different natural language based on the data stored in said language-expression converting data storage means, and providing the converted retrieved information as the result of the retrieval,

wherein said data storage means stores data coupling a string of characters expressed in one of the different natural languages and a symbol designating the one different natural language.

2. A knowledge base system comprising:

input means for inputting a series of words;

processing means for processing said inputted series of words; and

output means for outputting a result of the processing operation;

wherein said processing means comprises;

interface means for inputting a request for retrieval of information, said retrieval request corresponding to said series of words and outputting a result of a retrieval,

information storage means for storing information including a knowledge base which stores knowledge of a concept network forming a conceptual tree defined by nodes representing concepts and links representing relations among said concepts,

data storage means for storing data for language conversion for converting the stored information to strings of language characters, and

retrieval means for receiving the request for retrieval from said interface means and retrieving information stored on said information storage means,

wherein said retrieval means comprises;

analysis means for identifying a relation in meaning among words existing in said series of words by retrieving said concept network,

wherein said data storage means coexistently stores information in a plurality of different languages, thereby making it possible to retrieve identical information by use of the plurality of different languages.

3. A knowledge base system comprising:

input means for inputting a series of words;

processing means for processing said inputted series of words; and

output means for outputting a result of the processing operation;

wherein said processing means comprises;

interface means for inputting a request for retrieval of information, said retrieval request corresponding to said series of words and outputting a result of a retrieval,

an information storage means for storing information including a knowledge base which stores knowledge of a concept network forming a conceptual tree defined by nodes representing concepts and links representing relations among said concepts,

language-expression converting data storage means for storing data to convert a retrieval condition expressed in a string of characters in a language corresponding to an expression of data,

retrieval condition converting means for receiving the request for retrieval of information input from said interface means and converting the received request for retrieval to the expression of internal data by using said language-expression converting data storage means, and

retrieval means for retrieving information stored on said information storage means in response to the converted expression of internal data,

wherein said retrieval means comprises;

analysis means for identifying a relation in meaning among words existing in said series of words by retrieving said concept network,

wherein said language-expression converting data storage means coexistently stores information in a plurality of different languages, thereby making it possible to retrieve identical information by use of the plurality of different languages.

4. A knowledge base system comprising:

input means for inputting a series of words;

processing means for processing said inputted series of words; and

output means for outputting a result of the processing operation;

wherein said processing means comprises;

a knowledge base storing knowledge of a concept network forming a conceptual tree defined by nodes representing concepts and links representing relations among said concepts, and

analysis means for identifying a relation in meaning among words existing in said series of words by retrieving said concept network.

5. A knowledge base system according to claim 4, wherein said knowledge base comprises:

a concept name table storing names of said concepts in a plurality of languages;

a subsumption relation table storing subsumption relations among said concepts;

a general-relation defining table storing definitions of relations existing among said concepts other than said subsumption relations; and

a relation table storing information indicating the existence of the relations among said concepts.

6. A knowledge base system according to claim 5, further comprising:

a grammar file storing syntax rules, wherein said analysis means includes lexical analysis means for effecting lexical analysis of an attribute of each word of said series of words; and

syntax analysis means for effecting syntactic analysis of a syntax of said series of words by consulting said grammar file, said knowledge base and lexical analysis result of said lexical analysis means.

7. A knowledge base system according to claim 6, further comprising:

a structured concept file storing concepts structured by limiting a meaning of a concept by another concept; and

a lexicon file storing kinds of words,

wherein said lexical analysis means effects analysis of an attribute of each word of said series of words by consulting said structured concept file, said lexicon file and said concept name table.

8. An information retrieval system comprising:

input means for inputting a series of words;

a knowledge base including a concept name table storing concept names in a plurality of languages and storing knowledge of a concept network forming a concept tree defined by nodes representing concepts and links representing relations among said concepts;

analysis means for identifying a relation in meaning among words existing in said series of words by consulting said concept network;

retrieval means for retrieving said knowledge based on an output of said analysis means; and

output means for outputting a result retrieved by said retrieval means.

9. A multi-lingual information retrieval system, comprising:

an information storage which stores a plurality of concepts each defining relations between a plurality of items of information;

a data storage which stores in corresponding relation to each item of information stored in said information storage, data including translations of the item of information in a plurality of different natural languages, thereby making it possible to retrieve identical information for the same request for retrieval using any of the different natural languages;

an interface which inputs a request for retrieval of information in one of the different natural languages and outputs a result of a retrieval; and

a retrieval unit which receives the request for retrieval from said interface means, retrieves information stored on said information storage according to the request for retrieval, translates the retrieved information into the one different natural language based on the data stored in said data storage, and provides the translated retrieved information as the result of the retrieval,

wherein said information storage expresses stored information by coupling nodes and links.

10. A system according to claim 9, wherein each of said nodes expresses a concept, and each of said links expresses a relation between nodes.

11. A multi-lingual information retrieval system comprising:

an information storage which stores a plurality of concepts each defining relations between a plurality of items of information;

language-expression converting data storage which stores data to covert a retrieval condition expressed in any one of a plurality of different natural languages to an expression of internal data which is an interpretation of a natural language represented by a formula and stores data to convert information retrieved from said information storage based on an expression of internal data to any one of the different natural languages, thereby making it possible to retrieve identical information for the same request for retrieval using any one of the different natural languages;

an interface which inputs a request for retrieval of information in one of the different natural languages and outputs a result of the retrieval;

a retrieval condition converting unit which receives the request for retrieval from said interface and converts the received request for retrieval to the expression of the internal data by using said language-expression converting data storage; and

a retrieval unit which retrieves information stored on said information storage according to the converted expression of internal data, converts the retrieved information into the one different natural language based on the data stored in said language-expression converting data storage, and provides the converted retrieved information as the result of the retrieval.

14. A system according to claim 11, wherein said data storage stores data coupling a string of characters expressed in one of the different natural languages and a symbol designating the one different natural language.

12. A multi-lingual information retrieval system comprising:

an information storage which stores a plurality of concepts each defining relations between a plurality of items of information;

language-expression converting data storage which stores data to covert a retrieval condition expressed in any one of a plurality of different natural languages to an expression of internal data and stores data to convert information retrieved from said information storage based on an expression of internal data to any one of the different natural languages, thereby making it possible to retrieve identical information for the same request for retrieval using any one of the different natural languages;

an interface which inputs a request for retrieval of information in one of the different natural languages and outputs a result of the retrieval;

a retrieval condition converting unit which receives the request for retrieval from said interface and converts the received request for retrieval to the expression of the internal data by using said language-expression converting data storage; and

a retrieval unit which retrieves information stored on said information storage according to the converted expression of internal data, converts the retrieved information into the one different natural language based on the data stored in said language-expression converting data storage, and provides the converted retrieved information as the result of the retrieval,

wherein said information storage expresses stored information by coupling nodes and links.

13. A system according to claim 12, wherein each of said nodes expresses a concept, and each of said links expresses a relation between nodes.

15. A multi-media information retrieval system, comprising:

a user interface which inputs from a user a series of words from a plurality of natural languages, and displays responses from said system and information input by said user;

memory means for storing character strings, wherein each character string represents a word in one of said natural languages, together with a data item identifying said one natural language;

another memory means for storing concept identifiers, wherein each concept identifier is associated with one of said character strings;

query analysis means which identifies a natural language to which said series of words belongs, and creates an internal query condition which is an interpretation of a natural language represented by a formula based on said series of words correspondingly;

information storage means for storing multi-media files, wherein each multi-media file is associated with at least one of said concept identifiers;

information search means for identifying concept identifiers that match said internal query condition; and

information retrieval means for retrieving multi-media files from said information storage means which are associated with said identified concept identifiers.