Method and Apparatus for Expanding Similar Character Strings

1. A similar character string expanding method, comprising the steps of: preparing one or more similar sound tables according to a plurality of pronunciation similarity criteria, each of the similar sound tables indicating one or more similar sounds similar in pronunciation to one key sound for each of a plurality of key sounds; preparing a regular expression and an applicable range according to each of the pronunciation similarity criteria, each of the regular expressions indicating one or more strings of derivation types, each of derivation types indicating how each of a plurality of derived sounds placed at positions of an expanded pronunciation character string is derived from a remarked sound placed at a remarked position of a remarked pronunciation character string, and the number of sounds in the remarked pronunciation character string being within a corresponding applicable range; receiving a question pronunciation character string; producing a set of derivation elements for the question pronunciation character string according to the similar sound tables and the regular expressions corresponding to the pronunciation similarity criteria, each of the derivation elements in the set being composed of a derived sound derived from one of a plurality of sounds in the question pronunciation character string, a sound position of the derived sound in each of one or more expanded pronunciation character strings expanded from the question pronunciation character string and a list of one or more derivation types respectively used to derive the derived sound at the sound position of each expanded pronunciation character string, and the derivation types being extracted from the strings of derivation types indicated by each of the regular expressions prepared; producing a set of a plurality of regular expression elements respectively composed of one regular expression and an applicable sound number, the applicable sound number indicating the number of sounds in each of one or more expanded pronunciation character strings expanded from the question pronunciation character string; producing a finite state automaton indicating one or more applicable strings of derivation types from the set of regular expression elements, each applicable string of derivation types being indicated by one regular expression of one regular expression element on condition that the number of sounds in an expanded pronunciation character string expanded from the question pronunciation character string according to the applicable string of derivation types agrees with one applicable sound number of the regular expression element; and outputting the set of derivation elements and the finite state automaton to expand one or more expanded pronunciation character strings as one or more similar pronunciation character strings similar in pronunciation to the question pronunciation character string according to the set of derivation elements and the finite state automaton.

2. A similar character string expanding method according to claim 1, in which the step of outputting the set of derivation elements and the finite state automaton comprises the steps of: selecting a plurality of derived types from the set of derivation elements in increasing order of sound positions corresponding to the derived types on condition that one applicable string indicated by the finite state automaton is formed by the derived types; and serially arranging a plurality of derived sounds of the set of derivation elements corresponding to the derived types in correspondence order to form one expanded pronunciation character string corresponding to one applicable string indicated by the finite state automaton.

3. A similar character string expanding method according to claim 1, in which the step of preparing a regular expression and an applicable range includes the step of: making up each of the derivation types of a first symbol identifying one of the similar sound tables, a second symbol specifying a positional difference between a remarked position of one remarked sound placed in the remarked pronunciation character string and a derived position of one derived sound which is looked up in the similar sound table identified by the first symbol by using the remarked sound as one key sound and is placed in the expanded pronunciation character string, and a third symbol indicating a derivation constraint for the remarked sound.

4. A similar character string expanding method according to claim 3, in which the step of producing a set of derivation elements comprises the steps of: extracting a derivation type from a string of derivation types indicated by one of the regular expressions; selecting a remarked sound from among a plurality of sounds in the question pronunciation character string; judging whether or not the remarked sound satisfies a derivation constraint of the derivation type; determining a sound position according to a remarked position of the remarked sound in the question pronunciation character string and a positional difference of the derivation type in cases where the remarked sound satisfies the derivation constraint; looking up one or more similar sounds in a similar sound table identified by an identifying symbol of the derivation type by using the remarked sound as a key sound; determining each of the similar sounds as a derived sound in cases where the remarked sound satisfies the derivation constraint; producing a derivation element composed of the sound position, the derivation type and the derived sound; and registering the derivation element in the set of derivation elements.

5. A similar character string expanding method according to claim 3, in which the step of making up each of the derivation types include the step of: judging whether or not a remarked sound in the question pronunciation character string satisfies a derivation constraint of a derivation type according to a plurality of sounds in the question pronunciation character string regardless of any expanded pronunciation character string.

6. A similar character string expanding method according to claim 1, in which the step of preparing one or more similar sound tables includes the step of; preparing a same sound table indicating a sound which is the same as one key sound as a similar sound for each of the key sounds; preparing a weak sound table indicating one or more weak sounds which is the same as or similar to one key sound as one or more similar sounds for each of the key sounds; preparing a weak different sound table indicating one or more weak sounds similar in pronunciation to one key sound as one or more similar sounds for each of the key sounds; and preparing a same-line sound table indicating a plurality of sounds, which are placed in the same line as that of one key sound, as a plurality of similar sounds for each of the key sounds.

7. A similar character string expanding method according to claim 1, in which the step of producing a set of derivation elements includes the steps of: producing various predetermined derivation types required in a probability to determine a plurality of derived sounds from a plurality of remarked sounds of various remarked pronunciation character strings; and selecting the derivation types in the set of derivation elements from among the predetermined derivation types.

8. A similar character string expanding method according to claim 1, in which the step of producing a finite state automaton includes the steps of: producing a plurality of derivation type list numbers; indicating one or more derivation type lists respectively composed of the same group of derivation types in the set of derivation elements by the same derivation type list number; and indicating each of the applicable strings of derivation types indicated by the finite state automaton by a string of derivation type list numbers.

9. A similar character string expanding method according to claim 1, in which the step of producing a set of derivation elements comprises the steps of: producing a plurality of derivation elements in the order in which the number of the sound position is increased; and adding the derivation elements one after another in a set.

10. A similar character string expanding method according to claim 1, in which the step of producing a set of derivation elements comprises the steps of: indicating a remarked position of a remarked sound in the question pronunciation character string by a relative position to a sound position of a derived sound in an expanded pronunciation character string; determining the sound position of the derived sound; determining a derivation element relating to the sound position of the derived sound; and adding the derivation elements one after another in a set.

11. A similar character string expanding method according to claim 1, in which the step of producing a set of a plurality of regular expression elements comprises the steps of: selecting one or more particular regular expressions from the regular expressions, the number of sounds in the question pronunciation character string being within each of one or more applicable ranges corresponding to the particular regular expressions; determining each of the applicable sound numbers according to one corresponding particular regular expression and the number of sounds in the question pronunciation character string; and producing the regular expression elements respectively composed of one particular regular expression and one applicable sound number corresponding to the particular regular expression.

12. A similar character string expanding apparatus, comprising: similar sound table storing means for storing one or more similar sound tables prepared according to a plurality of pronunciation similarity criteria, each of the similar sound tables indicating one or more similar sounds similar in pronunciation to one key sound for each of a plurality of key sounds; regular expression storing means for storing a plurality of sets respectively composed of a regular expression and an applicable range prepared according to each of the pronunciation similarity criteria, each of the regular expressions indicating one or more strings of derivation types, each of derivation types indicating how each of a plurality of derived sounds placed at positions of an expanded pronunciation character string is derived from a remarked sound placed at a remarked position of a remarked pronunciation character string, and the number of sounds in the remarked pronunciation character string being within a corresponding applicable range; question pronunciation character string receiving means for receiving a question pronunciation character string; derivation element producing means for producing a set of derivation elements for the question pronunciation character string received by the derivation element producing means according to the similar sound tables stored in the similar sound table storing means and the regular expressions stored in the regular expression storing means, producing a set of regular expression elements from the regular expressions and the applicable ranges stored in the regular expression storing means and outputting the set of derivation elements, each of the derivation elements in the set being composed of a derived sound derived from one of a plurality of sounds in the question pronunciation character string, a sound position of the derived sound in each of one or more expanded pronunciation character strings expanded from the question pronunciation character string and a list of one or more derivation types respectively used to derive the derived sound at the sound position of each expanded pronunciation character string, the derivation types being extracted from the strings of derivation types indicated by the regular expressions prepared, each of the regular expression elements being composed of one regular expression and an applicable sound number and being produced from one regular expression and one applicable range prepared according to one of the pronunciation similarity criteria and the number of sounds in the question pronunciation character string, and the applicable sound number indicating the number of sounds in each of one or more expanded pronunciation character strings expanded from the question pronunciation character string; and finite state automaton producing means for producing a finite state automaton indicating one or more applicable strings of derivation types from the set of regular expression elements produced by the derivation element producing means and outputting the finite state automaton to expand one or more expanded pronunciation character strings as one or more similar pronunciation character strings similar in pronunciation to the question pronunciation character string according to the set of derivation elements output by the derivation element producing means and the finite state automaton, each applicable string of derivation types being indicated by one regular expression of one regular expression element on condition that the number of sounds in an expanded pronunciation character string expanded from the question pronunciation character string according to the applicable string of derivation types agrees with one applicable sound number of the regular expression element.

13. A similar character string expanding apparatus according to claim 12, in which each of the derivation types stored in the regular expression storing means is composed of a first symbol identifying one of the similar sound tables, a second symbol specifying a positional difference between a remarked position of one remarked sound placed in the remarked pronunciation character string and a derived position of one derived sound which is looked up in the similar sound table identified by the first symbol by using the remarked sound as one key sound and is placed in the expanded pronunciation character string and a third symbol indicating a derivation constraint for the remarked sound.

14. A similar character string expanding apparatus according to claim 12, in which the similar sound tables stored in the similar sound table storing means are classified into a same sound table indicating a sound which is the same as one key sound as a similar sound for each of the key sounds, a weak sound table indicating one or more weak sounds which is the same as or similar to one key sound as one or more similar sounds for each of the key sounds, a weak different sound table indicating one or more weak sounds similar in pronunciation to one key sound as one or more similar sounds for each of the key sounds, and a same-line sound table indicating a plurality of sounds, which are placed in the same line as that of one key sound, as a plurality of similar sounds for each of the key sounds.

15. A similar character string expanding method according to claim 13, in which the judgement whether or not a remarked sound in the question pronunciation character string satisfies one derivation constraint of one derivation type is performed according to a plurality of sounds in the question pronunciation character string regardless of any expanded pronunciation character string.

16. A similar character string expanding apparatus according to claim 12, further comprising: derivation type producing means for producing a plurality of derivation types, each of the derivation types in the set of derivation elements produced by the derivation element producing means being supplied from the derivation type producing means.

17. A similar character string expanding apparatus according to claim 12, further comprising: derivation type list number producing means for producing a plurality of derivation type list numbers to indicate one or more derivation type lists respectively composed of the same group of derivation types in the set of derivation elements produced by the derivation element producing means by the same derivation type list number, each of the applicable strings of derivation types indicated by the finite state automaton produced by the finite state automaton producing means being indicated by a string of derivation type list numbers.

18. A similar character string expanding apparatus according to claim 12, in which the derivation elements in the set of derivation elements are produced by the derivation element producing means in the order in which the number of the sound position is increased.

19. A similar character string expanding apparatus according to claim 12, in which a remarked position of a remarked sound in the question pronunciation character string is indicated by a relative position to a sound position of a derived sound in an expanded pronunciation character string by the derivation element producing means to determine a derivation element relating to the sound position.

20. A character string retrieving method for retrieving one or more registered character strings similar in pronunciation to a question pronunciation character string according to one or more pronunciation similarity criteria, comprising the steps of: preparing a set of derivation elements, each of the derivation elements in the set being composed of a derived sound derived from one of a plurality of sounds in the question pronunciation character string, a sound position of the derived sound in each of the registered character strings and a derivation type indicating both the derivation of the derived sound from a remarked sound placed at a remarked position in the question pronunciation character string and a positional difference between the sound position and the remarked position; preparing a finite state automaton indicating one or more groups of applicable strings of derivation types and one or more applicable sound numbers, each of the applicable strings of derivation types in one group being obtained from one of the pronunciation similarity criteria, and each of the applicable sound numbers denoting the limitation for one or more character strings derived from the applicable strings of derivation types in one group; producing a plurality of possible strings of derivation types, respectively obtained by arranging one or more derivation types of one or more derivation elements extracted from the set of derivation elements in order of the sound position, as all candidates for the registered character strings; judging whether or not each of the possible strings of derivation types agrees with one of the applicable strings of derivation types in the groups; extracting one or more possible strings of derivation types respectively agreeing with one of the applicable strings of derivation types as one or more probable strings of derivation types; judging whether or not each of probable character strings, respectively obtained by arranging one or more derived sounds relating to one or more derivation types of one probable string in order of the sound position, satisfies one corresponding applicable sound number; extracting one or more probable strings of derivation types, which correspond to one or more probable character strings respectively satisfying the corresponding applicable sound number, as one or more definite strings of derivation types; arranging one or more derived sounds corresponding to one or more derivation types of each definite string in order of the sound position to produce one or more definite character strings for the definite strings; and retrieving one or more character strings which each agree with one of the definite character strings and are registered in a data base, from the data base as the registered character strings.

21. A character string retrieving method according to claim 20 in which the step of retrieving one or more character strings comprises the steps of: preparing the data base in which a plurality of character strings respectively corresponding to a character string number; annexing a pronunciation number to each of the character strings stored in the data base in a retrieval structure form; relating each of the pronunciation numbers to one or more character string numbers; selecting one or more particular pronunciation numbers annexed to one or more character strings, which each agree with one of the definite character strings and are stored in the data base, from the pronunciation numbers; selecting one or more particular character string numbers relating to the particular pronunciation numbers; setting a plurality of particular character string numbers overlapped with each other to one character string number; and outputting the particular character string numbers which are not overlapped with each other and correspond to the definite character strings.

22. A character string retrieving method according to claim 20 in which the step of producing a plurality of possible strings comprises the steps of: selecting one particular derivation element from one or more derivation elements corresponding to each of the sound positions; arranging one or more particular derivation types of one or more particular derivation elements while adding the particular derivation types one by one in order of the sound position until one particular derivation type corresponding to a maximum sound position is arranged; producing a possible string of derivation types each time one particular derivation type is arranged; replacing a final particular derivation type, which is finally arranged and corresponds to the maximum sound position, with another derivation type corresponding to the maximum sound position to produce another possible string of derivation types each time the final particular derivation type is replaced with another derivation type; reducing the maximum sound position by one each time all derivation types corresponding to the maximum sound position are arranged to produce one or more possible strings of derivation types; adding another particular derivation type corresponding to the maximum sound position newly defined to produce another possible string of derivation types; and stopping the production of the possible strings of derivation types when all derivation types corresponding to a first sound position are arranged to produce one or more possible strings of derivation types.

23. A character string retrieving apparatus for retrieving one or more registered character strings similar in pronunciation to a question pronunciation character string according to one or more pronunciation similarity criteria, comprising: derivation element set storing means for storing a set of derivation elements, each of the derivation elements in the set being composed of a derived sound derived from one of a plurality of sounds in the question pronunciation character string, a sound position of the derived sound in each of the registered character strings and a derivation type indicating both the derivation of the derived sound from a remarked sound placed at a remarked position in the question pronunciation character string and a positional difference between the sound position and the remarked position; finite state automaton storing means for storing a finite state automaton indicating one or more groups of applicable strings of derivation types and one or more applicable sound numbers, each of the applicable strings of derivation types in one group being obtained from one of the pronunciation similarity criteria, and each of the applicable sound numbers denoting the limitation for one or more character strings derived from the applicable strings of derivation types in one group; derivation type string producing means for producing a plurality of possible strings of derivation types, respectively obtained by arranging one or more derivation types of one or more derivation elements extracted from the set of derivation elements stored in the derivation element set storing means in order of the sound position, as all candidates for the registered character strings; derivation type string judging means for judging whether or not each of the possible strings of derivation types produced by the derivation type string producing means agrees with one of the applicable strings of derivation types in the groups which are indicated by the finite state automaton stored in the finite state automaton storing means, extracting one or more possible strings of derivation types respectively agreeing with one of the applicable strings of derivation types as one or more probable strings of derivation types, judging whether or not each of probable character strings, respectively obtained by arranging one or more derived sounds relating to one or more derivation types of one probable string in order of the sound position, satisfies one corresponding applicable sound number which are indicated by the finite state automaton stored in the finite state automaton storing means, and extracting one or more probable strings of derivation types, which correspond to one or more probable character strings respectively satisfying the corresponding applicable sound number, as one or more definite strings of derivation types; and character string retrieving means for arranging one or more derived sounds corresponding to one or more derivation types of each definite string extracted by the derivation type string judging means in order of the sound position to produce one or more definite character strings for the definite strings and retrieving one or more character strings which each agree with one of the definite character strings and are registered in a data base, from the data base as the registered character strings.

24. A character string retrieving apparatus according to claim 23 in which the character string retrieving means comprises: pronunciation index storing means for storing a pronunciation index in which a pronunciation number is annexed to each of the character strings stored in the data base in a retrieval structure form; character string number list storing means for storing a character string number list in which a plurality of character string numbers, corresponding to the character strings stored in the data base, are made relation to a plurality of pronunciation numbers; character string number retrieving means for retrieving one or more particular pronunciation numbers annexed to the definite character strings from the pronunciation index stored in the pronunciation index storing means and retrieving one or more particular character string numbers relating to the particular pronunciation numbers from the character string number list stored in the character string number list storing means; and overlapped character string number removing means for removing a plurality of particular character string numbers overlapped with each other by setting the particular character string numbers overlapped with each other to one particular character string number and outputting the particular character string numbers which are not overlapped with each other and correspond to the definite character strings.

25. A similar character string expanding and retrieving method for retrieving one or more registered character strings similar in pronunciation to a question pronunciation character string according to one or more pronunciation similarity criteria, comprising the steps of: preparing one or more similar sound tables according to the pronunciation similarity criteria, each of the similar sound tables indicating one or more similar sounds similar in pronunciation to one key sound for each of a plurality of key sounds; preparing a regular expression and an applicable range according to each of the pronunciation similarity criteria, each of the regular expressions indicating one or more strings of derivation types, each of derivation types indicating how each of a plurality of derived sounds placed at positions of an expanded pronunciation character string is derived from a remarked sound placed at a remarked position of a remarked pronunciation character string, and the number of sounds in the remarked pronunciation character string being within a corresponding applicable range; receiving the question pronunciation character string; producing a set of derivation elements for the question pronunciation character string according to the similar sound tables, the applicable ranges and the regular expressions corresponding to the pronunciation similarity criteria, each of the derivation elements in the set being composed of a derived sound derived from one of a plurality of sounds in the question pronunciation character string, a sound position of the derived sound in each of one or more expanded pronunciation character strings expanded from the question pronunciation character string and a derivation type used to derive the derived sound at the sound position of each expanded pronunciation character string, and the derivation types of the derivation elements being extracted from the strings of derivation types indicated by each of the regular expressions prepared; selecting one or more particular regular expressions from the regular expressions, the number of sounds in the question pronunciation character string being within each of one or more applicable ranges corresponding to the particular regular expressions; determining an applicable sound number indicating the number of sounds in each of one or more expanded pronunciation character strings expanded from the question pronunciation character string according to one particular regular expression by using the question pronunciation character string and the particular regular expression; producing a group of a plurality of regular expression elements respectively composed of one particular regular expression and one applicable sound number corresponding to the particular regular expression; producing a finite state automaton indicating one or more groups of applicable strings of derivation types and one or more applicable sound numbers from the group of regular expression elements, each group of applicable strings of derivation types being indicated according to one particular regular expression of one regular expression element, and each of the applicable sound numbers denoting the limitation for one or more character strings derived from the applicable strings of derivation types in one corresponding group; producing a plurality of possible strings of derivation types, respectively obtained by arranging one or more derivation types of one or more derivation elements extracted from the set of derivation elements in order of the sound position; judging whether or not each of the possible strings of derivation types agrees with one of the applicable strings of derivation types in the groups; extracting one or more possible strings of derivation types respectively agreeing with one of the applicable strings of derivation types as one or more probable strings of derivation types; judging whether or not each of probable character strings, respectively obtained by arranging one or more derived sounds relating to one or more derivation types of one probable string in order of the sound position, satisfies one corresponding applicable sound number; extracting one or more probable strings of derivation types, which correspond to one or more probable character strings respectively satisfying the corresponding applicable sound number, as one or more definite strings of derivation types; arranging one or more derived sounds corresponding to one or more derivation types of each definite string in order of the sound position to produce one or more definite character strings for the definite strings; and retrieving one or more character strings which each agree with one of the definite character strings and are registered in a data base, from the data base as the registered character strings.

26. A similar character string expanding and retrieving method according to claim 25, in which the step of preparing a regular expression and an applicable range includes the step of: making up each of the derivation types of a first symbol identifying one of the similar sound tables, a second symbol specifying a positional difference between a remarked position of one remarked sound placed in the remarked pronunciation character string and a derived position of one derived sound which is looked up in the similar sound table identified by the first symbol by using the remarked sound as one key sound and is placed in the expanded pronunciation character string, and a third symbol indicating a derivation constraint for the remarked sound.

27. A similar character string expanding method according to claim 26, in which the step of producing a set of derivation elements comprises the steps of: extracting a derivation type from a string of derivation types indicated by one of the regular expressions; selecting a remarked sound from among a plurality of sounds in the question pronunciation character string; judging whether or not the remarked sound satisfies a derivation constraint of the derivation type; determining a sound position according to a remarked position of the remarked sound in the question pronunciation character string and a positional difference of the derivation type in cases where the remarked sound satisfies the derivation constraint; looking up one or more similar sounds in a similar sound table identified by an identifying symbol of the derivation type by using the remarked sound as a key sound; determining each of the similar sounds as a derived sound in cases where the remarked sound satisfies the derivation constraint; producing a derivation element composed of the sound position, the derivation type and the derived sound; and registering the derivation element in the set of derivation elements.

28. A similar character string expanding and retrieving method according to claim 26, in which the step of making up each of the derivation types include the step of: judging whether or not a remarked sound in the question pronunciation character string satisfies a derivation constraint of a derivation type according to a plurality of sounds in the question pronunciation character string regardless of any expanded pronunciation character string.

29. A similar character string expanding and retrieving method according to claim 25, in which the step of preparing one or more similar sound tables includes the steps of; preparing a same sound table indicating a sound which is the same as one key sound as a similar sound for each of the key sounds; preparing a weak sound table indicating one or more weak sounds which is the same as or similar to one key sound as one or more similar sounds for each of the key sounds; preparing a weak different sound table indicating one or more weak sounds similar in pronunciation to one key sound as one or more similar sounds for each of the key sounds; and preparing a same-line sound table indicating a plurality of sounds, which are placed in the same line as that of one key sound, as a plurality of similar sounds for each of the key sounds.

30. A similar character string expanding and retrieving method according to claim 25 in which the step of retrieving one or more character strings comprises the steps of: preparing the data base in which a plurality of character strings respectively corresponding to a character string number; annexing a pronunciation number to each of the character strings stored in the data base in a retrieval structure form; relating each of the pronunciation numbers to one or more character string numbers; selecting one or more particular pronunciation numbers annexed to one or more character strings, which each agree with one of the definite character strings and are stored in the data base, from the pronunciation numbers; selecting one or more particular character string numbers relating to the particular pronunciation numbers; setting a plurality of particular character string numbers overlapped with each other to one character string number; and outputting the particular character string numbers which are not overlapped with each other and correspond to the definite character strings.

31. A similar character string expanding and retrieving method according to claim 25 in which the step of producing a plurality of possible strings comprises the steps of: selecting one particular derivation element from one or more derivation elements corresponding to each of the sound positions; arranging one or more particular derivation types of one or more particular derivation elements while adding the particular derivation types one by one in order of the sound position until one particular derivation type corresponding to a maximum sound position is arranged; producing a possible string of derivation types each time one particular derivation type is arranged; replacing a final particular derivation type, which is finally arranged and corresponds to the maximum sound position, with another derivation type corresponding to the maximum sound position to produce another possible string of derivation types each time the final particular derivation type is replaced with another derivation type; reducing the maximum sound position by one each time all derivation types corresponding to the maximum sound position are arranged to produce one or more possible strings of derivation types; adding another particular derivation type corresponding to the maximum sound position newly defined to produce another possible string of derivation types; and stopping the production of the possible strings of derivation types when all derivation types corresponding to a first sound position are arranged to produce one or more possible strings of derivation types.

32. A similar character string expanding and retrieving apparatus for retrieving one or more registered character strings similar in pronunciation to a question pronunciation character string according to one or more pronunciation similarity criteria, comprising: similar sound table storing means for storing one or more similar sound tables prepared according to a plurality of pronunciation similarity criteria, each of the similar sound tables indicating one or more similar sounds similar in pronunciation to one key sound for each of a plurality of key sounds; regular expression storing means for storing a plurality of sets respectively composed of a regular expression and an applicable range prepared according to each of the pronunciation similarity criteria, each of the regular expressions indicating one or more strings of derivation types, each of derivation types indicating how each of a plurality of derived sounds placed at positions of an expanded pronunciation character string is derived from a remarked sound placed at a remarked position of a remarked pronunciation character string, and the number of sounds in the remarked pronunciation character string being within a corresponding applicable range; question pronunciation character string receiving means for receiving a question pronunciation character string; derivation element producing means for producing a set of derivation elements for the question pronunciation character string received by the derivation element producing means according to the similar sound tables stored in the similar sound table storing means and the regular expressions stored in the regular expression storing means, selecting one or more particular regular expressions from the regular expressions stored in the regular expression storing means on condition that the number of sounds in the question pronunciation character string is within each of one or more applicable ranges corresponding to the particular regular expressions, determining an applicable sound number indicating the number of sounds in each of one or more expanded pronunciation character strings expanded from the question pronunciation character string received according to one particular regular expression by using the question pronunciation character string received by the question pronunciation character string receiving means and the particular regular expression, producing a group of a plurality of regular expression elements respectively composed of one particular regular expression and one applicable sound number corresponding to the particular regular expression, each of the derivation elements in the set being composed of a derived sound derived from one of a plurality of sounds in the question pronunciation character string, a sound position of the derived sound in each of one or more expanded pronunciation character strings expanded from the question pronunciation character string and a derivation type used to derive the derived sound at the sound position of each expanded pronunciation character string, and the derivation types of the derivation elements being extracted from the strings of derivation types indicated by the regular expressions; finite state automaton producing means for producing a finite state automaton indicating one or more applicable strings of derivation types from the set of regular expression elements produced by the derivation element producing means, each applicable string of derivation types being indicated by one particular regular expression of one regular expression element on condition that the number of sounds in an expanded pronunciation character string expanded from the question pronunciation character string according to the applicable string of derivation types agrees with one applicable sound number of the regular expression element; derivation type string producing means for producing a plurality of possible strings of derivation types, respectively obtained by arranging one or more derivation types of one or more derivation elements extracted from the set of derivation elements produced by the derivation element producing means in order of the sound position; derivation type string judging means for judging whether or not each of the possible strings of derivation types produced by the derivation type string producing means agrees with one of the applicable strings of derivation types in the groups which are indicated by the finite state automaton produced by the finite state automaton producing means, extracting one or more possible strings of derivation types respectively agreeing with one of the applicable strings of derivation types as one or more probable strings of derivation types, judging whether or not each of probable character strings, respectively obtained by arranging one or more derived sounds relating to one or more derivation types of one probable string in order of the sound position, satisfies one corresponding applicable sound number which are indicated by the finite state automaton stored in the finite state automaton storing means, and extracting one or more probable strings of derivation types, which correspond to one or more probable character strings respectively satisfying the corresponding applicable sound number, as one or more definite strings of derivation types; and character string retrieving means for arranging one or more derived sounds corresponding to one or more derivation types of each definite string extracted by the derivation type string judging means in order of the sound position to produce one or more definite character strings for the definite strings and retrieving one or more character strings which each agree with one of the definite character strings and are registered in a data base, from the data base as the registered character strings.

33. A similar character string expanding and retrieving apparatus according to claim 32, in which each of the derivation types stored in the regular expression storing means is composed of a first symbol identifying one of the similar sound tables, a second symbol specifying a positional difference between a remarked position of one remarked sound placed in the remarked pronunciation character string and a derived position of one derived sound which is looked up in the similar sound table identified by the first symbol by using the remarked sound as one key sound and is placed in the expanded pronunciation character string, and a third symbol indicating a derivation constraint for the remarked sound.

34. A similar character string expanding and retrieving apparatus according to claim 33, in which the judgement whether or not a remarked sound in the question pronunciation character string satisfies one derivation constraint of one derivation type is performed according to a plurality of sounds in the question pronunciation character string regardless of any expanded pronunciation character string.

35. A similar character string expanding and retrieving apparatus according to claim 32, in which the similar sound tables stored in the similar sound table storing means are classified into a same sound table indicating a sound which is the same as one key sound as a similar sound for each of the key sounds, a weak sound table indicating one or more weak sounds which is the same as or similar to one key sound as one or more similar sounds for each of the key sounds, a weak different sound table indicating one or more weak sounds similar in pronunciation to one key sound as one or more similar sounds for each of the key sounds, and a same-line sound table indicating a plurality of sounds, which are placed in the same line as that of one key sound, as a plurality of similar sounds for each of the key sounds.

36. A similar character string expanding and retrieving apparatus according to claim 32, in which the character string retrieving means comprises: pronunciation index storing means for storing a pronunciation index in which a pronunciation number is annexed to each of the character strings stored in the data base in a retrieval structure form; character string number list storing means for storing a character string number list in which a plurality of character string numbers, corresponding to the character strings stored in the data base, are made relation to a plurality of pronunciation numbers; character string number retrieving means for retrieving one or more particular pronunciation numbers annexed to the definite character strings from the pronunciation index stored in the pronunciation index storing means and retrieving one or more particular character string numbers relating to the particular pronunciation numbers from the character string number list stored in the character string number list storing means; and overlapped character string number removing means for removing a plurality of particular character string numbers overlapped with each other by setting the particular character string numbers overlapped with each other to one particular character string number and outputting the particular character string numbers which are not overlapped with each other and correspond to the definite character strings.