System for creating expression in universal language, and recorded medium on which translation rules used for the system are recorded

[0002] The present invention relates to a technology for overcoming language problems in translation, interpretation and so on, for information exchanges and/or information storage. More particularly, the present invention relates to a language conversion technology, which is suitable for information exchanges and information storage performed in computers, networks and so on and which can be applied to machine translation, machine interpretation and so on.

[0003] Today in the Internet age, people's activities have been spread internationally. During international activities, communication is the most important. Accordingly, difference in language is an important matter. However, a perfect global common language does not exist at the present time. Even widely-used English cannot be the global common language. In fact, people belong to various language zones, and each of them desires communication by using his/her native language.

[0004] Currently, in order to achieve the communication between different languages, translation between these languages is necessary. In some cases, another language which can be translated from both languages must lie between. Translation generally relies on human power. Alternatively, translation software has been developed for some languages, and machine translation may be performed therefor.

[0005] In fact, conventional translation software analyzes a construction and a meaning of a sentence written in a certain language. Then, words to be translated into are used to rewrite elements of the sentence according to the grammar thereof. Therefore, perfect translation is significantly difficult to achieve. Even the translation between very similar languages can hardly reach to 100%.

[0006] Human beings create information expressed in languages under various conditions in various fields such as reporting, transaction, literature, and entertainment regardless of their used languages. Information written in a language is generally stored individually in each form such as printed form, digital data form, or the like in the language the information is originally written. The information may be translated into a different language and may be stored in translated form as well. Therefore, having the same content information may be stored in each different language. This cannot be considered efficient way in the network age. Especially, when the content is changed, not only the original text must be changed but also the changed part must be translated in all of the translated text. Accordingly, it is not efficient way either in view of document management.

[0007] In general, information originally written in one language is not translated into a different language in many cases. Especially, it is rare to translate it into many languages. Therefore, exchanges of information between people using different languages are naturally limited.

[0008] Under this condition it is desirable to develop a system that can achieve communication among people using various kinds of languages easily and accurately with small cost. Further, it is desired to accumulate information independent of a specific language and to convert the information into a desired language as necessary.

[0009] One of the objectives of the present invention to provide a system for creating representation in universal language, whereby communication among various kinds of language can be achieved easily and accurately with small costs. Another objective of the present invention to provide a recording medium for recording rewriting rules used therefor.

[0010] In order to achieve the objective, according to a first aspect of the present invention, it is provided a system for creating representation in universal language, which can be handled by a computer and on which conversion to the other language can be based, comprising a conversion rule portion for, corresponding to at least one kind of natural language, storing rewriting rules which are referred to for converting representation in natural language to representation in universal language and a conversion processing portion for converting representation in given natural language to representation in universal language by referring to the rewriting rules stored in the conversion rule portion, wherein the conversion rule portion has, as the rewriting rules, a universal language dictionary for defining natural language elements constructing natural language by universal language elements constructing universal language, which correspond to information that each of natural language elements has, and rewriting rules for stipulating a join relation between the universal language elements arranged in accordance with a sentence construction in natural language and for representing the join relation in a binary relation including two universal words and a relation indicating that they have a certain relation, and wherein the conversion processing portion performs processing for rewriting components constructing the natural language representation of the representation in given natural language to universal language elements corresponding thereto in the universal language dictionary with reference to the universal language dictionary and processing for joining multiple rewritten universal language elements by applying the rewriting rules in accordance with an order in universal language element and for creating representation in universal language, which is represented in the binary relation, in accordance with a rule for representation in the binary relation.

[0011] According to a second aspect of the present invention, it is provided a recording medium for recording a rewriting rule compliant with at least one kind of natural language, which is referred to for mutually enconverting between representation in the natural language and representation in universal language, which can be handled by a computer, the rewriting rule has a universal language dictionary for defining natural language elements constructing natural language by universal language elements constructing universal language, which correspond to information that each of natural language elements has, and rewriting rules for stipulating a joint relation between the universal language elements arranged in accordance with a sentence construction in natural language and for representing the join relation in a binary relation including two universal words and a relation indicating that they have a certain relation.

[0012] FIG. 1 is a block diagram showing an example of the construction of a hardware system used in the present invention.

[0013] FIG. 2 is a block diagram of the functional construction of a system for creating representation in a universal language according to the present invention.

[0014] FIG. 3 is an explanatory diagram schematically showing an example of the construction of a universal language dictionary used in the present invention.

[0015] FIG. 4 is an explanatory diagram schematically showing an example of a construction of a rewriting rule used in the present invention.

[0016] An embodiment of the present invention will be descried below with reference to drawings. The present invention will be described by using, as an example, a case where representation in English is used as representation in the natural language and representation in universal language is created based thereon. Of course, it is possible to create representation in the universal language based on various languages.

[0017] According to the present invention, the universal language to be used to create for representation is a language that can be basically handled by computers and also used for communication via network. Therefore, notations for the universal language may be enough if they can be handled by computers, such as numerals. However, it may be represented in a human-understandable manner. For example, codes in the universal language can be coordinated to notations in a specific natural language. In this embodiment, an example where notations in the universal languages are based on English is used. This is because English is a widely used language and is also widely used for notation of the computer language. Of course, the notation of the universal language may be based on a non-English language. Alternatively, it may be a notation in an artificial language, which itself can be understood by human beings.

[0018] FIG. 1 shows an example of a construction of a hardware system, which can be used to construct a representation system in the universal language according to the present invention. A generic computer system is shown in FIG. 1. That is, the hardware system shown in FIG. 1 has an operational device 10, an input device 20 having a keyboard 21 and a mouse 22, an output device 30 having a display device 31 and a printer 32, a memory device 40 including a hard disk device and so on, and a communication control device 50 for connecting to networks. Here, the input device 20 and the output device 30 are both generic terms. Accordingly, for example, the output device 30 does not have to be a device having the display device 31 and the printer 32.

[0019] The operational device 10 has a central processing unit (CPU) 11, a read only memory (ROM) 12 and a random access memory (RAM) 13. The CPU 11 loads application programs into the RAM 13 and executes the same installed in the memory device 40 in order to perform processing of different kinds of applications. One of the application programs is a program for performing processing for creating representation in universal language used in this embodiment. The program is provided by a recording medium such as CD-ROM and a flexible disk. Further, the program may be obtained by downloading it from a server, which provides programs, over a network or the like. In both cases, it is converted into an executable form and is stored in the memory device 40 according to this embodiment.

[0020] Furthermore, according to this embodiment, rewriting rules are used for the convert between natural language and universal language. Like the above-mentioned programs, the rewriting rules may be provided in a form recorded in a recording medium. Alternatively, they can be obtained over a network. Then, the rewriting rules are stored in the memory device 40.

[0021] FIG. 2 shows a functional construction of this embodiment. As shown in FIG. 2, the construction includes a conversion preparation processing portion 100 for preparing data represented in natural language, on which representation in universal language is based, a conversion rule portion 200 for storing rewriting rules to be referred to for converting representation in natural language to representation in universal language so as to be compliant with at least one kind natural language, a conversion processing portion 300 for performing conversion processing of converting representation in natural language given by the conversion preparation processing portion 100 into representation in universal language by referring to the rewriting rules stored in the conversion rule portion 200, a storage processing portion 400 for performing processing to store the created representation in universal language in the memory device 40, and an output processing portion 500 for performing processing for outputting representation in the process of the creation processing and after creation processing.

[0022] The conversion preparation processing portion 100 performs processing for capturing data represented in natural language to which the conversion processing portion 300 refers. The capturing processing may include following processes, for example: There may be a process for receiving inputs of character string data representing representation in natural language, which has been input through the keyboard 21. Further, there may be a process for reading from the memory device 40 character string data representing representation in natural language held in the memory device 40. Furthermore, there may be a process for capturing character string data received through the communication control device 50.

[0023] The conversion rule portion 200 has a universal language dictionary 210 which defines natural language elements constructing natural language in universal language elements constructing universal language, corresponding to information pieces that each of the natural language elements has, and rewriting rules 220 which defines join relations among the universal language elements aligned in accordance with a sentence construction in natural language, and which are used for writing a join relation in a binary relation including two universal languages and a relation indicating they have a specific relation. Both of them are stored in the memory device 40. Here, the universal language dictionary 210 and the rewriting rules 220 are dealt separately in this embodiment. However, The present invention is not limited thereto. For example, the natural language dictionary 210 and the rewriting rules 220 can be dealt collectively as one rewrite rule without making a distinction. Alternatively, each of the universal language dictionary 210 and the rewriting rules 220 can be subdivided furthermore.

[0024] The universal language dictionary 210 includes a collection of universal language elements written in symbols used in universal language corresponding to natural language components. As a universal language element, at least one of information defining a meaning indicated by a corresponding component of natural language and information defining an evaluation for a corresponding component of natural language, is written. Each universal language element is provided with a sentence element symbol indicating which element of a sentence in natural language the universal language element corresponds to.

[0025] FIG. 3 schematically shows an example of a structure of the universal language dictionary. In the example shown in FIG. 3, only parts of dictionary elements are shown, of course. The universal language dictionary in the example shown in FIG. 3 includes descriptions such as ‘a (“.@indef”), DET’ and ‘man(“person”), N’ and so on. That is, the descriptions are in form of:

[0026] X(“Y”),Z

[0027] Here, an element in natural language, that is, in this embodiment, an element in English is written for X. In the above-described example, ‘a’ and ‘man’ correspond thereto. Many of the elements in natural language are words.

[0028] For example, elements having the same spelling but having different meanings are defined separately. Further, root form, past form and past participle of an irregular verb are dealt as separate elements. Here, elements described herein are not limited to independent words. For example, parts such as a stem and a suffix are also dealt separately. In this case, they are added to each element as information indicating the characteristic thereof. For example, information indicating a tense, or information indicating a plural form and so on are added thereto. The stem and the suffix of a regular verb are separated and are defined as independent elements, respectively. ‘ask’ and ‘ed’ shown in FIG. 3 correspond to the stem and the suffix. Furthermore, adding ‘ed’ at the end of a noun causes the noun to function as an adjective. In this case, it is defined as a suffix having a different characteristic from the one at the end of the verb.

[0029] For Y, at least one of information defining a meaning indicated by a corresponding component of natural language and information defining an evaluation for a corresponding component of natural language is written.

[0030] The information defining a meaning corresponds to ‘person’ in the above-described example. Of course, the information defining a meaning is not limited to a noun, but may be a verb or an adjective.

[0031] The information defining an evaluation corresponds to ‘.@indef’ in the above-described example. In other words, it defines a characteristic of a noun as an indefinite article and can be information defining a meaning. According to the present invention, they are called attributer.

[0032] Information indicating various characteristics such as a tense, an intention, a reason and an event can be written as the information defining a meaning. Further, depending on a condition of a target, multiple kinds of information can be written together by, for example, separating them with a comma or the like. For example, ‘question’, which is an English word shown in FIG. 3, is defined in the binary relation. That is, it is written as question (“question(icl>thing)”). This defines that ‘question’ in English is written as ‘question’ in universal language. Furthermore, it defines that ‘question’ in universal language has a characteristic of ‘(icl>thing)’. The ‘(icl>thing)’ is a symbol relating to a concept, and ‘question’ is defined to be included in a concept of ‘thing’. Here, ‘thing’ indicates a concept of a thing in universal language.

[0033] For Y, information defining a meaning and information defining the evaluation can be written together by separating them, for example, by using a comma.

[0034] The sentence element symbol is written for Z. It corresponds to ‘DET’ or ‘N’ in the above example. The sentence element symbol corresponds to information indicating, for example, a noun in natural language. For example, ‘DET’ is a symbol meaning a determiner and indicating that it is a sentence element which determines a characteristic of a target noun, such as an article. ‘N’ is a symbol indicating that it is a sentence element meaning a noun. ‘VTSTM’ and ‘ENDING’ are symbols indicating the stem of a transitive verb and a suffix element, respectively.

[0035] The rewriting rules 220 stipulates join relations between the universal language elements aligned in accordance with a sentence structure in natural language as indicated by numbers 1 to 10 shown in FIG. 4. For example, in No. 1 in FIG. 4, a relation is shown that a sentence S has NP which is a subject part, and VP which is a verb part, and they are joined by a relation, agt. In No. 2, a relation is shown that VP has VT that is a transitive verb, and NP that is a noun phrase, and they are joined by a relation, obj. The relation will be described later.

[0036] In No. 3, a relation is shown that NP corresponding to a noun phrase is joined with DET. In No. 4, a relation is shown that VT corresponding to a transitive verb is joined with VTSTM and ENDING.

[0037] Those in No. 5 or below are stipulated in rules different from the rule for No. 1 to 4. For example, in No. 5, a relation is shown that DET is joined with ‘a(“.@indef”)’ by removing an element in natural language, that is ‘DET(“.@indef”)’. The same is true in those in No. 6 and after.

[0038] According to the present invention, the universal language dictionary 210 and the rewriting rules 220 are used when representation in universal language is created from representation in natural language. The same rewriting rule can be used when representation in natural rule is created from representation in universal language. Therefore, once representation in universal language is created, various representations in natural language can be easily created. How to create representation in universal language by using the universal language dictionary 210 and the rewriting rules 220 will be described later.

[0039] The conversion processing portion 300 performs a universal language element rewrite processing 310 for referring the universal language dictionary, to rewrite components constructing a given natural language representation to corresponding universal languages elements in the universal language dictionary. Further, the conversion processing portion 300 performs binary relation creation processing 320 for joining the multiple rewritten universal language elements by using the rewrite rule in accordance with the alignment and for creating representation in universal language represented in the binary relation in accordance with the rule for the notation in the binary relation. The conversion processing is performed by executing a program mainly in the operational device 10.

[0040] The storage processing portion 400 performs processing for storing a result processed in the operational device 10 in the memory device 40. The processing is performed by the operational device 10, and the memory device 40 stores data output from the operation device in response to the processing. The data is read as required.

[0041] The output processing portion 500 performs processing for displaying input data, processing results and so on in the display device 31 and for causing the printer 32 to print them. The displaying and the printing are performed in separate processes.

[0042] Next, the binary relation, which is a feature of the present invention, will be described. The binary relation used in the present invention is defined as a following equation:

[0043] where

[0044] ::=indicates that . . . is defined as . . . ;

[0045] | indicates disjunction, “or”;

[0046] [ ] indicates an optional element which can be added as desired

[0047] { } indicates an alternative element;

[0048] “ ” indicates a character and/or a symbol is included therein; and

[0049] <xxx> is a variable name.

[0050] By using the above equation, various representations can be achieved accurately. The simple examples include:

[0051] and(quickly, easily),

[0052] and(Mary(icl>person), John(icl>person)).

[0053] The former indicates meanings of “quickly” and “easily”. The latter indicates meanings of “Mary” and “John”. Here, the symbol ( ) indicates that there is a constraint, which is written within the ( ) about the element. This will be described together with the description on universal language.

[0054] A compound universal word is a collection based on the binary relation for indicating the compound concept. The compound universal word ID is an identifier for identifying the binary relation belonging to the collection of compound universal words. By specifying the identifier, the compound universal words can be used as a universal word. Here, the compound universal word ID is represented by, for example, alphabet capital letters or two-digit numbers.

[0055] Incidentally, also for the universal words, representation of the same content can be referred to by using the universal word ID as shown in Equation 6, which is described later.

[0056] The binary relation is a representation in universal language, namely, a sentence block. The binary relation is constructed of two universal words and a relation. In universal language, the binary relation is predefined and stipulated by a symbol having three or fewer characters, which is called a relation. The relation between universal words differs depending on behaviors of the universal words, and a corresponding relation such as agt, and, aoj, obj and icl is used. These relations are predefined, for example, as follows:

[0057] RELATIONS

[0058] agt: a thing which causes an event

[0059] and: a conjunctive relation between concepts

[0060] aoj: a thing having an attribute

[0061] bas: a thing used as the basis and/or degree of comparison

[0062] cag: a thing which initiates an implicit event

[0063] cao: a thing having an implicit attribute

[0064] cnt: an equivalent concept

[0065] cob: target of an implicit event

[0066] con: an event and/or a state which conditions a focused event or state

[0067] coo: a co-occurred event or state for a focused event or state

[0068] dur: a period of time while an event occurs or a state exists

[0069] fmt: a range between two things

[0070] frm: an origin of a thing

[0071] gol: the final state of object or the thing associated with object

[0072] ins: the instrument to carry out an event

[0073] man: the way to carry out an event or a characteristics of a state

[0074] met: means to carry out an event

[0075] mod: a thing which restricts a focused object

[0076] nam: a name of a thing

[0077] obj: a thing in focus which is directly affected by an event

[0078] or: select relation between concepts

[0079] per: reference unit

[0080] plc: the place where an event occurs or a place where a state exists

[0081] plf: the place where an event begins or a place where a state becomes true

[0082] plt: a place where an event ends or a place where a state becomes false

[0083] pof: whole and part relation

[0084] pos: possessor of a thing

[0085] ptn: unfocused thing which initiates an event together

[0086] pur: the purpose of existence of an event or a thing

[0087] qua: quantity of a thing or unit

[0088] rsn: a reason that an event or a state happens

[0089] scn: a scene where an event occurs

[0090] seq: a prior event or state of a focused event or state

[0091] src: the initial state of object or a thing associated with object

[0092] tim: the time when an event occurs or a state is true

[0093] tmf: the time when an event starts or a state becomes true

[0094] tmt: the time when an event ends or a state becomes false

[0095] to: a destination of a thing

[0096] via: an intermediate place or state of an event

[0097] Next, universal language used in the present invention will be described. The universal language will be defined in the equation below. The notation rule for the equation is the same as that in Equation 1.

[0098] <universal words>::=<title>[<constraint list>][“:”<universal word ID>][“.”<attribute list>]  [EQ 2]

[0099] The <title> is formed by a character string, and may include buy, or book, for example. The universal language may only need to be handled by computers. Thus, it is not always necessary to represent it in human readable form. However, it is more convenient to define it in human readable form. Therefore, in this embodiment, universal language is defined based on English as described above. No problems will occur even if it is defined in other language, for example, in Japanese.

[0100] As apparent from the form of definition equations, in the simplest form, universal language can be constructed only by the title.

[0101] In general, there are various constraints on language elements.

[0102] Accordingly, in this embodiment, one or more constraints are indicated in ( ) as a constraint list. Multiple constraints can be written together. In that case, they are separated by commas. The constraint list is defined as follows:

[0103] <constraint list>::=“(”<constraint>[“,”<constraint>] . . . “)” where . . . means that one or more subjects appear.  [EQ 3]

[0104] A constraint listed in the constraint list is defined as follows:

[0105] <constraint>::=<relation>{“>”|“<”}<universal word>[<constraint list>]|<relation>{“>”|“<”}<universal word>[<constraint list>][{“>”|“<”}<universal word>[<constraint list>]]  [EQ 4]

[0106] In this way, universal language includes unconstrained basic universal words, restricted universal words having some constraints, and a special type of restricted universal words. The basic universal words include ‘go, take, house’ in English, for example. The restricted universal words include various universal words, which can be represented by the above-described definitions. As an example, a case where ‘state’ has four different constraints will be shown as follows.

[0107] state (agt>person,obj>information)

[0108] state (equ>nation)

[0109] state (icl>situation)

[0110] state (icl>government)

[0111] An example of the special restricted universal words will be represented as follows:

[0112] ikebana (icl>activity,obj>flowers)

[0113] samba (icl>dance)

[0114] As information identifying the characteristic includes an attribute. The attribute is defined by following equation:

[0115] <attribute list>::=<attributer>[“.”<attributer>]  [EQ 5]

[0116] Attributes in universal language include the time relating to a speaker such as past, present and future, perspective of the speaker regarding whether he/she is in hurry, speaker's reference opinion regarding a common topic, speaker's focus on a point that the speaker pays attention to, how strongly the speaker is interested in a topic, and speaker's opinion. Typically, they are defined as follows:

[0117] Next, universal word ID is used for identifying object by alphabets, numerals and so on in order to specify the object, when, for example, there is a representation that can be referred to, or when there is the same representation. The universal word ID can be defined as follows:

[0118] <universal word ID>::={alphabet capital letter>|<numeral>}{alphabet small letter>|<numeral>}  [EQ 6]

[0119] In addition to alphabet capital letters and small letters and numerals, following symbols can be used along with the letters and numerals to indicate the binary relations applied in this embodiment:

[0120] “_”|“*”|“#”|“!”|“$”|“%”|“=”|“{circumflex over ( )}”|“˜”|“|”|“@”|“+”|“−”|“<”|“>”|“?”  [EQ 7]

[0121] Next, a specific example of the creation of representation in universal language according to the present invention will be described. In the example below, a case will be described where representation (a sentence) is created in universal language for English representation (a sentence) of ‘the man asked a question’. The universal language dictionary shown in FIG. 3 and the rewriting rule shown in FIG. 4 is used for the creation. Here, the rewriting rule in FIG. 4 is applied from right to left in the creation in universal language. In other words, it is applied in the direction contrary to the direction indicated by an arrow (→) described in the rewrite rule. Here, the conversion from the universal language to the natural language is performed in the direction indicated by the arrow.

[0122] The operational device 10 captures a target sentence to be translated, ‘The man asked a question’, into the RAM 13 through the conversion preparation processing portion 100. In this example, it is input through the keyboard 21. Then, the CPU 11 of the operational device 10 searches for the first word in the English sentence inputted through the universal language dictionary 210.

[0123] Then, the(“.@def”), DET′ is retrieved. Then, the rewriting rule 6 in FIG. 4 is applied to obtain the following result:

[0124] DET(“.@def”) man asked a question  [EQ 8]

[0125] Next, the CPU 11 searches for the next word, ‘man’, in the English sentence through the universal language dictionary 210. Then, ‘man (“person”), N’ is retrieved. Then, the rewrite rule 8 in FIG. 4 is applied to obtain the following result:

[0126] DET(“.@def”)N(“person”) asked a question  [EQ 9]

[0127] Next, the CPU 11 searches for the next word, ‘asked’, in the English sentence through the universal language dictionary 210. Then, ‘ask (“ask(icl>do)”), VTSTM’ is retrieved. Then, the rewrite rule 9 in FIG. 4 is applied to obtain the following result:

[0128] DET(“.@def”)N(“person”)VTSTM(“ask(icl>do)”)ed a question  [EQ 10]

[0129] Next, the CPU 11 searches for ‘ed’, the ending of the word, ‘asked’, in the above English sentence through the universal language dictionary 210. Then, ‘ed(“.@past”), ENDING’ is retrieved. Then, the rewrite rule 10 in FIG. 4 is applied to obtain the following result:

[0130] DET(“.@def”)N(“person”)VTSTM(“ask(icl>do)”)ENDING(“.@past”) a question  [EQ 11]

[0131] Next, the CPU 11 searches for the next word, ‘a’, in the English sentence through the universal language dictionary 210. Then, ‘a(“.@def”), DET’ is retrieved. Then, the rewrite rule 5 in FIG. 4 is applied to obtain the following result:

[0132] DET(“.@def”)N(“person”)VTSTM(“ask(icl>do)”)ENDING(“.@past”)DET(“.@indef”) question  [EQ 12]

[0133] Next, the CPU 11 searches for the next word, ‘question’, in the English sentence through the universal language dictionary 210. Then, “question(icl>thing)” is retrieved. Then, the rewrite rule 7 in FIG. 4 is applied to obtain the following result:

[0134] DET(“.@def”)N(“person”)VTSTM(“ask(icl>do)”)ENDING(“.@past”)DET(“.@indef”)N(“question(icl>thing)”)  [EQ 13]

[0135] Next, after confirming that there is no words after the ‘question’ in the English sentence, the CPU 11 stops dictionary searches and performs further rewrite processing on the rewrite result. The determination that there is no more word to be searched in the English sentence can be performed based on the presence of a period, a question mark, a colon, or the like.

[0136] After that, the CPU 11 further performs rewriting by applying rewriting rules 1 to 4.

[0137] The CPU 11 reads our sentence element symbols from the beginning of the rewrite result and determines relations between adjacent sentence element symbols from the beginning of the sentence with reference to rewrite rule shown in FIG. 4. For example, in case of the above rewrite result, order of the arrangement sentence element symbols is ‘DET, N, VTSTM, ENDING, DET, N’ from the beginning. Then, rewrite rule 3 is applied to ‘DET’ and ‘N’ to obtain ‘NP(“person.@def”)’. As a result, the following result is obtained:

[0138] NP(“person.@def”)VTSTM(“ask(icl>do)”)ENDING(“.@past”)DET(“.@indef”)N(“question(icl>thing)”)  [EQ 14]

[0139] Next, the CPU 11 applies rewrite rule 4 to ‘VTSTM’ and ‘ENDING’ to obtain ‘VT (“ask(icl>do).@past”). As a result, the following result is obtained:

[0140] NP(“person.@def”)VT(“ask(icl>do).@past”)DET(“.@indef”)N(“question(icl>thing)”)  [EQ 15]

[0141] Next, the CPU 11 applies rewrite rule 3 to ‘DET’ and ‘N’ to obtain ‘NP(“question.@indef”)’. As a result, the following result is obtained:

[0142] NP(“person.@def”)VT(“ask(icl>do).@past”)NP(“question.@indef”)  [EQ 16]

[0143] Next, the CPU 11 determines that there is no target to be rewritten at the ending. Then, the CPU 11 reads out sentence element symbols and determines relations between adjacent sentence element symbols from the beginning of the sentence with reference to rewrite rule shown in FIG. 4. For example, the order of arrangement sentence element symbols is ‘NP’, ‘VT’, and ‘NP’ in case of the above rewrite result. Here, the rewrite rule cannot be applied to ‘NP’ and ‘VT’ on the beginning side. Thus, the CPU 11 first applies rewrite rule 2 to ‘VT’ and ‘NP’ in order to obtain the following result:

[0144] NP(“person.@def”)

[0145] VP(“ask(icl>do)(obj>question.@indef).@past”)  [EQ 17]

[0146] Furthermore, the CPU 11 reads out sentence element symbols included in the rewrite results and rule 1 is applied thereto. Then, the following result is obtained:

[0147] S(“ask.@entry(agt>person.def,obj>question.@indef)”)  [EQ 18]

[0148] In this way, the representation in universal language is created. The universal language can be interpreted as following: The word ‘ask’ in universal language means ‘ask’ in natural language, and the speaker is focused on ‘ask’ based on the presence of the attributer ‘entry’. The universal word ‘ask’ has a constraint, and the content of the constraint means that a thing (agt) initiating an event, ‘ask’ is a definite (.@def), ‘person’, and that a thing directly influenced by the event is indefinite (.@indef), ‘question’.

[0149] As apparent from the components, this representation in universal language does not include components in natural language. Therefore, when natural language is created from once-obtained universal language, the conversion becomes possible from/to arbitrary language if there is the rewriting rule corresponding to the language to be translated.

[0150] The universal language dictionary and rewriting rules used in the above examples are only parts of the whole. For example, about one million words are prepared for the universal language dictionary. Incidentally, the smaller number of words can be used when the present invention is used to a specific purpose.

[0151] In this way, according to the present invention, by preparing rewriting rules corresponding to each natural languages, any natural language can be represented in universal language. Also, the universal language is written in a form that it is not influenced by elements and grammar of original natural language. Therefore, when information in one language is communicated to the other language, international communication and efficient storage of the information are allowed by using representation in universal language. The content in language having less translated information can be communicated to many people using different kinds of language easily with low costs like communication in English.

[0152] As the universal language of the present invention can be converted to natural language, it can be determined easily whether original representation in natural language is communicated accurately through the representation in universal language by checking reverse-converted representation. Thus, the representation in universal language can be accurate. Furthermore, universal language is written clearly by using a binary relation, and thus the conversion from universal language to natural language can be performed accurately. As a result, having the representation in universal language as the original is useful when the same information is communicated internationally.

[0153] Although natural language is described in the above example, the present invention can be applied to representation in language, which may not be natural language strictly; for example, artificial language such as Esperanto language, or language for machine such as computer language.

[0154] According to the present invention, communication among various languages easily and accurately with small costs.