JPS58219670A - Dictionary for translation and dictionary retrieval system using it - Google Patents

Abstract

PURPOSE:To decrease the necessary number of dictionaries, by describing the meanings of words and phrases in dictionaries for translation not in other languages, in codes for expressing meaning conceptions. CONSTITUTION:An electronic dictionary consists of, for example, a processor 1, display 2, input device 3, dictionary memories 4-7 of English, Japanese, Germany, and French, working area 8, and memory 9 for program storage. The memories 4-7 each have the 1st section area for the correspondence relation between coded characters forming words described in the 1st language and ideographic codes for expressing meaning conceptions of those words and the 2nd section area for the correspondence relation between coded characters forming words described in the 2nd languages and their ideographic codes. Consequently, the number of dictionaries necessary for translation among (n) kinds of language is reduced from 2XnC2 to (n).

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an electronic dictionary and dictionary search method used in a device for translating from one language to another, and is particularly suitable for multilingual translation and reduces storage capacity. This is related to a dictionary structure that can be used in a variety of ways.

In recent years, so-called electronic dictionaries and translators have been developed that translate the first input word, idiom, or sentence into a second language (output language). uses a dictionary in which words or idioms in the second language (output language) corresponding to the meanings of words or idioms in the first language (input language) are stored.

However, since the second language is used to describe the meaning, such dictionaries do not allow the translation from the second language to the first language.
Or, it cannot be used as a device for translating word idioms or sentences from a first language to a third language. Therefore, n(n
is a natural number greater than or equal to 2] To perform translation between languages, 2XnC! (i.e., the number of bidirectional X language combinations) dictionaries must be prepared, and 11111□ for translation between n languages - It requires a lot of effort to create a dictionary, and the capacity of the dictionary is large. There was a drawback.

SUMMARY OF THE INVENTION It is an object of the present invention to overcome the above-mentioned drawbacks, to provide a translation dictionary that is suitable for multilingual translation, and has a small storage capacity, and a search method using the same. The present invention is particularly effective for cases in which the concepts of words and idioms are less likely to deviate between languages, such as the names of things.

In order to achieve the above object, in the present invention, the meanings of words and phrases in translation dictionaries are not described in other languages, but the code representing the meaning concept, for example, a binary bit string c or less, is written as a code for simplicity. is described using binary bits as an example, and this code is called an ideographic code). According to the present invention, when translating from a first business word (input word) to a second head word (output language), a dictionary of the first language (input language) whose meaning is described in ideographic codes and a second language By using a dictionary with a total output of 1, the number of dictionaries required for translation between n languages can be reduced from t-2Xnc! to n.Furthermore, By representing the ideographic code as a bit string itself rather than as characters, it is possible to reduce the memory capacity Mkt required for semantic description, and by providing it with a reverse lookup dictionary, it is possible to speed up retrieval. It is also possible to

Hereinafter, the present invention will be explained in detail with reference to Examples. 1st
The figure shows an electronic dictionary according to an embodiment of the present invention. In FIG. 1, 1 is a processor, 2 is a display device, 3 is an input device such as a keyboard, 4 is an English dictionary memory, 5 is a Japanese dictionary memory, 6 is a German dictionary memory, 7 is a French dictionary memory, 8 9 represents a work area and h represents a program storage memory. The dictionary memory for each language stores a dictionary in the format shown in FIG. Tsumashi,
Heading at key: part-of-speech name, part-of-speech base, irregular declension, ka (for nouns in German and French)
, an ideographic code (the ideographic code will be described later), the number of idioms, and an idiom pattern are stored. The part-of-speech name is written after the first G) symbol. This is provided so that it is possible to describe the same entry word with multiple headings @, and the minimum value is i. The part-of-speech base is written in the column of the symbol Q) from the m2th onward based on the part-of-speech code b shown in Table 1.

When the part of speech base is a verb, if the headword changes irregularly in the past tense and past participle form, place the symbol Q) and change the past tense and past participle form in this order with spaces wo' and /. Write in sections. If the part of speech base is a noun and the plural form changes irregularly, place the symbol Q) and write the plural form after it.

If it does not change irregularly, the symbol (D) is omitted and the past tense, past perfect tense, or plural form is not written.

German and 7 Reims nouns have genders such as masculine, neuter, and feminine, so if the part of speech base is noun in the dictionaries in German dictionary memory 6 and French dictionary memory 7, 0) Enter the symbol ti followed by the gender code shown in Table 2. If there is no gender distinction, omit the symbol ■). The meaning corresponding to the part of speech base is
This is done using ideographic codes as shown in Table 3, and θ)
It is written after the symbol. If there are multiple meanings, write one ideographic code after the 0+ symbol to express it.
1 Repeat the description pattern shown in Table 2. In the case of a multi-part-of-speech word with multiple parts of speech, we will explain the part-of-speech name part, then the □ idiom description. An idiom is registered with one of the words making up the idiom as a headword. The idiom number is written after the first ■) symbol. If there is no idiom to be registered, the number of idioms is set to 0. Therefore, the minimum value of the number of idioms is 0. The idiom pattern (form of the idiom) is written after the second and subsequent Gφ symbols.However, the idiom pattern must be written in the alphabet,
Alternatively, write in katakana, separate each word with a space, and for the part of the idiom pattern that corresponds to a headword, write ``water'' instead of writing the entire headword. Fill out. The meaning corresponding to an idiom pattern is written using the symbol G) followed by an ideographic code. If an idiom pattern has multiple meanings, the description pattern of writing one ideographic code after the symbol qφ is repeated. If there are multiple registered idioms, repeat the above (9 idiom pattern and subsequent parts). Then, at the end, put a symbol ■ to indicate the end.

Next, the ideographic codes shown in Table 3 will be explained in detail.

An ideographic code is a binary bit string for expressing a semantic concept common to each language, and in the case of Table 3, it has a length of 16 bits. Therefore, with this ideographic code, 2111 fr
It can represent IA (approximately 64,000 semantic concepts). In the conventional method, the meaning is directly written in the characters of the second language (
JIs, ASCII, EBCDIC code, etc.), it took 7 to 16 bits to write just one character of a sentence, candy, or word that describes the meaning, resulting in very poor memory efficiency. However, according to this method, approximately 64,000 semantic concepts can be described using only 16 bits, and memory efficiency can be improved. Furthermore, by encoding the semantic concepts common to each language, it is possible to use a dictionary for one language for translation into various languages. Tsumashi, a dictionary for translating from a first language to a second language using the conventional method is a second language table.
It is difficult to use for translation from 3 to the first language, and furthermore,
The dictionary cannot be used at all when translating the 1d word into other languages. Therefore, according to the conventional method, nl [
% for translation between languages of 1il (n is a natural number greater than or equal to 2)
2XflCt dictionaries are required, but according to this method, only n dictionaries are required. Therefore, there is an advantage that the effort of creating a dictionary can be saved. Now, in Table 3, the semantic concept of the ideographic code '1010000000000011' is 1 sphere), but (sphere)
What this means is that the meaning concept pointed to by this ideographic code is written in Japanese as 'ball', and it does not encode character information determined by Japanese words. The ideographic code represents a common semantic concept independent of language examples, and is determined by the semantic conceptual structure of the thing itself.

In addition, regarding coding in ideographic codes, as shown in Table 4, ideographic codes expressing similar semantic concepts are expressed as "close" if the distance between the codes is small.
Another possible method is to place it in This arrangement makes it easier to search for synonyms.

Figure 3 shows the entry words of the dictionary in the English dictionary memory 4.
This shows the contents of ace/.

In the case of ゝplace /, there are two part-of-speech names, a noun and a verb, so the part-of-speech base is 2, ■21r.

Describe it first. After the next [F], a code 'oooo' indicating a noun is written. In this case, there are no irregular changes, and since it is English, there is no gender, so ■, q
) symbols are omitted. Next ■ Relaxation 10100000
00000010 means (location) of the semantic concept, and the next e 1010000000000100 indicates that the second meaning is (position). At this point, the description of the noun part ends, and next the description of the verb part begins. ■su0
010 means that the part of speech name is a verb. Here too, there is no irregular change, and the verb is taω, ■
Reference symbols are omitted. Next 10100000
00000110 indicates that the meaning of the verb is (to put). (The part after m is the idiom description part. In this case, the idiom is take pl
ace and in place of), so
First it is written as 6]2. Next (81ake * is an idiom, pattern that indicates the shape of an idiom.
This is a description of place /. Xplace
' part is 'Place/ is a headword, so
It has been replaced with ``*''. Next ■toioooo
oooooi1i indicates that it has the meaning of (happening).

The following ■in * of is an idiom pattern '
This means that it is in plBce of t. The following G 1101111111111111 shows that the meaning of the idiom is (instead of). The last 6) means the end.

As described above, the dictionary registration of the word pHce' is performed. Dictionaries for other languages are constructed in a similar manner. Note that compound words, ren M, etc. are written in the idiom section of the Japanese dictionary.

The English dictionary shown in FIG. 3 is stored as variable length data in the English dictionary memory 4 with p13ce' as the headword. Therefore, here, (Eri.

j),...,GI? ) symbols act as delimiters and make it possible to identify various types of data.

Now, next, let us explain the operation of the electronic dictionary shown in Fig. 1 at 7 in Fig. 5.
0-・Explain while referring to the chart.

First, an initial screen shown in FIG. 6 is displayed on the screen of the display device 2. Then, key in the number of the language you want to translate into the 5OUR, CF fields, select the language you want to translate into by number, and key in the TAB, GET fields.
In. Based on that information, it is determined which dictionary will be used. Key in the word or idiom you want to translate and press the return key to translate that word into 5OUs.
A search is made from the dictionary of the language specified in the R and CB columns, and all information stored in the headword or idiom pattern is transferred to the work area 9. Next, read out the ideographic code from it and enter the key in the language, tab, and TABGET fields specified in 100 of FIG.
Search the dictionary memory of the loaded language, find the headword and idiom pattern with the ideographic code,
This is transferred to work area 9 as a translated word. By again specifying the language of shi station language on the keyboard, the specified language, the translation of shi, and the information obtained in step 102 are displayed on the screen of display group #9 from the work area.

Next, the dictionary search portion of FIG. 5 101 will be described in detail. There are two types of search input: words and idioms.
First, let's consider the case where a word is entered at 70-
・Explain while referring to the chart. First, it is searched to see if the input word is in the dictionary (201). If it is not stored as a headword in the dictionary, it is an error and the process proceeds to error processing (208). If it is stored as a headword, all information within the headword is read into a buffer in the work area 8 (202).

Next, data is read sequentially from the beginning of the buffer.

If there is no Q sign at the beginning of the buffer (203), this is an error, and the process proceeds to error processing (209).

(If the symbol 3 is at the beginning, the next number in the buffer is the number of part-of-speech names, so it is read as the number of part-of-speech names (20
9) Assign to part-of-speech name counter C1 (205)
. If C1 is 0, jump to the idiom part reading process (2). If C1 is not 0,
The next delimiter symbol in the buffer is read (206), and if it is not ■, it is an error and the process proceeds to error processing (210). When the delimiter symbol is d-e, the next b movement in the buffer is a part-of-speech code, so read it, change it to a part-of-speech name, and write it to the work area (20
7). If the next delimiter in the buffer is ■, then (
211) First, determine whether the part of speech name is a noun or not.212) If it is a noun, read the string in the buffer between this ■ symbol and the next delimiter symbol as a plural. , is written to the work area (213). If the part-of-speech name is a verb (214), the character string in the buffer sandwiched between the ■ symbol and the next delimiter symbol is past tense,
It is read as a past participle form and written to the work area (215). However, the past tense and past perfect tense are separated by ゝ, ', and the first half is the past tense and the second half is the past participle. After loading this irregular form,
The next delimiter in the buffer is read and it is determined whether it is Jj) (216). However, if the symbol %(D) does not exist, skip the part of reading the irregular inflection and proceed directly to the judgment of ω.

When the delimiter is (, read the gender code,
Convert to character string and write to work area (217)
. If the delimiter is (not boiled), proceed to the next step.

The next step is to determine whether the next delimiter in the buffer is a ■ (218). If the delimiter is not (, it is an error and the process proceeds to error processing (219).
). When the delimiter is ``k'', the next binary bit string is an ideographic code, so it is read and written to the area under work. If the next delimiter in the buffer is (ED), the next binary bit string in the buffer is also an ideographic code, so it is read as an ideographic code and written to the work area (220). , it is read as an ideographic code and the work area Vcm
! ! : Repeat the process of loading. When the next delimiter symbol in the buffer is no longer D (221), the loop is exited and the counter CI is decremented by one (222), and the process returns to ◯ in Figure 7. Then, the above-described word information acquisition process is repeated until C1 is OK. When C1 becomes 0, proceed to the idiom reading process (■).

Now, let's move on to the next step, the process of loading the idiom. First, the next delimiter in the buffer is read (230), and if it is not cleared, it is an error, so proceed to error processing (234), and when the delimiter is good, read the next delimiter in the buffer. Since the numerical value is the number of idioms, it is read as the number of idioms (r231) and assigned to the number of idioms counter C2 (232). If C2 is not 0 (233), the next delimiter in the buffer is read and it is determined whether the delimiter is 0 (235). If C2=0, the process proceeds to ■, and the dictionary search in the case of a word ends. If the read delimiter is not ■, an error occurs and the process proceeds to error processing (243).

When the delimiter p symbol is ■, the character string sandwiched between the delimiter ■ and the next delimiter is an idiom pattern, so it is read as an idiom pattern (2
36), replaces the symbol * part in the pattern with a headword (237), and writes it in the work area (238). The next delimiter in the buffer is ■
If not (239), it is an error and the process proceeds to error processing (244). If the delimiter is 61,
Since the next binary bit string is an ideographic code, it is read as an ideographic code and written to the work area (2
40). If the next delimiter in the buffer is less than (241), the next binary bit string in the buffer is also an ideographic code, so it is read as an ideographic code and inserted into the work area. While the delimiter read from the buffer is Gα, the process of reading it as this ideographic code and writing it to the work 4 area is repeated, and when the delimiter read out is no longer used, the loop starts. , and decrement the value of the idiom number counter C2 by one (242).Then, proceed to ■ in Figure 7 and continue the process of reading the idioms described above until the value of C2 becomes 0. value is 0
When the word is reached, the dictionary search for the word is completed.

A search method when an idiom is input as a search input will be explained with reference to chart 70-- in FIG.

First, the input idiom is broken down into words using spaces as delimiters, and stored in the array WI (301).
. Next, the number of words in the idiom entered? Substitute it into N1 (302). Input idiom word counter 1 in 5
is substituted with 1 (303), and it is determined whether 1 is greater than the number of words N1 of the input idiom (304). If it is larger, it means that there is no idiom corresponding to the input idiom in the dictionary, and the process proceeds to error processing (309). If ■ is not υ larger than N1, search for the fifth word in the input idiom, that is, WI (it is stored in the dictionary as a headword (305).It does not exist as a headword in the dictionary. If (306), proceed to ■, increase the value of 5 by one, and return to ■.If WI (1) is registered as a headword in the dictionary, all contents of that headword are processed.・Read into the buffer in area 8 (307).Then, read the delimiter symbol from within the buffer (308), and continue reading the delimiter symbol in the buffer until the delimiter symbol becomes (. If the ``S'' symbol is found (310), the next number in the buffer is the idiom number, so it is read as the idiom number (311) and assigned to the idiom number counter C (312).Next, ,
It is determined whether the idiom number counter C is 0 or not (313). If it is 0, the process proceeds to ■, increments the value by one, and returns to ■. Idiom number counter C is O
If not, read the next delimiter in the buffer, and if it is not ■ (314), it is an error, so
Proceed to error processing (319). If the delimiter is ■, the string between this symbol and the next delimiter in the buffer is an idiom pattern, so
Load it as an idiom pattern (315)
, and store it in the array WU) (316
). Then, further substitute the number of words tN2 (3
17).

Next, compare the number of words N1 of the input idiom with the number of words N2 of the idiom pattern you are currently focusing on.318)
If they are not equal, proceed to ■ and reduce the value of C by one (3
26) Return to ■ and start comparing with the new idiom pattern. If N1 and N2 are equal, the process proceeds to (■) (Fig. 8), and 1 is assigned to the word counter j (320). Next, it is determined whether the word stored in W(j) is equal to the symbol ゝ*'321), and if they are equal, W(
Resubstitute the headword, that is, WI(ri) for j) (3
22)'. Then, the words stored in WIO) and W
I want to determine whether the words stored in O) are equal or not.3
23) If they are not equal, the shapes of the idioms are different, so proceed to ■ and do the idiom number crunch.
Decrease the value of 〇 by one, return it to ■, load the next idiom pattern, and start comparing. If wi (if jt and w<j> are equal, determine whether j is greater than Nl 324), and if J force is smaller than 5N1, increase the value of j by one (325), and w< Return to the step of comparing j> with the symbol *l, and repeat the word-by-word matching process described above. If j≧N1, it means that the shape of the input idiom and the shape of the idiom we are currently focusing on match, so we proceed to ■ and enter the process of reading the semantic code. . In ■, first read all of the next delimiter symbol in the buffer and judge whether it is (327), (
If not, it is an error and proceed to error handling (
328). If the delimiter is (, then the next binary b
Since the it column is an ideographic code, it is read and written to the work area (329).

Additionally, reads the next delimiter in the buffer, and if it is (330), then the next binary bit in the buffer
The columns are also ideographic codes, read them and write them to the work area. While the delimiter read from the buffer is valid, this ideographic code is read and the work
The process of filling the area is repeated, and when the delimiter symbol read out is no longer a ■, the loop is exited and the dictionary search for idioms is completed.

Next, the part of searching for a translated word from the ideographic code 102 in FIG. 5 will be described in detail with reference to 70- and char[i- in FIG. 9. Here, a method is used in which all ideographic codes stored in the dictionary of the specified language (output language) are compared to see if they are the desired ideographic code. First, the ideographic code to be examined is assigned to the variable COI (401). Next, 1 is assigned to headword counter-i (402
). Then, it is determined whether or not i is greater than the total number of headwords registered in the dictionary (403), and if it is, the process proceeds to ■ and ends the search. If i is smaller than the number of headword lines, all contents of the i-th headword in the dictionary are read into the buffer (404). Next, read the separator symbol from the buffer, and if the read separator symbol is (g), proceed to (405)■ and enter the process of searching for the idiom part.If the separator symbol is not ■, then " Determine whether the delimiter is (B) or not (406) If it is not e, go back to reading the delimiter, read the delimiter, and repeat the comparison with (D). If the delimiter symbol or 3), then the next binary bi in the buffer
Since the t column is an ideographic code, it fr: read variable C
Assign to O2 (4073゜Next, compare the COI containing the ideographic code to be searched with the CO2 containing the ideographic code you are currently focusing on, and if they are not equal, read the delimiter symbol again. Go back to step 1 and repeat the process of ideographic code comparison described above until COI and CO2 are equal. When CO1 and CO2 are equal,
This means that the headword has the desired ideographic code, so the headword is written as a translation into the work area (4093°). Read the delimiters in the buffer until the delimiter becomes 1. When the delimiter becomes ■ (410), read the next delimiter in the buffer and check if it is 0. Determine whether or not (411).If the delimiter is 0,
This means that all the contents of the headword have been checked, so the value of i is increased by one and the process returns to (419) (419), and a search is started for the contents of the new headword. If the delimiter is not ■, it is fully determined whether the delimiter is (), and if it is, this () in the buffer is
The string between D and the next delimiter is an idiom.
Is that because it's a pattern? Read and assign to variable ID (413). In the next step, it is determined whether or not the delimiter is a stream (414), and if the delimiter is not 0), the process returns to ① and starts reading the entire next delimiter in the buffer. When the delimiter is (, the next binary bit string in the buffer is an ideographic code, so read it all and assign it to variable CO2 (415). coi and CO
Compare 2 (416), and if they are not equal, return to ■,
Begin reading the next delimiter in the buffer. COI and C
If O2 is equal, it means that you have the ideographic code required for the idiom/pattern force currently stored in the ID, so replace the symbol ゜*' in the ID with a headword (417) as a translated word. Color the work area (41B). After that, continue searching for the ■mu section. The read delimiter is (D), υ, and the literal of l is 1.
If the total number of entry words becomes i>, the process of searching for translated words from the ideographic code ends.

Next, a specific sequence of search operations for the electronic dictionary shown in FIG. 1 will be explained with reference to FIG. 10.

Figure 10 shows the dictionary memory 4゜5.6 for each language in Figure 1.
．． It shows a part of the contents of the dictionary in 7, 10
is an English dictionary, 11 is a Japanese dictionary, 12 is a German dictionary,
13 represents a French dictionary. The character strings surrounded by squares in each dictionary are headwords, and the meanings of each delimiter and code are shown in FIG. 2 and Table 1.2.3.

Now, specify English as the input language, Japanese, German, and French as the output languages, and use 'ball' as the input word.
Assume that you input the entire English word /. Then, according to the process described above (see the flow chart in FIG. 7), the contents of the headword "1" all/ are imported into the work area. In other words, the ideographic code for the noun form of tb, ll/ is も1010000000000011'.

The information that input 1010000000000101'... is taken into the work area. Next, the first ideographic code imported into the work area
The dictionary of each language is searched for 010000000000011' according to the process described above (see the flow chart in FIG. 9). Then, in the Japanese dictionary, since the headword 'ball' has the same table and meaning code, 'ball' is selected as one of the translated words in Japanese, and similarly, one of the translated words in German is selected. As one'B
a l l /, 7 'boule' is selected as one of the translated words in Lance. After searching for all translated words for the first ideographic code, all translated words are also searched for the second ideographic code, and subsequently translated words are found for all the obtained ideographic codes.

In this way, one-to-many translation (translation of one input language into multiple languages) is possible.

Incidentally, the sentence translation is performed based on the 70- chart shown in FIG. That is, an input language text vi is first input, divided into words, and a dictionary search is performed for each word.

Based on the information obtained as a result of the dictionary search, syntax analysis is performed, and then text in the output language is generated. In this process, a dictionary in which the meaning is described using the ideographic code of the present invention can be used.

Now, in the above example, when searching for a translated word from an ideographic code, it is necessary to search for all the ideographic codes stored in the dictionary, but as an advantageous method, a reverse lookup dictionary is prepared to save the search effort. There are possible ways. 1st
Figure 2 shows the basic configuration of the reverse lookup dictionary. The ideographic code is stored as a headword, and a word or idiom is written after the ED delimiter.

If multiple words or idioms exist for one ideographic code, repeat the writing pattern of writing one word (or idiom) after the ■ symbol.

And finally, (D's delimiter symbol device.) Figure 13 shows an example of an English and Japanese dictionary equipped with this type of reverse dictionary.

14 represents an English dictionary, and 15 represents a Japanese dictionary. In the dictionaries 14 and 15 for each language, the part above the dashed-dotted line is the dictionary in the format shown in FIG. 2, and the part below the dashed-dotted line is the dictionary in the format shown in FIG. 12.

Next, Figure 13? A dictionary search operation using a reverse dictionary will be explained with reference to the following. English word input Xbal
l / Consider the case of translating into all Japanese.

When you input 'ball', the ideographic code becomes '101' according to the process described above (see flow chart in Figure 7).
0000000000011 ' etc. is obtained, and then the Japanese dictionary for the ideographic code ' 1010000000000011 ' is obtained! Search for reverse lookup dictionaries in 15. And headword' 1010000000000011
By reading out the contents of / and treating it as a whole delimiter symbol, we can obtain a translated word such as 1 ball'. In this way, by providing a reverse lookup dictionary, the task of searching for a translated word from an ideographic code can be performed quickly and easily. In this case, when performing translation between n languages, there are two dictionaries for each language: a dictionary for ideographic codes (or a translation part written in ideographic codes), and a reverse lookup dictionary for finding strings to be translated words from ideographic codes. A total of 21 [I dictionaries are required. On the other hand, in the conventional method, 2Xn
C! Therefore, this method is more advantageous when there are four or more languages.

In this method, it is sufficient to prepare as many ideographic codes as ``the number of words that can be translated between n languages''. For example, when translating from Japanese, which has many translations for rain, to English, the conventional method for translating ``vermicelli'' is 'Spring.
In this method, an ideographic code of 'Vermicelli' is provided.

Then, register @Spring Shower'' in the reverse lookup dictionary from the ideographic code to each language (English in this case). The method of storing the code in the output language may make the operation impossible because the expression method of the output language differs depending on each language.In other words, the ideographic code, which can be called a general expression, is used to express the concepts of all the languages necessary for use. It is preferable to prepare it so that it can be easily identified.

Even in that case, for example, considering that n bits in computer memory represent 2r#lJ (for example, 4 million for 32 bits), this method is advantageous because it is easy to operate. .

In addition to the fact that a single word or idiom can be expressed as multiple words or phrases as described above, a major problem in translation is that a single word has multiple meanings. This is a case where there are multiple ideographic codes corresponding to the word on the input side.

In this case, the method for resolving these ambiguity is the 11th
The 7 aids of syntactic analysis in the diagram include syntactic information (for example, rules that depend on each language, such as certain prepositions not modifying certain nouns), and semantic constraints (in other words, unless something has a will), In this case, in the conventional method, the semantic constraint rules are written and executed using some language-dependent code (in simple cases, However, in the method described in this invention,
The part that does not depend on each language can be commonly described as a constraint between ideographic codes, and even in the case of translation between n languages, only one part is required.

The embodiments of the present invention have been described above, and according to this method, the following effects can be obtained.

(1) The meaning of the dictionary is not described in the output language, but is described using pure semantic concepts, so it is suitable for translation between multiple languages. Tsumashi, if we apply the conventional dictionary format, n(
n is a natural number greater than or equal to 2) To translate languages: % 2X
nC* dictionaries are required, but if the dictionary of the form of the present invention is used, only n dictionaries need to be prepared. Therefore, the effort of creating a dictionary can be saved, and the capacity of the entire dictionary can also be reduced.

(2) Since the meaning description of the dictionary is performed using the ideographic code C2 (daily), it is possible to describe 2nWA semantic concepts with n (n is a natural number of 1 or more) bits of semantic information, reducing the dictionary capacity. Leads to.

(3) Furthermore, since the ideographic code is a binary bit string, it is easy to apply information compression techniques.

(4) Similar words can be easily searched using a coding method in which ideographic codes representing similar meaning concepts are set to "close location" (meaning "small distance between codes").

(5) Furthermore, by providing a reverse color dictionary, dictionary searches can be speeded up.

(6) Rules such as disambiguation of translations determined by semantic constraints can be written universally, independent of each language, and this part can be made independent of dictionaries that depend on each language. In addition, it is possible to write only one rule for translation between n languages.

[Brief explanation of the drawing]

Figure 1 shows the embodiment of the present invention. I! 2 is a diagram showing the basic format of the dictionary of the present invention, FIG. 3 is a diagram showing specific columns of the dictionary of the present invention, and FIG. 4 is a diagram showing how dictionary data is stored in memory. FIG. 5 is a diagram showing the operation flow of the electronic dictionary in FIG. 1, FIG. 6 is a diagram showing the initial screen on the display device of the electronic dictionary in FIG. ), Figure 7 (■) is a diagram showing the flow of processing to perform a dictionary search and obtain an ideographic code from a word, and Figure 8 (4)
), Figure 8 is a diagram showing the flow of processing to perform a dictionary search and obtain an ideographic code from an idiom, Figure 9 (4), Figure 9
Figure ■ shows the flow of processing to perform a dictionary search and obtain words and idioms from ideographic codes, Figure 10 shows the implementation of one-to-many translation, and Figure 11 shows the flow of processing for sentence translation. Figure 12 is a diagram showing the basic format of a reverse lookup dictionary, and Figure 13 is a diagram showing the basic format of a reverse lookup dictionary. Figure 7 (B) $ 8 Figure CB) Yyq Figure (A) Figure 9 (B)

Claims (1)

[Scope of Claims] 1. A device for translating words, idioms, or sentences written in a first language (hereinafter referred to as "words") into words, idioms, or sentences (hereinafter referred to as "words" below) in a second language; an input device for inputting words in the first language to be translated;
A dictionary memory that describes the correspondence between words in a first language and words in a second language in a predetermined manner, a processing unit that performs processing to search for a second language corresponding to an input word, and In the electronic dictionary device, the dictionary memory includes a device for displaying words in a second language written in the first language, and a device for displaying words in a second language written in the first language. The first segmented area describes the correspondence between the ideographic code representing the concept, and the correspondence between the coded characters forming the words written in the second language and the above ideographic code. A translation dictionary characterized by having a second segmented area. 2. A code for each character forming a word written in the first language, and an ideographic code representing the meaning concept of the word. The first segmented area describes the correspondence relationship between the coded characters forming the word written in the second language and the ideographic code representing the semantic concept of the word. In a search method for a dictionary memory having two divided areas, a first step compares an input word with a word in the first language and searches for an ideographic code corresponding to the matched word; A search method for a translation dictionary, comprising a second step of comparing the ideographic codes of the two divided areas and searching for a word in a second language corresponding to a match.