JPS6389976A - Language analyzer - Google Patents

Abstract

PURPOSE:To analyze a various kinds of character strings without storing a large amount of data, by retrieving a fundamental dictionary means and analyzing the character string for another part of the character string similarly in a case that the fundamental dictionary means is retrieved, and a part of the character string is retrieved. CONSTITUTION:A processing part 16 divides an English text inputted from an input part 14 into retrieval key character strings, and retrieves them first by a dictionary file 22 through a dictionary retrieving part 20, and when no entry to the file 22 exists, recognizes a unit, and outputs the retrieval key character string to a unit recognizing part 24. The recognizing part 24 divides the retrieval key character string, and retrieves a fundamental dictionary file 26 by every divided character string by indicating with a pointer P, and recognizes the character string recorded on the file 26, or the one consisting of the continuity of the recorded string as the character string expressing the unit. Therefore, for example, it is enough to store only (k), (m), or (s), etc., in the file 26, and it is not necessary to store the combination of them, km, km/s, etc., thereby, it is possible to reduce the capacity of the file.

Description

TECHNICAL FIELD The present invention relates to a language analysis device, particularly to a language analysis device useful for automatic translation devices.

Prior Art For example, when creating a corresponding Japanese sentence from a sentence in a foreign language such as English, the morphemes of the input English sentence are analyzed,
It analyzes the syntax, converts the sentence structure, and then generates a Japanese translation.

That is, by searching a dictionary, the morphemes of each word constituting the input sentence are analyzed, and information such as the part of speech of these morphemes is obtained. Then, based on the obtained information such as parts of speech, the structure of the input sentence is analyzed by analyzing the modifying relationships between each word or the blocks formed by these words using grammatical rules. , Furthermore, based on the parsed syntax, the sentence structure of the input sentence is converted into the order of the Japanese syntactic arrangement, and Japanese morphemes are generated according to the converted arrangement to generate a Japanese sentence.

In the morphological analysis performed when generating a translated sentence, the part of speech and other information of morphemes such as words are obtained by searching a dictionary. In the case of ordinary words such as nouns and verbs, many of them can be stored in a dictionary, so they can be easily searched and information obtained.

However, since there are so many types of expressions for units such as length, velocity, acceleration, etc., storing them all in a dictionary will unnecessarily increase the storage capacity of dictionary information. , is inefficient. This is because many of these units are complex expressions that are a combination of multiple unit expressions, such as m/s, km/s, etc.

However, in the case of a device that stores only some of these units in a dictionary, it is not possible to obtain information on the units of these complex expressions contained in the input sentence, making it impossible to perform morphological analysis. , there was a risk of incorrect language analysis.

Purpose The present invention solves the drawbacks of the prior art, and makes it possible to perform morphological analysis of input sentences containing strings of complex expressions without having to store all strings of complex expressions in a dictionary. The purpose is to provide a language analysis device that can perform

Structure In order to achieve the above object, the present invention includes an input means for inputting a character string in a predetermined language, a basic dictionary means for storing basic data and used for searching the input character string,
and an analysis means for analyzing a character string by searching the basic dictionary means for the input character string, and the analysis means analyzes a part of the character string by searching the basic dictionary means for the input character string. When a character string is retrieved, the character string is analyzed by similarly searching the basic dictionary means for other parts of the character string. Hereinafter, a detailed explanation will be given based on one embodiment of the present invention.

FIG. 1 shows an embodiment in which a language analysis device according to the present invention is applied to an English-Japanese visual translation device. Note that the present invention can be used for purposes other than morphological analysis of English input sentences, and can be used not only as an English-Japanese translation device that translates English into Japanese.
Needless to say, the present invention can also be effectively applied to an automatic translation device that translates one language into another language.

The wood flow side has an input section 14 into which data is input from the input device 10 or the input document file 12. The input device 10 includes, for example, a keyboard having character keys such as alphanumeric keys, function keys, etc., an optical character reading device for reading alphabetic text recorded on paper, etc. The input document file 12 is stored in a magnetic disk or the like. A storage device that records English text on a medium.

The input unit 14 has an input character string buffer 14a, and stores an English input sentence input from the input device 10 or the input document file 12 in the input character string buffer 14a. The input unit 14 reads the input sentence stored in the input character string buffer 14a and outputs it to the processing unit 16.

The processing unit 18 searches the dictionary file to input the input unit 1.
This is a functional unit that performs morphological analysis of the input sentence sent from 4. The processing unit te has a dictionary information storage table lea, and stores a dictionary file 22 or a basic unit dictionary file 2, which will be described later.
The information obtained by searching 6 is stored in the dictionary information storage table lea.

The processing unit 18 searches for a search key character string that is a unit when searching a dictionary from the character strings that constitute the input sentence input from the input unit 14 . The search for the search key string is performed in accordance with a predetermined search rule, starting from the first character of the string constituting the input sentence.

For example, an input sentence is divided sequentially from the beginning of the sentence using delimiters such as spaces and commas, and each of the divided character strings is used as a search key character string. In this case ■, km, mis
Character strings representing units such as , etc. are each used as a search key character string. The processing unit 1G sends the search key character string searched from the character strings forming the input sentence to the dictionary search unit 20.

The dictionary file 22 is searched based on the character string. As shown in FIG. 2, the dictionary file 22 stores entries and grammatical information such as parts of speech. If there is an entry in the dictionary file 22, the dictionary search unit 20 reads part-of-speech information, etc. of the entry, and outputs this to the processing unit 1B. As a result of searching the dictionary file 22, the dictionary search unit 20 outputs a notification to that effect to the processing unit 1B if there is no entry in the dictionary file 22.

The processing unit 1B stores the part-of-speech information etc. searched by the dictionary search unit 20 in the dictionary information storage table lea. If there is no entry for the search key string in the dictionary file 22, the processing section 1B stores the search key string in the unit recognition II section 2.
Output to 4.

The unit recognition unit 24 searches the basic unit dictionary file 2B based on the search key character string sent from the processing unit 1B. The basic unit dictionary file 28 stores basic unit entries as shown in FIG. 883. If there is a basic unit entry in the basic unit dictionary file 2B, the unit recognition unit 24 reads out the basic unit entry. If there is no entry in the basic unit dictionary file 2B, the unit recognition unit 24 reads out a plurality of search key character strings as described later. search the basic unit dictionary file 2B multiple times, and if there are basic unit entries in the basic unit dictionary file 2B in the multiple searches, complex unit information is obtained from these basic unit entries. . If there is no basic unit entry in any of the multiple searches, information indicating that the word is not registered in the dictionary is obtained.

Unit recognition i! 1ffi24 sends the basic unit entry, composite unit information, and information indicating that the word is not registered in the dictionary to the processing unit IB.
Output to. The processing unit IB stores this information input from the unit recognition unit 24 in the dictionary information storage table IEla.

As shown in FIG. 4, the dictionary information storage table 18a stores and stores entries for search key strings and grammatical information such as parts of speech obtained by searching the dictionary file 22 or basic unit dictionary file 26 for the search key string. do. Processing section 16
After these data are stored in the dictionary information storage table 18a, these data are output to the output interface 1B together with the input sentence. Output interface 1
8 outputs the input sentence and morphological analysis data output from the processing unit 1B to an output device 30 such as a printer or a display.
Alternatively, it is output to a storage file 32 such as a magnetic disk.

Or, without providing the output interface 18,
The input sentence and morphological analysis data output from the processing unit 16 are directly input to a syntax analysis means (not shown), the input sentence is parsed by the syntax analysis means, and a translated text is generated based on the syntax analysis. You may also do so.

The control section 2B controls the operation of each functional section of this device, and is advantageously configured by a microprocessor.

The operation of this device will be explained with reference to the flowchart shown in FIG.

First, an English input sentence is read into the input unit 14 from the input device 10 or the input document file 12 (100), and the input sentence read into the input unit 14 is transferred to the input character string buffer 14.
It is stored in a. The input sentence stored in the input character string buffer 14a is read out and output to the processing unit 18.

In the processing unit 1B, when an input sentence is input, extraction is performed in dictionary lookup units (102). In other words, the character strings that make up the input sentence are
It is determined whether or not the search key string, which is the unit when searching the dictionary file 22 or the basic unit dictionary file 2B, has a search key string that is divided into zeros, which are divided sequentially from the beginning of the string (104). , in some cases, sends the search key string to the dictionary search unit 20.

When a search key string is sent to the dictionary search section 20, the dictionary search section 20 searches the dictionary file 2 for this search key string.
2), it is determined whether or not there is a search key character string in the entry of the dictionary file 22 as shown in FIG. The grammatical information, such as the part of speech, is read out, and the read data is sent to the processing unit 16 and recorded in the dictionary information storage table 16a (110).Then, the process returns to step 102 and the dictionary lookup unit is cut out again.

If there is no entry in the dictionary file 22, the dictionary search section 20 sends the search key string back to the processing section 1B, the processing section 1B sends this search key string to the unit recognition section 24,
Unit recognition is performed in the unit recognition unit 24 (112)
.

When the search key string sent to the dictionary search unit 20 is a word such as a normal noun or verb, there are almost always entries in the dictionary file 22, so grammatical information such as part of speech is read from the dictionary file 22 and this data is is sent to the processing unit 1B and recorded in the dictionary information storage table tea. Dictionary file 2
2, as mentioned above, entries for ordinary words such as nouns and verbs are formed, but entries for character strings representing units are not formed. Therefore, the search key string represents units such as km, ■/S, etc. If it is a character string, the dictionary file 22
Since there is no entry, the process proceeds to step 112, where the unit is recognized.

The unit recognition operation in step 112 will be explained with reference to FIG.

In the search of the dictionary file 22, the search key string for which no entry exists in the dictionary file 22 is searched by the processing unit 1B.
When the search key character string is sent to the unit recognition unit 24, the unit recognition unit 24 sets a pointer P to the first character of the search key character string (200).

Next, the unit recognition unit 24 searches the basic unit dictionary file 26 for a character string starting from the character to which the pointer P is set (202). This search is performed to determine whether the basic unit whose entry exists in the basic unit dictionary file 2B appears as a complete character string in a character string starting from the character set by pointer P, and whether the character set by pointer P is the starting point. Search for whether or not. That is, in this search, a character string of one or more characters starting from the character to which the pointer P is set is found in the basic unit dictionary file 2B.
Search to see if it matches any of the basic units in which an entry exists. For example, if the character to which the pointer P is set is k, sho, S, etc., the basic unit for these characters starting from the character to which the pointer P is set is shown in Figure 3. An entry exists in the dictionary file 26.

The unit recognition unit 24 determines whether or not an entry exists in the basic unit dictionary file 2B as a result of searching the basic unit dictionary file 2B (204), and if the entry exists, the length of the recognized basic unit is determined. The pointer P is advanced by that amount (208). Therefore, when the basic unit is k, otori, S, etc., the pointer P is advanced by one character and set to the next character in the search key character string.

The unit recognition unit 24 determines whether there are any further character strings starting from the character to which pointer P is set (
208). If there are more character strings like this, the process returns to step 202, and the basic unit dictionary file 2 is rewritten with the character string starting from the character to which the pointer P is set.
Search for 8. Then, as a result of searching the basic unit dictionary file 26, it is determined whether an entry exists in the basic unit (204), and if an entry exists, the pointer P is advanced by the length of the recognized basic unit. .

In step 208, if there is no longer a character string starting from the character to which the pointer P is set, it means that the search in the basic unit dictionary file 2B has been completed and the recognition of the compound unit has been successful.

For example, if the search key string sent to the unit recognition unit 24 is km/s, which represents the unit, an entry exists in the basic unit dictionary file 2B because the unit is a complex unit. do not. Therefore, first set the pointer P to k200), and set k to the basic unit dictionary file28.
The existence of the entry is confirmed by searching (202).

Next, the pointer P is set to l (20B), the layer is searched by the basic unit dictionary file 28 (202), and the existence of the entry is similarly confirmed. Since the unit recognition unit 24 considers slash l, bullets, etc. to be part of the unit, next it skips l in km/s and sets pointer P to S (2H). Then, search for S using the basic unit dictionary file 26 and confirm the existence of the entry in the same manner (202).
, As a result, since entries for all of k, m, and S were found by searching the basic unit dictionary file 2B, it is determined that k-/S is a character string representing a unit.

In this way, when entries exist in the basic unit dictionary file 2B for all the characters that make up the search key character string.

Or, if there are entries in the basic unit dictionary file 2B for all characters other than symbols that are considered to be part of units such as slashes and bullets, the search key string is determined to be a string representing a unit. be done.

When the unit recognition unit 24 finishes searching the basic unit dictionary file 26 and successfully recognizes the composite unit, it sends the obtained unit information to the processing unit 1B and stores it in the dictionary information storage table 18a (210). This completes unit recognition.

In step 204, as a result of searching the basic unit dictionary file 28 for a character string starting from the character to which the pointer P is set, if there is no entry in the basic unit dictionary file 2B, the character string in l is used as the basic unit dictionary file 28. Since this means that it could not be recognized as a unit or a compound unit, the unit recognition section 24 sends information that this character string is a word not registered in the dictionary, that is, information that it does not represent a unit, to the processing section 1B. , by storing it in the dictionary information storage table lea of the processing unit 18 (212), the intermediate recognition is completed.

Returning to FIG. 5, when the unit recognition (112) is completed,
Returning to step 102, the processing section 16 performs the extraction of each dictionary lookup unit again.

After extracting the dictionary lookup units, the processing unit 16 determines whether there are any more extracted units (104), and if there are no more extracted units, that is, search key strings,
The information stored in the dictionary information storage table lea,
It is output to the output device 30 through the output interface 18 (114). This completes the analysis of the input sentence.

As described above, according to this embodiment, an English input sentence is divided into search key character strings, first searched using the normal dictionary file 22, and when there is no entry in the dictionary file 22, unit recognition is performed. In unit recognition, the search key string is divided and indicated by pointer P, the basic unit dictionary file 2B is searched for each divided string, and the basic unit dictionary file 2B is searched for the one recorded in the basic unit dictionary file 28 or the basic unit. A string consisting of consecutive items recorded in the unit dictionary file 26 is determined to be a character string representing a unit.

Therefore, even if the character string represents a complex unit, the unit can be recognized by combining the basic units stored in the basic unit dictionary file 26.
Analysis can be performed in response to a variety of unit expressions. Moreover, '' only needs to be stored in the basic unit dictionary file 2B, such as only certain basic units, such as +w, s, etc.
Complex units that combine these, such as ni■, ni■/
Since there is no need to store S, etc., the capacity of the dictionary file can be reduced.

Effects According to the present invention, the basic dictionary means is searched for an input character string, and when a part of the character string is retrieved, the basic dictionary means is similarly searched for other parts of the character string, and the character string is searched. Analyze. Therefore, many types of character strings can be analyzed without storing a large amount of data in the basic dictionary means.

[Brief explanation of the drawing]

Figure 1 is a block diagram showing an embodiment of the language analysis device according to the present invention, Figure 2 is a diagram showing an example of data stored in the dictionary file of Figure 1, and Figure 3 is the basic unit of Figure 1. FIG. 4 is a diagram showing an example of data stored in the dictionary information storage table of FIG. 1. FIG. 5 is a flowchart showing the operation of the device of FIG. 1. . FIG. 6 is a flowchart showing unit recognition in the operation shown in FIG. 585. 18. Processing unit 20, Dictionary search unit 22, Dictionary file 24, Unit recognition unit

Claims (1)

[Scope of Claims] 1. Input means for inputting a character string in a predetermined language; basic dictionary means used for searching the input character string and storing basic data; and the input characters. an analysis means for analyzing the character string by searching the basic dictionary means for the input character string, the analysis means analyzing the character string by searching the basic dictionary means for the input character string. 2. A language analysis device characterized in that when a part of the character string is retrieved, the basic dictionary means is similarly searched for other parts of the character string to analyze the character string. 2. In the device according to claim 1, the basic dictionary means is a basic unit dictionary means for storing data representing units, and the analysis means is configured to analyze the input character string from the basic unit dictionary. A language analysis device characterized in that it analyzes whether or not the character string represents a unit by searching for a means. 3. In the apparatus according to claim 2, the analysis means searches the basic unit dictionary means, and, as a result of searching the basic unit dictionary means, the character strings are combinations of character strings representing units stored in the basic unit dictionary means. A language analysis device characterized in that a character string is determined to be a character string representing a unit when the character string consists only of characters. 4. In the device according to any one of claims 1 to 3, the analysis means has a pointer,
When a part of the character string is retrieved by setting the pointer to the first character of the input character string and searching the basic dictionary means for a character string starting from the character to which the pointer is set. , a language analysis device characterized in that the pointer is set to a character string following a part of the searched character string, and the basic dictionary means is searched for the character string to which the pointer is set. 5. In the device according to any one of claims 1 to 4, the input character string is searched by a normal dictionary means and is not stored in the dictionary means. A language analysis device characterized by: