JP2500680B2 - Data name assignment registration device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data name assigning and standardizing device in a computer system which can standardize data names while maintaining uniqueness without the user's judgment.

[0002]

2. Description of the Related Art In machine translation systems, automatic keyword extraction systems, intelligent search systems for document databases, automatic report generation systems, document creation support systems, etc., data names are standardized. If you do not register it, the standardization of data names is important because users and readers will be confused by the fact that different terms are translated for the same object or different search documents are extracted. Is. Conventionally, a data name giving / registering device has been developed as a device for standardizing a data name. This data name assignment / registration device has standardized the data name by performing a compound word analysis using a dictionary or a grammar rule and then detecting a related phrase by a partial match search. Note that literatures relating to the data name assignment / registration device include, for example, William R.
"Data Resource Management" written by Shigeomi Ajimura, edited by Durrell, "Data Resource Management", Nikkei McGraw-Hill, Inc. Sho 62.1.27, Chapter 3, p. 57-.

[0003]

In the conventional data name assignment / registration apparatus, the user can select the target term based on a plurality of word decomposition candidates obtained by compound word analysis using grammatical rules. Must be properly selected. However, since it is difficult to set the selection criteria, various selections are made by people. Also, in the partial match search used when determining the structure of words, it is necessary to output unnecessary large numbers of candidates in order to extract related words regardless of the classification and appearance position of words that are the search conditions. become. For this reason, when the data name is assigned by using the conventional data name assigning / registering device, there are places where different candidates are selected depending on the judgment of the user, and in a large-scale system, the whole is not selected. There was a problem that the uniqueness of could not be maintained.

An object of the present invention is to solve such conventional problems, prevent different candidates from being selected by a user's judgment, and increase the capability of the apparatus to improve operability and efficiency. It is to provide a data name giving / registering device capable of performing.

[0005]

In order to achieve the above object, the data name assignment / registration apparatus of the present invention comprises a basic word table (7 in FIG. 1) defining a word to be used and a classification of the word, and a composite word. A compound word table (8 in FIG. 1) that describes the constituent words related to the word,
A table of word construction rules for standardizing data items (see FIG. 1).
9), a data item management table (10 in FIG. 1) for storing existing data items, and a standard word correspondence table (11 in FIG. 1) in which the relationship between similar terms is arranged. A phrase decomposition module (1 in FIG. 1) that decomposes an input phrase based on a basic word table and a compound word table into basic words, and when the decomposition by the phrase decomposition module fails, the input phrase is a basic word. The unknown word registration module (2 in FIG. 1) to be registered in the table, the classification information about the input words, and the word structure rule table based on the arrangement of the input words are used to determine whether the input words are valid as data items. When it is determined that the word structure confirmation module to be confirmed (3 in FIG. 1) or the word structure confirmation module violates the rules, the data item management table is searched, and the data item management table is deleted based on the prioritized classification. Similar word search to extract information about existing words Ju - Le (4 in FIG. 1), a base word in the input word satisfying the phrase structure rules, the phrase standardized module mapped to the standard terminology of the standard language in the corresponding table - Le (Fig. 1
5), and a data item generation module (6 in FIG. 1) for accumulating standard data items corresponding to input phrases in the data item management table.

[0006]

In the present invention, with respect to the words and phrases forming the data items, a basic word table accumulating words and phrases of a related field,
By utilizing the compound word decomposition table, the decomposition candidate of the input phrase can be made the minimum number, that is, almost unique. In addition, by utilizing the classification information in the basic word and giving a priority order to the classification of words and the position where the words appear, as a word construction rule that captures the structural characteristics of words related to compound words, It is possible to set a clear assignment standard regarding the generation of data items. In addition, by searching the data item table generated by the phrase construction rule using the proposed similar phrase search module, extraction of related data items (phrases) is facilitated and the input phrase The correction can be made reliably. Further, by sequentially accumulating the data items output by the data item generation module in the data item management table, the knowledge amount can be increased and the system can be changed to a more intelligent registration system.

[0007]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 is an overall block diagram of a data name assignment / registration apparatus showing an embodiment of the present invention. The data name assignment / registration apparatus of the present invention is, as shown in FIG.
Module 1, unknown word registration module 2, word structure confirmation module
Module 3, similar phrase search module 4, phrase standardization module
6 program modules of data 5 and data item generation module 6, basic word table 7, compound word table 8, word structure rule table 9, data item management table 10, and standard words The correspondence table 11 is composed of five table tables. The basic word table 7 is composed of the basic words of the world to be input and information indicating the classification thereof, the compound word table 8 is composed of compound words such as abbreviations and their constituent word strings, and the standard word correspondence table 11 is It consists of a table summarizing the relationships between similar terms.
It is a rule provided to realize the standardization of data items, which is composed of rules for forming words and phrases, but may include a prohibited word table that defines prohibited words that should not be used. De-
The data item management table 10 is provided to store existing data items. The word decomposition module 1 refers to the basic word table 7 to extract the composed words and the classification information thereof from the input words input by the user. If the input phrase has a compound word, it is decomposed into basic words by the compound word table 8. When a forbidden word or an unknown word is detected, a new word is registered by the unknown word registration module 2 and then the input word is instructed to be re-input. The phrase construction confirmation module 2 uses the phrase construction rule table 9 to confirm that the arrangement of the classification information of the phrases in the input phrase conforms to the rule.
The similar word search module 4 searches for existing data items stored in the data item management table 10 and corrects or corrects them when a structure error is detected by the word structure confirmation module 3. The analogy of the words to be added is made, and the input word is re-input. If the input word order conforms to the word construction rules, the word construction module 5 searches the standard word correspondence table 11 in which the relationships between similar terms are arranged, and standardizes the input words. Finally, the data item generation module 6 stores the corresponding input phrase in the data item management table 10.

FIG. 7 is an operation flowchart in FIG. When the user inputs a word or phrase into the data name assignment / registration device of the present invention, first, the word / phrase division module 1 collates with the word / phrase held in the basic word table 7, and the corresponding word / phrase and its classification. Is extracted (step 101). The word or phrase that fails to match with the basic word table 7 is interpreted as an unknown word and sent to the unknown word registration module 2 (step 102). In the unknown word registration module 2, the unknown word and the classification information set by the user are registered in the basic word table 7 (step 10).
3). In step 101, when the classification is a compound word (step 104), it is decomposed into the constituent words by the compound word table 8 and is input again to the phrase division module 1. If the unknown word is a compound word, the user adds a word string constituting the compound word and registers it in the compound word table 8 (step 10).
5). In the phrase division module 1, the word string is collated with the basic words in the basic word table 7, and if they do not match, the above process is repeated (steps 101 to 105). When the matching with the basic word in the basic word table 7 is successful, the new registration of the phrase is completed. The input words and their classification information that have been normally processed in the basic word table 7 are sent to the word structure confirmation module 3. The word / phrase structure confirmation module 3 checks whether or not the input word / phrase satisfies the structure condition as a data item using the classification information and the word / phrase structure rule table 9 based on the classification information and the classification information (step 106). If the configuration conditions are not met, a message to that effect is output, and
The analysis information is sent to the similar phrase search module 4. In the similar phrase search module 4, the similar search is executed for the data items in the data item management table 10 which are already held in accordance with the priority order of the phrases, and the candidate of the related phrase is extracted. Output (step 107). The user tries to input again into the phrase division module 1 based on this information. On the other hand, the input phrase satisfying the condition of the data item in the phrase structure confirmation module 3 is sent to the phrase standardization module 5. In the phrase standardization module 5, the standard word of the corresponding input phrase is extracted based on the standard word correspondence table 11 in which the relationships between similar terms are organized (step 108). The sequence of standard words corresponding to this data item is the data item generation module.
Sent to Le 6. In the data item generation module 6, the standard data are replaced by replacing the corresponding terms with standard words.
A data item is generated and is registered in the data item management table 10 together with the corresponding data item (step 109).

FIG. 2 is a diagram showing terms and examples of classifications thereof stored in the basic word table in FIG. 1, and FIG. 3 is a diagram of constituent words also stored in the compound word table, and FIG.
Is a diagram of rules also stored in the phrase construction rule table, FIG. 5 is a diagram of terms and standard terms also stored in the standard word correspondence table, and FIG. 6 is also a data item management table. FIG. 6 is a view of the same data item held in the. The basic word table 7 holds terms and categories as shown in FIG. It is assumed that the compound word table 8 holds compound words and word structures as shown in FIG. 2, the terminology, the main term,
The classification of the modifiers is utilized in the phrase construction rule table 9. Now, the operation when the user inputs the phrase "subscriber telephone number" will be described. Phrase division module 1
Then, by collating with the terms in the basic word table 7 in FIG. 2, (subscriber, main word) and (telephone number, compound word) are obtained. Since a compound word was found, it was compared with each compound word in the compound word table 8 in FIG.
After obtaining (telephone number: telephone, number), it is collated again with the terms in the basic word table 7 in FIG. As a result, (subscriber, main word), (telephone, main word), (number, section word) are obtained for the "subscriber telephone number". Since no unknown word was detected,
Next, the above information is input to the phrase composition module 3. In the phrase construction module 3, it is confirmed whether or not the "subscriber telephone number" conforms to the phrase construction rule table 9 in FIG. "number"
Since the delimiter word appears only once and at the right end,
Satisfy rule 1. Rule 2 is satisfied because the main words "telephone" and "subscriber" appear twice to the left of the delimiter. In addition, since no modifier appears, Rule 3 is also satisfied. Since all the rules are satisfied, the word structure is recognized as correct, and the analysis information is input to the next word standardization module 5. The phrase standardization module 5 collates the terms in the standard word correspondence table 10 in FIG. 5 and extracts the corresponding standard terms. In the example of "subscriber telephone number", (customer, telephone, number) is extracted as a standard term. In the next data item generation module 6, the standard customer data item name "customer telephone number" and the input phrase (data item name)
"Subscriber's telephone number" is sent. Data item generation module
In the rule 6, the data item management table 11 of FIG. 6 is collated, and the corresponding standard data item "customer telephone number" already exists, so this is not registered. Next, it is checked whether or not "subscriber telephone number" exists in the same data item. In this case, since there is no corresponding item in FIG. 6, "subscriber telephone number" is added to the same data item column, and the registration process is completed.

As described above, according to the present invention, since the phrase division module 1 and the unknown word registration module 2 are set, the registration of the term relating to the target field and the setting of the classification standard thereof are prepared and input. It is also possible to efficiently perform the analysis of the generated compound phrase. Further, since the phrase confirmation module 3 and the similar phrase search module 4 are provided, even for incompatible input phrases, similar search is performed for existing data items created regularly, Information about the related word and its appearance position can be obtained, and the standard data item can be easily generated. Further, since the phrase standardization module 5 is provided, the unregistered data decreases and the capability of the registration device increases as the number of terms and data items held increases.

[0011]

As described above, according to the present invention,
Standard data items can be easily generated, different candidates are not selected according to the judgment of the user, the capability of the registration device can be increased, and the operability and efficiency of the data name assignment registration device are improved. Can be made.

[0012]

[Brief description of drawings]

FIG. 1 is an overall block diagram of a data name assignment / registration apparatus showing an embodiment of the present invention.

2 is a diagram showing terms held in a basic word table in FIG. 1 and examples of classification thereof.

FIG. 3 is a diagram showing an example of terms held in a compound word table in FIG.

FIG. 4 is a diagram showing an example of a word structure rule table in FIG. 1.

5 is a diagram showing an example of a standard word correspondence table in FIG.

FIG. 6 is a diagram showing an example of a data item management table in FIG.

7 is an operation flow chart of the data name assignment / registration apparatus of FIG.

[Explanation of symbols]

 1 word division module 2 unknown word registration module 3 word structure confirmation module 4 similar word search module 5 word standardization module 6 data item generation module 7 basic word table 8 compound word table 9 Word composition rule table 10 Data item management table 11 Standard word correspondence table

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Mitsuru Kawashita 1-1-6 Uchisaiwaicho, Chiyoda-ku, Tokyo Inside Nippon Telegraph and Telephone Corporation (56) Reference JP-A-57-137965 (JP, A) JP-A 2-176865 (JP, A)

Claims (1)

(57) [Claims] 1. A word / phrase is input and a standard data is input based on the word / phrase.
In a data name assignment / registration device that generates data item names, a standard word table that defines the words to be used and the classification of the words, a compound word table that describes the constituent words related to the compound words, and standardize the data items It is equipped with a word structure rule table for storing, a data item management table for storing existing data items, and a standard word correspondence table in which relationships between similar terms are arranged. A word decomposition module that decomposes an originally input word into basic words, an unknown word registration module that registers the input word in the basic word table when the decomposition by the word decomposition module fails, A phrase structure confirmation module for confirming whether or not the input phrase is valid as a data item according to the classification information relating to the input phrase and the phrase structure rule table based on the arrangement of the input phrase, and the phrase structure confirmation By module When it is determined that the violation of the rule,
The similar word search module for searching the data item management table to extract information about the missing word based on the prioritized classification, and the basic word in the input word satisfying the word construction rule as described above. A standardized module for words and phrases corresponding to standard terms in the standard word correspondence table, and a data item generation module for accumulating standard data items corresponding to the input words and phrases in the data item management table are provided. A data name giving / registering device characterized by the above.