JPH0793362A - Data base retrieval system - Google Patents

Abstract

PURPOSE:To perform retrieval that a user intends by stating information on a vague word and asking a user back when which data base item the word in a retrieval sentence corresponds to is uncertain, and storing, reflecting, and deleting contents that the user selects as a history. CONSTITUTION:A correspondence table selecting means 27 narrows down a table, group by group, according to information on a syntax analysis result, but when all vagueness is not narrowed down, an asking-back device 29 states the information on the vague word to make the user make a choice. A history storage device 30 stores the asking-back result as a history and if similar vagueness appears next, a history reflecting device 28 employs it. Further, a history deleting device can deletes the history as the user intends before retrieval.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information retrieval system.

[0002]

2. Description of the Related Art FIG. 19 shows, for example, the following document (Takashi: Database for using correspondence table, Japanese Patent Laid-Open No. 4-348470).
Is an information retrieval method for a tabular database in natural language. In the figure, 16 is an example of a search sentence, 1
Reference numeral 7 is an input device for inputting a search sentence, 18 is an interactive control device for receiving and processing an input sentence, 21 is term information including a term dictionary, and 19 is an input sentence divided into words with reference to the term information 21. A morphological analysis device, 22 is a grammar table, and 20 is a syntactic analysis device that converts a search sentence input from the input device 17 into a syntactic analysis result 23 using the term information 21 and the grammar table 22. Reference numeral 200 is a collating unit that correlates the term information 21 and the syntactic analysis result 23 with the correspondence table 24. 26 is a grouping unit that classifies words in the input sentence by particles of the input sentence or the corresponding table. 27 is a case where there are a plurality of combinations of tables. Is a correspondence table selecting means for narrowing down the correspondence table for each group. 50 is a database search formula generation unit that generates a database search formula from the result of the collation unit; 51 is a search device that executes a database search using the database search formula;
Reference numeral 52 is a database to be searched.

Next, the operation will be described. When the search sentence 16 is input from the input device 17, the syntactic analysis device 20 uses the term information 21 and the grammar table 22 to obtain the syntactic analysis result 2
3 is obtained. In the term information 21, in addition to the definition of the part of speech, meaning, etc. of each word, the correspondence table corresponding to each word and the information of that item are defined, and a plurality of items can be defined for one word. Is. Next, the collating unit 20
Before selecting the correspondence table, the grouping means 26 of 0 divides each clause into groups according to the information of the syntax tree, and makes them correspond to expressions such as "from to" of comparison and time. When a plurality of items correspond to one word and there is ambiguity, the correspondence table selection means 27 preferentially selects a correspondence table containing a large number of words in the verb correspondence or syntax in the table.

[0004]

Since the conventional apparatus is configured as described above, the correspondence table selecting means 27 preferentially selects a verb correspondence table when a verb exists, and a plurality of words can be selected by one word. When it corresponds to the items of the table, the correspondence table is determined by preferentially selecting the correspondence table containing many words in the syntax in the table. Therefore, there is a problem that the correspondence table cannot be determined when the number of words included in the table is the same or when the same table has ambiguity such as proper nouns.

The present invention has been made to solve the above problems, and when the correspondence table cannot be determined due to ambiguity in the selection of the correspondence table, a part of the term information of the ambiguous word is used. Is displayed and the user is queried, the correspondence table is determined according to the contents selected by the user, and the search is performed. Also, the contents selected by the user are stored as history data,
The purpose is to allow the user to delete the history once stored, or to manage the history for each identifier so that the search can be performed according to the user's intention.

[0006]

A database search system according to a first aspect of the present invention includes an input means for inputting a natural language, a tabular database having a plurality of database items for storing data, and a natural language. And a correspondence table storage means for storing a correspondence table for generating a database search expression, a term information storage means for storing term information including information for associating words with the correspondence table, and analyzing the input natural language. And a parsing unit that outputs the parsing result, and a collating unit that determines the correspondence table using the term information from the parsing result,
From the result of the collation unit, in the database retrieval system including a database retrieval formula generation unit that generates a database retrieval formula using the correspondence table, and a database retrieval device that performs a database retrieval by the database retrieval formula, the collation unit is When selecting a correspondence table using the above term information, if the word is an ambiguous word that corresponds to multiple correspondence tables and the database item cannot be specified, the content of the term information for that ambiguous word is presented and selected. It is characterized in that it is provided with a question and answer device.

The term information in the database search system according to the second invention has information for associating a word with a correspondence table divided into several stages according to items, and the inquiry device, when an ambiguous word occurs, It is characterized in that the data necessary for narrowing the ambiguity is presented to the user step by step from the term information, and the user decides based on the data.

A database retrieval system according to a third aspect of the present invention is, when a plurality of database items corresponding to one word are ambiguous, presenting a part of the term information of the ambiguous word to the user for asking the question and answering device. And a history storage device for storing the selection result of the inquiry as a history, and a history reflection device capable of automatically reflecting the stored history when the ambiguity occurs again to eliminate the ambiguity. Characterize.

A database retrieval system according to a fourth aspect of the present invention is, when a plurality of database items corresponding to one word are ambiguous, presenting a part of term information of the ambiguous word to the user for asking a question and answering device. And a history reflection device that can store the selection result of the question and answer as a history and automatically resolve the ambiguity when the ambiguity occurs again, and the history according to the user's intention. It is characterized by having a history deleting device that can delete.

The history reflection device in the database search system according to the fifth aspect of the present invention is adopted if the latest history is found out from among the histories stored in the history storage device that have the same ambiguous word, and the word is ambiguous. It is characterized by specifying a database item corresponding to a word.

The history reflection device in the database retrieval system according to the sixth aspect of the present invention examines the history stored in the history storage device from the newest one, and adopts the one having the same ambiguous word, which is adopted. If not, next check the new history. In this way, the feature is that the database item corresponding to the ambiguous word is specified using the history.

A history deleting device in a database search system according to a seventh aspect of the present invention is characterized in that all or all of the histories stored in the history storing device are deleted.

A database search system according to an eighth aspect of the present invention is characterized in that, in a history storage device and a history reflection device, an identifier is attached to history information, and a history is stored / used for each identifier.

A database search system according to a ninth aspect of the present invention further comprises a history deletion device for deleting the history stored in the history storage device by using the above identifier.

[0015]

In the database search system according to the first and second aspects of the present invention, the term information about the word is grouped into several stages according to items for the word whose ambiguity has not been resolved by the processing before the inquiry device. Then, effective information for narrowing down the ambiguity for each item is sequentially presented as options, and the user is queried back to determine the ambiguous word according to the intention of the user.

In the database search system according to the third aspect of the present invention, the history storage device stores the result of the inquiry, that is, the ambiguous word and the selected information as a history. When ambiguity occurs again, if the history of the same ambiguous word is stored by the history reflecting device, the history data is reflected in the determination of the ambiguous word.

In the database retrieval system according to the fourth aspect of the present invention, the history deleting device can delete the history that the user does not want to reflect.

The database search system according to the fifth aspect of the present invention enables the history to be searched in order from the latest when the history is reflected, and when only the latest history is to be adopted, the latest available history is automatically selected. To resolve the ambiguity.

The database search system according to the sixth aspect of the present invention enables the history to be searched in order from the latest when the history is reflected, and the history to be adapted is checked in order from the latest history. If there is something, it will be reflected automatically and the ambiguity will be resolved.

The database search system according to the seventh aspect of the present invention can delete and initialize all the history stored by the history deleting device according to the intention of the user. Alternatively, the stored history can be displayed on the screen, and the user can view the stored history and select the history to be deleted.

The database search system according to the eighth aspect of the present invention can have a history for each identifier by adding an identifier when storing the history of inquiry, and the history contents for each identifier can also be reflected when reflecting the history. Can be used.

In the database retrieval system according to the ninth aspect of the present invention, by adding an identifier when storing the history of inquiry, it is possible to delete only the history having the same identifier when deleting the history.

[0023]

Embodiments of the present invention will be described below with reference to the drawings. It should be noted that substantially the same elements as those in the conventional example are denoted by the same reference numerals, and description thereof will be omitted.

Example 1. FIG. 1 is a diagram showing an inquiry device according to this embodiment. The Japanese interrogation device is similar to the conventional ones: 16, 17, 18, 19, 20, 21,
Has 22, 24, 50, 51, 52. The collating unit 25 has 26 and 27, which are almost the same as those of the conventional one, and then has an inquiry returning device 29 for asking the user when there is an ambiguous word.

Next, the operation will be described.

"Suzuki" in the search sentence "Suzuki's sales" in FIG. 1 is ambiguous because it has a plurality of term information as shown in FIG. 3, and narrowing down is not possible even by the correspondence table selecting means 27. Is. This "ambiguity" occurs because "Suzuki" represents the product name "Suzuki-kun" when it is a proper noun such as "Suzuki", or represents the shipping destination "Suzuki Shokai", and has many uses. Further, for example, if the search sentence is “Sales in the last month”, the sales data is stored in the data accumulation table corresponding to this sales, and the sales data is stored in the actual results table. In this case, the data is retrieved from which table. The correspondence table selecting means 27 cannot determine the correspondence table. In this way, ambiguity occurs when data is held in duplicate.

In such a case, the inquiry returning device 29
In the rough classification item order in the term information that could not be narrowed down by the correspondence table selection means 27, in this case,
Check each table ID, representative word, and code, and if there is any difference, present it to the user as a candidate word and let it be selected,
The correspondence table has been decided.

FIG. 2 is a flow chart showing the procedure in the inquiry device 29, which will be described below. Ambiguity occurs (s1). Here, it is assumed that the term information data that cannot be narrowed down by the correspondence table selection means 27 is as shown in FIG. Table I, which is the most rough classification item
All the data of D are compared (s2). In this example, there is a difference in the data of the table ID, so "Product""Shipto"
"Customer" and "person in charge" are candidate words, and shown to the user as shown in FIG. 4 (a) (s3) and asked to select a number (s).
4). It is checked whether the correspondence table can be determined by the user's selection (s5). If it can be determined, the correspondence table is determined (s15), and this process is terminated (s16). For example,
If the user selects 3, the customer "Suzuki Shoten"
And the correspondence table is determined, and this processing ends.

If the user selects 4, "Suzuki"
Is determined to be "the person in charge" Suzuki "", but this "person in charge" Suzuki "" has two codes and still contains ambiguity. In such a case, in s7,
Check to see if there is a difference in the representative word that is the next item.
If there is, it is displayed to the user (s8), the user selects (s9), it is judged whether the correspondence table can be decided (s10), and if so, the correspondence table is decided (s1).
5) Then, this process ends (s16). In this case, as shown in FIG. 3, the representative word of the person in charge is “Suzuki” and there is no difference. Therefore, next, in s12, a different code is searched for. In the case of this example, since there are two 6A1 and 6A2, the code is displayed to the user as shown in FIG. 4B (s13), and the user is asked to select the code (s1.
4). In this way, the correspondence table is narrowed down until there is no ambiguity, the correspondence table is determined (s15), and the processing ends (s1).
6).

As described above, in this embodiment, when one word corresponds to a plurality of database items and is ambiguous, the contents of the term information that cannot be narrowed down by the correspondence table selecting means 27 are "table". The inquiring device has been described that sequentially inquires from more comprehensive classification items such as “ID”, “representative word”, and “code” and presents necessary data as options to return an inquiry to the user.

Example 2. FIG. 5 is a diagram showing a question and answer device according to the present invention. In addition to the device of Example 1 above,
It has a history reflection device 28 for reflecting the history and a history storage device 30 for storing the result of inquiry.

Next, the operation will be described. Example 1 above
After asking the question as shown in FIG.
The ambiguous word and the selected word selected by the user are paired and stored as a history. Each time a question is asked, the ambiguous word and the selected word are paired and stored in order. Then, when an ambiguous word occurs next, before the inquiry returning device 29 performs the inquiry, the history reflecting device 28 checks whether or not the same ambiguous word exists in the selection history, and determines whether or not the history is reflected. If the same ambiguous word exists in the history and the selected word can be adopted, the question data is not asked and the history data is used for narrowing down the ambiguous word. By storing and reflecting the history, it is possible to effectively use the result once selected by the user for narrowing the ambiguity, without repeating the same inquiry repeatedly.

FIG. 7 is a diagram showing the flow of processing when only the latest selection result from the same ambiguous word in the history is applied, and the diagram will be described. First, ambiguity occurs (s20). Search history data. The history is stored so that the latest history can be searched in order (s21). Next, it is checked whether there is search data (s22), and if there is no search data, the user is queried (s27). Then, the correspondence table is determined (s25), and the process ends (s2
6). If there is, next, it is checked whether the history data is the same as the ambiguous word (s23). If the ambiguous words are not the same, the process returns to s21 and the history data is searched. If it is the same as the ambiguous word, it is checked whether the selected word of the history is in the term information narrowed down by the correspondence table selecting means 27 (s2).
4). If not, the inquiry is returned to the user by the inquiry returning device 29 (s27). This processing is as described in the first embodiment. Thereby, the correspondence table can be determined, and this processing ends (s25, s26). When the selected word in the history can be adopted, the correspondence table can be determined without asking the user back (s25), and the process ends (s26).

For example, it is assumed that the ambiguity in FIG. 3 occurs again when the history is stored as shown in FIG. Comparing the history from the latest, the first history 1201 "Choco ｜ 1101"
Is an ambiguous word. Next, the new history 1202 is "Suzuki |
“Customer”, the ambiguous words match, and the selected word is also included in the data of the term information in FIG.
Is adopted, it becomes “sales of“ Suzuki ”who is a customer” without asking the user, and the correspondence table can be determined.

Thus, this embodiment is a case where only the latest history that can be adapted in the stored history is examined.

Example 3. This embodiment uses the same device as in the second embodiment, but applies the latest history when using the stored history, but if it cannot be adopted, the history with the same ambiguous word is searched for. FIG. 9 is a diagram showing the flow of this process, and the diagram will be described below.

Ambiguity occurs (s30). Search history data. The history is stored so that the latest history can be searched in order (s31). It is checked whether there is search data (s32), and if not, the user is queried (s).
37). Then, the correspondence table is determined (s35), and the process ends (s36). If there is search data, it is checked whether the history data is the same as the ambiguous word (s33). If it is not the same as the ambiguous word, the process returns to s31 again to search the history data.
If it is the same as the ambiguous word, it is checked whether or not the selected word of the history data is in the term information narrowed down by the correspondence table selecting means 27 (s34). If not, the process returns to s31 and the history data is searched. When the selected word in the history can be adopted, the correspondence table can be determined without asking the user (s3
5) The process ends (s36).

For example, the history is stored as shown in FIG. 10, and the result in which the ambiguity is narrowed down by the correspondence table selection means 27 by the search sentence "Suzuki's kakewa" is shown in FIG. The first history 1301 “Suzuki | Product” has the same ambiguous word, but the selected word “Product” is not included in the term information data of FIG. 11, so this history cannot be adopted. In the case of the second embodiment, the history search is stopped here and the user is queried again, but in this embodiment, the next new history is searched. The next history 1302 is “Suzuki | customer”, and the ambiguous words match and the selected word is also included. Therefore, this history is adopted, and “the customer's“ Suzuki ” It becomes the "multiplication rate" and the correspondence table can be determined.

As described above, in this embodiment, when the stored history is used, the latest adaptable history is automatically reflected from the matching history to eliminate the ambiguity.

Example 4. FIG. 12 is a diagram showing an inquiry device according to this embodiment. In addition to the device of the above-mentioned embodiment, the device of this embodiment has a history deletion device 10 capable of deleting a history before analyzing a search sentence.

Next, the operation will be described.

In the above-described embodiment, after asking a question, the history can be stored and the history that matches can be reflected when ambiguity occurs. However, if the topic changes or if the wrong selection is made, the history deletion device 1
The history stored by 0 can be deleted. This makes it easy for the user to manage the history and search for his or her intention.

FIG. 13 is a diagram showing the flow of processing in this embodiment. In this case, if history deletion is selected, the history up to this point is lost and you can search for a different topic.

Example 5. This embodiment relates to the deletion of the history as in the case of the fourth embodiment, but in this case, a part of the history is deleted. FIG. 14 is a diagram showing a process flow in this embodiment, which will be described with reference to the diagram.

When the user selects history deletion before performing the search (s51), the entire history is displayed on the screen as shown in FIG. 15 (s52). Therefore, the NO. By inputting, the history to be deleted or the history to be left can be selected (s53). Delete the history selected by the user (s5
4), the process ends. Since the history stored in this way is displayed on the screen and the user can select the history to be deleted depending on the storage status of the history, it is possible to change a part or delete the result selected by mistake. Becomes

Example 6. The above-described fourth and fifth embodiments may be performed separately, or a menu screen for selecting all history deletion and partial history deletion may be displayed and selected as shown in FIG.

Example 7. FIG. 17 is a diagram showing an inquiry device according to this embodiment. This apparatus has an ID-based history reflection apparatus 101, an ID-based history storage apparatus 102, and an ID-based history deletion apparatus 100.

Next, the operation will be described.

In this embodiment, first, the ID-based history storage device 10
2, the ID is stored as a history in addition to the ambiguous word and the selected word so far as shown in FIG. This ID may be any user name, user ID, time, season, or the like. Based on this history, the history reflecting apparatus for each ID 101 retrieves a history that applies to the ID during the actual search, and searches for a selection result that can be adopted in that history. For example, in FIG.
When the user with the ID R00 in 8 is searching, only the history with the ID R00 is the search target. In addition, the ID-based history deletion device 100
It recognizes and displays and deletes only the applicable history. This enables history management under various conditions.

Example 8. In this embodiment,
But using the stored history,
This is a case of designating a plurality of IDs as search targets. For example, if a user having an ID of R00 is R00, R02
It is also possible to retrieve the history having the two IDs and search for a selection result that can be adopted.

[0051]

As described above, according to the first and second aspects of the invention, even when the words in the search sentence are ambiguous, the choices are made from the contents of the term information and the user is queried back to ask the user. You can search according to your intentions.

According to the third invention, when the question is returned, the data selected by the user can be left as a history, and when the ambiguity occurs next time, the history can be reflected as long as it fits. Therefore, the same question can be returned. It is possible to effectively use the result once selected by the user for narrowing the ambiguity without repeating it again and again.

According to the fourth invention and the seventh invention, since the history can be deleted all or selectively by the inquiry history deleting device, it is possible to deal with the case where the topic is changed or the wrong selection is made.

According to the fifth aspect of the invention, even if the ambiguous words have the same ambiguity in the stored history, the latest one is checked and adopted if it matches. Therefore, only the latest history can be reflected in the history.

According to the sixth aspect of the invention, the stored history is examined from the newest one, and the one having the same ambiguous word is adopted, but if it cannot be adopted, the next new history is searched. Therefore, the history can be reflected in all the histories.

According to the eighth invention and the ninth invention, since the identifiers are added to the history storage device, the history reflection device, and the history deletion device, it becomes possible to manage the history for each identifier. This enables history management under various conditions.

[Brief description of drawings]

FIG. 1 is a block diagram including an inquiry returning device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a process flow of the above-described embodiment.

FIG. 3 is a diagram showing term information data.

FIG. 4 is a diagram showing the contents of an inquiry to the user.

FIG. 5 is a block diagram of an apparatus including a history storage device and a history reflection device according to an embodiment of the present invention.

FIG. 6 is a diagram showing the contents of a stored history.

FIG. 7 is a diagram showing a process flow of the above-described embodiment.

FIG. 8 is a diagram showing the contents of a stored history.

FIG. 9 is a diagram showing a process flow of the above-described embodiment.

FIG. 10 is a diagram showing the contents of a stored history.

FIG. 11 is a diagram showing term information data.

FIG. 12 is a block diagram including a history storage device, a history reflection device, and a history deletion device in one embodiment of the present invention.

FIG. 13 is a diagram showing a process flow of the above-described embodiment.

FIG. 14 is a diagram showing a process flow of the above-described embodiment.

FIG. 15 is a diagram showing information presented to a user when deleting a history.

FIG. 16 is a diagram showing a menu screen for selecting the type of history deletion.

FIG. 17 is a block diagram of an apparatus including an ID-based history storage device, an ID-based history reflection device, and an ID-based history deletion device according to an embodiment of the present invention.

FIG. 18 is a diagram showing the contents of a stored history.

FIG. 19 is a block diagram of a conventional correspondence table use database search device.

[Explanation of symbols]

 10 history deletion device 16 search sentence 17 input device 18 dialogue control device 19 morphological analysis device 20 syntactic analysis device 21 term information 22 grammar table 23 syntactic analysis result 24 correspondence table 25 collation unit 26 grouping means 27 correspondence table selection means 28 history reflection Device 29 Inquiry device 30 History storage device 50 Database search formula generation unit 51 Database search device 52 Database 100 ID-based history deletion device 101 ID-based history reflection device 102 ID-based history storage device

Claims (9)

[Claims] 1. An input means for inputting a natural language, a tabular database having a plurality of database items for storing data, and a correspondence table for generating a database search expression based on the natural language are stored. Correspondence table storage means, term information storage means for storing term information including information for associating words with the correspondence table, syntactic analysis unit for parsing input natural language and outputting syntactic analysis result, syntactic analysis result The collation unit that uses the term information to determine the correspondence table, the database retrieval formula generation unit that generates the database retrieval formula from the result of the collation unit using the above correspondence table, and the database retrieval formula is executed. In a database search system equipped with a database search device, when the matching unit uses the term information to select a correspondence table, a word is If ambiguous word database entry can not be identified correspond to the number of the correspondence table, a database search system which comprising the asking back device to select and present the contents of the term information for the ambiguous word. 2. The term information has information for associating words with a correspondence table divided into several stages, and when the ambiguous word occurs, necessary data is selected from the term information. 2. The database search system according to claim 1, wherein is presented step by step. 3. The database search system further uses a history storage device that stores a history of inquiry responses of the inquiry response device and an ambiguous word using the history stored in the history storage device when an ambiguous word occurs. 3. The database search system according to claim 2, further comprising a history reflecting device that specifies a database item corresponding to. 4. The database search system according to claim 3, further comprising a history deletion device that deletes the history stored in the history storage device. 5. The history reflection device specifies the database item corresponding to the ambiguous word using the latest history in the history of the ambiguous word stored in the history storage device. The database search system described in 3. 6. The history reflecting device specifies the database item corresponding to the ambiguous word using the latest history in the history applicable to the ambiguous word stored in the history storage device. The database search system according to claim 3, which is characterized in that. 7. The database search system according to claim 4, wherein the history deletion device deletes all or all of the histories stored in the history storage device. 8. The database search system according to claim 3, wherein the history storage device stores the history by adding an identifier thereto, and the history reflection device adapts the history by using the identifier. . 9. The database search system according to claim 8, further comprising a history deletion device that deletes the history stored in the history storage device using the identifier.