JP2001147940A - Retrieval system and retrieval method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a retrieval system capable of easily managing and reutilizing data. SOLUTION: Data stored in a data storage part 10 are grouped by a grouping part 30, overlap information, which evaluates the grouped data are included to be overlapped over plural pieces of data belonging to the same group is extracted y an information extracting part 40, and edited information is prepared by editing the extracted overlap information by an information editing part 50 and stored in the data storage part 10. A data retrieving part 20 can retrieve data stored in the data storage part 10. Only the edited information stored in the data storage part 10 can be retrieved from the outside.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates to a search system and a search method for searching stored information.

[0002]

2. Description of the Related Art In recent years, diversification of products, enhancement of functions, and expansion of users have been progressing. For this reason, the company's internal and external consultation (help desk) departments are instructed on how to use the product,
Inquiries about defects are increasing. Help desk operations require a large number of highly specialized personnel,
The increase in cost is regarded as a problem. Due to stressful work such as troubleshooting, it is difficult to have many specialized staffs. There is an urgent need for help desk operations to respond quickly and reduce costs.

As a solution to such a problem, there is a help desk support system for electronically recording past cases such as inquiries and answers, and reusing them. Staff who receive inquiries can respond quickly by referring to past cases. Some systems have a function to search past cases in natural language, so that even non-specialized staff can respond to some inquiries. However, the past cases that have been accumulated are records relating to situations that differ for each customer, and many of them cannot be reused as they are. Problems such as many similar cases are included, and old and useless cases occur, making it difficult to find the information to be sought. In order to improve efficiency, it is necessary to maintain a huge number of accumulated cases, but because of the need for specialty and complicated work, at present it is rarely performed.

[0004] On the other hand, in order to reduce costs, attempts have been made to collect and organize answers to frequently asked questions and to publish them on a WWW page or the like in the form of FAQ. As a result, it is possible to reduce the number of cases of inquiring directly to the counter, and to expect quicker response and cost reduction. But actually,
Creating and managing FAQs has the same problems as maintenance of past cases, and requires complicated work. further,
In the case of information that is disclosed directly to customers, past inquiry cases cannot be disclosed directly. It is necessary to select the cases to be included in the FAQ, replace the general situation, and create an easy-to-understand and concise explanation. This is a task that requires a high degree of specialty and expressive power, and it is difficult to achieve cost reductions by opening FAQs. FAQ
No effective method of supporting creation has been realized.

[0005] Of course, the above-mentioned problem occurs not only in a system that handles information related to the help desk business but also in a system that handles other information.

[0006]

As described above,
In a conventional search system, management for reusing a huge amount of accumulated data is not easy, and it is difficult to achieve cost reduction by reusing.

[0007] The present invention has been made in view of the above circumstances, and has as its object to provide a search system and a search method capable of facilitating data management and reuse.

[0008]

A search system according to the present invention comprises: a grouping means for grouping data (for example, data such as text, voice, image, etc.) stored in a predetermined storage device; Extraction means for extracting duplicate information included in a plurality of data belonging to the same group with respect to data grouped by the means, and editing means for editing the duplicate information extracted by the extraction means. It is characterized by having.

Preferably, the editing means may include means for presenting a data group from which duplicate information to be edited is extracted.

Preferably, the grouping means detects, from the data having a structure, core data (for example, a title) serving as a core of the group based on the structure,
Means for extracting features (for example, word sets) from the core data, means for grouping by including data having core data including the features in the same group,
Means for assigning a predetermined range including the feature of the nuclear data as a group label.

[0011] Preferably, each data stored in the predetermined storage device is provided with a unique identifier, and the grouping means includes the other data described as the content in a certain data. The certain data and the other data may be put into the same group based on an identifier given to the data or information capable of specifying the data.

[0012] Preferably, the extracting means converts the data in the grouped data group into a predefined form, an item of the form overlapping at least two data, and the contents thereof. May be extracted as the duplicate information.

[0013] Preferably, the information processing apparatus may further include search means for searching only information extracted by the extraction means and edited by the editing means as a search target.

Preferably, the editing means includes an instruction means for instructing whether or not the information extracted by the extracting means can be made public, and the search means makes only the information made public by the instruction means a search target. You may make it search.

[0015] Preferably, the information processing apparatus may further include supplementary information extracting means for extracting information other than the overlapping information and satisfying a predetermined criterion as supplementary information.

Preferably, each data in the grouped data group is converted into a predefined form, and in at least one data, the items of the form that do not overlap with other data and the contents thereof are stored in the form. The information processing apparatus further includes supplementary information extracting means for extracting as complementary information of the group, wherein the editing means can also edit the complementary information, and the search means can search only the complementary information. You may.

Further, the search method according to the present invention provides a method for storing data (for example, text, voice,
Image data, etc.), and for the grouped data, duplicate information that is evaluated to be included in multiple data belonging to the same group is extracted and edited. It is characterized by the following.

The present invention relating to the apparatus is also realized as an invention relating to a method, and the present invention relating to a method is also realized as an invention relating to an apparatus.

The present invention relating to an apparatus or a method is provided for causing a computer to execute a procedure corresponding to the present invention (or for causing a computer to function as means corresponding to the present invention, or for causing a computer to correspond to the present invention). The present invention is also realized as a computer-readable recording medium in which a program for realizing the function of performing the above is recorded.

According to the present invention, accumulated data is grouped, duplicate information is extracted from the grouped data, and editing is performed based on the extracted duplicate information. It is easy to sort out useful information and edit it into a reusable form. Also, only information after editing can be set as a search target.

According to the present invention, complicated operations related to information reuse, such as management of stored data, are reduced. Also, for example, even a staff member who does not have a high level of specialized knowledge can easily create a FAQ or the like. Also, for example,
Not only can the help desk department reduce costs, but it can also enhance the service to customers by releasing high-quality information quickly.

[0022]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows a configuration example of a search system according to an embodiment of the present invention.

This search system assumes, for example, a data manager as a user (a person other than a manager (for example, a general user such as a customer or a person other than a company administrator). The components assumed as (user) will be described later.)

As shown in FIG. 1, the search system includes a data storage unit 10, a data search unit 20, a grouping unit 30, an information extraction unit 40, and an information editing unit 50.

The part for performing each processing can be realized by software, but various forms of realization are conceivable. For example, (1) data search unit 20 / grouping unit 3
0 / Information extraction unit 40 / Information editing unit 50 are all mounted on one computer, and the data storage unit 10 is a local storage device of this computer. (2) Data search unit 20 / Grouping unit 30 / A method in which the information extracting unit 40 / information editing unit 50 are all mounted on one computer and the data storage unit 10 is managed by another computer on the network. (3) In the above (1) or (2), Various forms are conceivable, such as a method in which only the search unit 20 is mounted on another computer, a method in which the above-described units are combined and mounted in one or a plurality of computers, and whether or not a network is used.

In FIG. 1, a data input section for inputting data to be stored in the data storage section 10 is omitted. This data input unit may be mounted on a computer in which any of the above-mentioned parts is mounted, or may be mounted independently on the computer.

As a user interface,
It is preferable to use a GUI (graphical user interface).

The data storage section 10 is for storing data such as text, voice, images, etc., provided via the data input section.

The data search unit 20 is for receiving and analyzing a search command based on a natural language sentence or the like, searching for data matching the content of the command from the data stored in the data storage unit 10, and presenting the data. .

The grouping unit 30 is for creating a group (group) including a plurality of data from the data stored in the data storage unit 10.

Note that there are a method of performing grouping every time data is input and a method of performing grouping at appropriate timing. It is also possible to rearrange groups for some or all of the data.

The information extracting section 40 refers to the data belonging to the group created by the grouping section 30 and extracts information (duplicate information) redundantly included in a plurality of data. The duplicate information is not necessarily limited to exactly the same sentence, but has a meaning content such as information having a content that can be evaluated as duplicate. In addition, the duplication information is not necessarily limited to the case where it is included in all data belonging to one group. For example, the duplication information is included in a certain number (or may be two or more) of data belonging to one group. Case, the data may be included in data that occupies a certain percentage of the data belonging to one group. Since the duplicate information is information that appears commonly to all or a (many) part of the data belonging to the group, for example, the content of the information is for the group (or for the theme or characteristic of the group). , Information that is commonly valid, important information, or useful information, or information that should or should be included in common. Therefore, for example, duplicate information is used as a material, and this is edited (processed).
By doing so, it is possible to create useful information such as representative information of the group, organized information, or information that can be browsed by general users.

The information editing section 50 is for editing based on the duplicate information extracted from the data belonging to a certain group by the information extracting section 40. For example, the extracted duplicate information is presented, an instruction for information processing such as addition, change, or deletion by the user is received, and the information is processed as instructed. Information extracted by the information extracting unit 40 and edited by the information editing unit 50 (hereinafter, referred to as edited information) is stored in the data storage unit 10.

In the present embodiment, it is assumed that data having duplicate information as a material is not directly edited, but newly edited information is created based on the data. It is assumed that the data used as the material is stored even after the edited information is created.

The data search unit 20 can search the edited information stored in the data storage unit 10.

In the following, the flow of the processing of this embodiment will be described.

FIG. 2 shows an example of the processing procedure of this embodiment. This procedure represents a flow of a process of grouping data as materials, extracting duplicate information from data belonging to a certain group, and creating edited information.

First, the information stored in the data storage unit 10 is read (step S11), and grouping is performed (step S12).

Hereinafter, examples of some grouping methods will be described.

Here, as an example of data, “text data” (101) including information of each item of “title”, “question text”, and “answer text” as shown in FIG. It is assumed that a large amount of data is accumulated.

In the first grouping method, a data group included in one generated cluster is defined as one group by using a cluster generation algorithm based on “title” as exemplified below.

(1) For a title tj of certain data, all word sets Wi satisfying the following conditions are generated. 1 ≦ N (Wi) ≦ P1 Wi is included in Wtj. N (Wi) / N (Wtj)> P2 Wi: an arbitrary word set (a candidate for a reference for cluster generation) Wtj: a word set of a title tj of certain data N: returns the number of words constituting the word set Functions P1 and P2: Constant This is performed for all or some data to be grouped.

(2) A cluster is generated from data having a title including each word set created in the procedure (1).

For example, Wtj = {2000, question, test} is obtained from the title “About the test for the year 2000 question”. If P1 = 2, W
i is {2000, problem}, {2000, test}, {problem, test}, {2000}, {problem},
{Test} is obtained. If P2 = 0.25, N (Wtj) = 3, N (Wi) = 1 or 2, and in this example, any word set is N (Wi) / N (Wt
j)> P2 is satisfied. Such a word set is generated for other data. Then, for all of the generated word sets or a part of the word sets, clusters are generated from data having titles that include the word sets. For example, if there is a large number of data having a title including the word set of {2000, question}, {2000, question} is adopted. Generate a problem cluster. For example, the data with the title "About the year 2000 test" is $ 2000,
Problem I belong to the cluster.

In the above algorithm, one data can be included in a plurality of clusters. However, one data may be included in a plurality of clusters, or one data may be included in one cluster. It may be limited to be included only in the cluster. In the latter case, it is determined to which cluster data that can be included in a plurality of clusters belongs according to another predetermined criterion.

Also, a hierarchy may be created between clusters. For example, when there are clusters Ci and Cj generated from Wi and Wj, if Wi is included in Wj, Ci includes Cj. For example, the {2000, Question, Test} cluster is a child of the {2000, Question} cluster.

Before the above algorithm is applied, by converting synonyms to the same words using a synonym dictionary or the like, it is possible to further increase the number of data included in one cluster. For example, in the above example,
If “Y2K” is described in the synonym dictionary of “2000”, data having a title such as “Y2K problem patch” is also classified into the same cluster.

Next, as a second grouping method, words included in the data can be used to classify the data into a plurality of groups by a hierarchical clustering method that is common in the field of multivariate analysis.

Next, as a third grouping method, a category structure to be classified can be determined in advance, and data can be allocated to each category. For example, a dictionary of important words is prepared, and a weight such as a probability that each word is included in a category is obtained. Then, each time a word described in the dictionary appears in each data, the weight for each category is added, and each data is classified into the category with the highest weight.

Next, as a fourth grouping method, if there is a description referring to a previous inquiry in data such as a title, this can be used. For example, if a unique ID is added to the inquiry record,
A description such as "continuation of Y00000000082" is found in the data, and this data and referenced past data (data having ID "Y0000000282") are grouped. The same applies to the case where information capable of specifying a unique ID of other data is described in the data.

Various other methods are conceivable.
In addition, these grouping methods can be used by appropriately combining a plurality of methods. For example, first, clusters based on titles are determined using the above-described first grouping method, and then data including a description referring to data included in each cluster is searched for in accordance with the above-described fourth grouping method. It is also possible to make adjustments to add to the same cluster.

Next, duplicate information is extracted from the grouped data (step S13).

Various methods can be considered for extracting duplicate information. For example, when grouping is performed by the above-mentioned “title” cluster generation algorithm, the same title is assigned to the data group belonging to the group. The word set is included. Therefore, from the titles of the data group, when the elements of the word set of interest are adjacent, or when the elements of the word set of interest are within a certain range, the adjacent part or the part of the certain range is included. The part can be extracted as duplicate information. For example, information such as "about the year 2000 problem" is extracted from the group generated from the word set {2000, problem}.

Also, for example, of the grouped data, it is checked whether or not each of the data of the item of “question text” and the data of the item of “answer text” includes an overlapping word set. Is an information constituent unit (for example,
One sentence, one paragraph, or the like), or an information constituent unit in which an overlapping word set exists adjacently, or an information constituent unit in which an overlapping word set exists within a certain range can be extracted as duplicate information.

Also, based on the existence of individual words instead of word sets, it is possible to extract duplicate information such as question sentences and answer sentences by the following method. (1) Find a word that is duplicated in a plurality of data (or the same or similar item of a plurality of data). For example, when there is an answer 1, an answer 2, and an answer 3, a word that is duplicated between the answer 1, the answer 2, and the answer 3 (that is, appears in a plurality of answers) is found. When the number of answers is large, the answer may not be limited to only appearing in a plurality of answers, but may be limited to appearing at a certain ratio or more.

(2) Extract a sentence having many duplicate words For example, if all the words constituting the sentence meet the above condition (1), the sentence is extracted unconditionally. If there is no such sentence, one sentence that has the most is extracted.

(3) The duplicate of the extracted sentence is deleted. For example, when a plurality of sentences having exactly the same words (that satisfy the condition (1)) constituting the extracted sentence are extracted, only the first sentence is extracted.

As another method for extracting duplicate information, there is a method using the following form.

In this case, for example, the data of the inquiry and the answer as shown in FIG. 3 are analyzed, and the form (111) representing the meaning structure of the inquiry as shown in FIG.
Conversion is performed in the following manner.

A characteristic expression of each item (112) of the form is described as a rule based on a part of speech or a sequence of words. After each unit sentence in the data is decomposed into parts of speech by morphological analysis or the like, the rule is applied.
904). A unit statement including a description that matches the rule is extracted as information (113) of the item. For example, if you write a rule that includes words such as "manual", "page", and "reference",
A sentence such as “Please refer to page XX of manual XX” in the data can be extracted as reference information in the form of FIG.

In this way, all data in the group is converted into a form by the above-described method.

A plurality of forms may be prepared according to the contents of the inquiry.

If there is information extracted from a plurality of data for each item of the form, the degree of duplication of the two pieces of information is calculated by the following formula, for example. Degree of duplication = (number of words included in two pieces of information) /
(Total number of words included in two pieces of information) Then, a unit sentence whose degree of duplication is equal to or more than a threshold is extracted as duplication information. In this case, a plurality of pieces of duplicate information are extracted as they are.

Note that the extracted duplicate information may be ranked (for presentation or use at the time of editing, searching, etc.). For example, the degree of duplication may be set as a rank for each item.

Further, for example, one data containing many unit sentences having a high degree of duplication is selected, and among the unit statements contained in the data, only those having a degree of duplication higher than a threshold are extracted in the order of appearance in the data. You may make it.

FIG. 5 shows a display example of the overlapping information along the form.

In addition to the above, various methods are conceivable.
In addition, these overlapping information extraction methods can be used by appropriately combining a plurality of methods. For example, after converting the data into the form as described above, for the information of each item, it is also possible to extract the range including the overlapping word set as the overlapping information by the method using the word co-occurrence relation as described above. it can. For example, in the case of the form shown in FIG. 4, it is also possible to extract redundant information on items such as a product name, a request, a situation, and a countermeasure, and provide an overview of data included in the group.

Next, the extracted duplicate information is edited (step S14). The edited information is stored in the data storage unit 10 (Step S15).

In the present embodiment, in editing the extracted duplicate information, edited information is created for a certain group based on the duplicate information extracted from the data belonging to the group. For example, extracted from data including “a sentence that is not supposed to be viewed by general users” or “a sentence that directly describes individual and specific Q & A results” belonging to a group related to the “2000 problem”. Using the duplicate information as a material, it is possible to easily create edited information having contents assumed to be viewed by a general user.

The editing of duplicate information (creation of edited information) will be described below.

For example, a screen (131) as shown in FIG.
Then, (duplicate) of the extracted duplication information is displayed, and the content is corrected according to the user's editing instruction. In this example, at the upper part (132) of the screen, the duplicate information extracted from the items of "title", "question sentence" and "answer sentence" for a certain data is presented. The user can directly edit the information in the frame. Data belonging to the same group can be displayed in the lower part (133) of the screen, and the information in the upper part (132) can be edited while referring to the data as appropriate.

Further, as shown in the upper part (132) of FIG. 6, the user can select "public" or "non-public" (134, 135). If the information is to be disclosed to a person other than the person, a designation may be made that some information is not disclosed.
Thereby, only the administrator can refer to the edited information. Further, with respect to in-company information and the like, it may be possible to specify a person or a department that can be viewed in fine units.

When grouping by title is performed as described above, the contents of the titles of the data constituting the group are uniform, so that the characteristics of the group can be easily grasped.
There is an advantage that the user can easily edit.

Further, for example, in addition to the screen as shown in FIG. 6, a group of buttons (142 to 144) is prepared in the middle (423) of the edit screen (141) as shown in FIG. Editing can also be enabled.

For example, when the first data among the data belonging to the group displayed in the lower part (424) of FIG. 7 is selected, the selected data is displayed. Here, when the "delete" button (142) is clicked, the function of deleting the selected first data from the group can be provided. .

Similarly, when one or a plurality of data is selected and the “split” button (143) is pressed, the selected data is deleted from the group, and a new group including the deleted data is deleted. It is also possible to have a function of creating.

By clicking an "add" button (144), a screen for accepting a data search is started, and data selected from the search results by inputting a natural language sentence or the like is added to the group being edited. It is also possible to have a function such as addition.

After the group is edited, the duplicate information may be extracted again.

For example, when duplicate information is extracted for each item of the form as shown in FIG. 4, for example, on the screen (121) as shown in FIG. Is presented and the user edits it.

As described above, it is possible to present the duplication information of each item for data including many unit sentences having a high degree of duplication, or to present the item having the highest degree of duplication for each item. You may make it.

On the screen shown in FIG. 5, information not presented is displayed by clicking the button (123) next to the item frame, for example, as shown at 152 in FIG. As a result, it is possible to select other duplicate information (for example, the second or subsequent item), and it is possible to edit the information of each item while referring to other data.

The created edited information is stored in the data storage unit 10 in association with the grouped data by pressing a button such as “Register”. At this time, the created edited information is registered with an identifier different from the data (which is the material of editing) added.

Incidentally, the duplication information extracted by the information extraction unit 40 may be stored in the data storage unit 10 instead of being edited immediately after the extraction, and may be edited later. FIG. 9 shows an example of a processing procedure in this case. That is, a natural language sentence or the like is input to search for data stored in the data storage unit 10, and the search results are displayed (steps S21 to S23). When the user selects data from the search results (step S24), an edit screen of a group to which the selected data belongs (for example, FIGS. 5, 6, 7,
8 and the like are displayed to facilitate the user's editing work.
Then, the user edits based on the duplicate information (step S25), and the edited information is stored in the data storage unit 10 with an identifier attached (step S26). in this case,
If the information edited by the user is added and registered with information that can be identified as “edited”, the information after editing can be identified. For example, as described above, identification can be performed by adding identification information different from data to edited information.

Further, an editing operation can be performed at the same time as registering data (in the data storage unit 10).
FIG. 10 shows an example of the processing procedure in this case. That is,
When handling the data as shown in FIG. 3, after the user who receives the question inputs the data, the data stored in the data storage unit 10 is searched by using the input data, and the search result is displayed. (Steps S31 to S3
3). When the user selects data from the search result (step S34), an editing screen is displayed for the group selected by the user as illustrated in FIG. Subsequently, the user can perform editing (step S35). At this time, when the user clicks the "Add" button (144), the new data being input is added to the group. If there is edited information to be stored, it is stored in the data storage unit 10 (step S36). In the above description, the "answer" of the data being input can be created and corrected while referring to the data of the group to which the data being input by the user belongs. That is, before the user who accepts the question inputs “answer” or after inputting the draft of “answer”, the data stored in the data storage unit 10 is searched by using the input data, and the search is performed. The result is displayed (steps S31 to S33).
When the user selects data from the search results (step S
34), an editing screen is displayed for the group selected by the user as illustrated in FIG. Subsequently, the user can perform editing (step S35). that time,
When the user clicks the "Add" button (144),
New data being entered is added to the group. The user can create an “answer” of the input data or modify the original “answer” while referring to the data of the same group. If there is edited information to be stored, the edited information is stored in the data storage unit 10 (step S36).

Here, the data search result display screen may display a list of data titles in descending order of score, as in the conventional search result screen. , And the title of the data can be displayed together with the title of the group. For example, as shown in FIG. 11, when the user performs an operation for search on the upper part (202) of the display screen (201), the search is performed according to the search condition. When the search is completed, the data of the search results are arranged in descending order of the score (regardless of the group), and then grouped for each group (here, displayed as FAQ). 2
03). At that time, for example, after being arranged in descending order of the score, for example, the first place and the tenth are arranged in the descending order of the score.
If the data belonging to the same rank belongs to the same group, the data of the first place may be displayed at the top (204), and the data of the tenth place may be displayed below the data (205). Alternatively, only the first data may be displayed, and the tenth data may not be displayed on the search result screen. By grouping and displaying in this way, the user can collectively confirm similar data, and the trouble of searching for desired information is reduced.

A screen for editing the entire structure of the group may be prepared. For example, when the “all display” button (145) in FIG. 7 is clicked, an editing screen for the hierarchical relationship of groups as shown in FIG. 12 is started. The upper part (212) of the display screen (211) of FIG.
The title of the group being edited is displayed, and editing can be performed. The lower part of the display screen (211) (2
13) shows the hierarchical relationship of the groups.

In the example shown in FIG. 12, characters surrounded by a frame are the title of the group and the like. Further, the number of data included in the group may be displayed as a bar graph next to each title. Thus, the distribution of the stored data can be confirmed.

In the example of FIG. 12, for example, the following group editing work can be performed.

When a group displayed at the bottom (213) of the display screen is selected, the display such as the title at the top (212) of the display screen is updated for the selected group.

When an "open" button (218) or the like is selected with a group selected, if there is a lower group in the selected group, a group hierarchy including the lower group is displayed at the lower part (213) of the display screen. . If there is no group below the selected group, the screen is switched to the screen shown in FIG. 7 and data included in the group is presented.

When a "delete" button (214) is clicked while a group is selected, the group is deleted.

When one or more groups are selected and the "split" button (215) is clicked, a new group containing the selected groups is created along with the current parent group. For example, in FIG.
When “aaa” and “bbbb” are selected and “split” is clicked, a new group of “XXXX” groups is created, and the “aaaa” and “bbbb” groups are moved below them.

When one or more groups are selected and the "Insert" button (216) is clicked, a new group containing that group is created as a child of the current parent group. For example, in FIG.
When "aa" and "bbbb" are selected and "insert" is clicked, a new group is created as a child of the "XXXX" group, and the "aaaa" and "bbbb" groups are moved below them.

When one or a plurality of groups are selected and the "join" button (217) is clicked, the members of the selected groups are integrated into one group. For example, in FIG. 12, "aaa" and "b"
When "bbb" is selected and the "join" button (217) is clicked, data of the "bbbb" group is added to the "aaaa" group to form one group. on the other hand,
The “bbbb” group itself will be deleted.

Further, terms indicating the characteristics of the groups may be written beside the titles of the groups in FIG.
For example, when grouping is performed using titles as described above, a word set such as {2000, question} for determining a group may be written as it is. Alternatively, a term (label) representing a feature of the group may be obtained by a known probability model or the like, and may be displayed.

Up to this point, the description has been given of a portion in which a data manager or the like is assumed to be a user. For example, there has been a case where data itself can be viewed only by an administrator, and the administrator creates information that can be viewed by anyone other than the administrator based on the data.

Hereinafter, a case will be described in which a user other than the administrator searches the created edited information.

FIG. 13 shows an example of the system configuration in this case. That is, FIG. 13 is obtained by adding the information search unit 60 to the system of FIG.

The information retrieval unit 60 receives and analyzes a retrieval command based on a natural language sentence or the like, and stores it (stored in the data storage unit 10, extracted by the information editing unit 50, edited by the information editing unit 50, and This is for searching for and presenting data that matches the content of the instruction from the edited information (accumulated in 10).

The information retrieval unit 60 can be realized by software. For example, a computer used by a user other than the administrator (for example, a computer used by a general user on the Internet, an intra-company LA to which the system is connected).
N, etc.). When the system in FIG. 1 is a stand-alone device, a mode in which the information search unit 60 is further mounted on the stand-alone device is also possible.

FIG. 14 shows an example of a processing procedure when a user other than the administrator searches the stored edited information.

A user other than the administrator inputs a natural language search sentence, for example. The information search unit 60 reads the search sentence (step S41), and matches the search request among the edited information stored in the data storage unit 10 (the original data is not set as a search target here). The searched information is searched for (step S42) and displayed to the user (step S43).

FIG. 15 shows an example of the search screen. For example, the upper frame (302) of the display screen (301) in FIG.
And a search button (303) is selected. Then, the search sentence is read and matched with the edited information stored in the data storage unit 10, and the search results are displayed at the bottom of the screen (304) in the order in which they apply.
As each search result, for example, a title, a question sentence, and a part of the answer sentence are displayed.

The user selects desired information from the search results and browses it.

As a result, users other than the administrator can
For example, only edited information that is supposed to be viewed by a user other than the administrator can be made public, without viewing the unedited data, such as the response history with another customer, as it is.

Further, as described above, if "public" or "non-public" information is added and registered, only the information which has been made "public" among the edited information is presented. can do.

The case where edited information is created on the basis of the duplicate information extracted from the data group belonging to the same group has been described above. In the following, the case where a part other than the duplicate information included in the data is used will be described. Will be described.

FIG. 16 shows an example of the system configuration in this case. That is, FIG. 16 is obtained by adding an information search unit 60 to the system of FIG.

The data storage unit 10, the data search unit 20, the grouping unit 30, and the information extraction unit 40 have basically the same functions as those of the respective components in FIG.

In the following, description will be made focusing on portions different from those described above.

The supplementary information extraction unit 41 added in FIG.
Is for extracting, as supplementary information, all pieces of information other than duplicate information or information other than duplicate information and satisfying a predetermined criterion. For example, among the data belonging to the group created by the grouping unit 30, information that does not overlap in at least some two data is extracted as complementary information.

The information editing unit 50 edits the duplicate information extracted by the information extracting unit 40 and the complementary information extracted by the complementary information extracting unit 41.

The overlapping information extracted by the information extracting unit 40, the complementary information extracted by the complementary information extracting unit 41,
Edited information edited on the basis of the duplicate information by the information editing unit 50 (hereinafter, referred to as edited duplicate information to distinguish it from the edited complementary information), and complemented information edited based on the complementary information Information (hereinafter referred to as edited supplementary information)
Are stored in the data storage unit 10.

FIG. 17 shows an example of the processing procedure in this case. Steps S51 to S53 correspond to steps S11 to S11 in FIG.
It is the same as S13.

Step S56 is the same as step S15 in FIG.

The extraction of the complementary information in step S54 can be carried out, for example, by a method using the following form.

The data of the inquiry and the answer as shown in FIG. 3 are analyzed and converted into a form representing the meaning structure corresponding to the inquiry as shown in FIG. 4, for example. This conversion is the same as that already described.

Here, the degree of duplication of a certain item of the form between any two pieces of data in the group created by the grouping unit 30 is focused on the information converted to the item for each data. For example, it is calculated by the following equation in the same manner as described above. Degree of duplication = (number of words commonly included in two pieces of information) /
(Number of words included in two pieces of information) When the degree of duplication of the information of a certain item of certain data with the same item of all other data in the group does not exceed a predetermined threshold, a complementary information extracting unit 41 extracts the information A as complementary information on the item of the group.

Next, in step S55, the information extraction unit 4
Duplicate information and complementary information extraction unit 4 extracted by 0
The supplementary information extracted by 1 is edited as follows.

For example, as shown in FIG.
Information and complementary information extraction unit 41 extracted by
Display the supplementary information extracted in the form.

In FIG. 18, the duplicate information extracted by the information extracting unit 40 is displayed on the left side (402) of the display screen (401), and the complementary information extracted by the complementary information extracting unit 41 is displayed on the right side (403). , The contents of each item can be edited. When there is a plurality of pieces of information to be edited for each item, a button (40
By pressing (40), information to be edited can be exchanged.

FIG. 19 shows another example of the display screen. The left side (412) and the right side (413) of the display screen (411)
It has the same form configuration, duplicate information is in the left frame,
The supplementary information is displayed in the right frame. This makes it possible to smoothly move, for example, from complementary information to duplicate information, or from duplicate information to complementary information, in addition to editing each piece of information.

The display screen example of FIG. 20 is an example in which the display at the upper part (422) of the display screen of FIG. 7 is replaced with the display of FIG. In this case, for example, while editing the “operation” item of the supplementary information at the upper part (425) of the display screen of FIG. 20, the original data from which the data being edited is extracted is displayed at the lower part (424) of the screen. You.

The display screen example of FIG. 21 is an example corresponding to a display area of supplementary information added below the display area of overlapping information at the upper part (422) of the display screen of FIG.
The description of the buttons and the like is omitted in FIG. 21). In this case, for example, the supplementary information is selected and the up arrow button (4
When the user presses 34), the user can move to the overlapping information, or when the user selects the overlapping information and presses the down arrow button (435), the operation moves to the complementary information.

FIG. 22 shows an example in which a supplementary information display area is added to the screen in the format of FIG. In the example of FIG. 22, the distinction between the two is made clear by displaying the overlapping information in a bold frame and displaying the complementary information in a thin frame. If both duplicate information and complementary information exist for the same item,
Display two frames.

Next, a case will be described in which the information search unit 60 is connected to the search system of FIG. 16 and searches for edited duplicate information and edited complementary information.

FIG. 23 shows an example of the processing procedure in this case.

For example, the upper frame (30) of the display screen (301) of the search interface similar to that shown in FIG.
Input a search sentence in 2) and select a search button (303). Then, the search sentence is read and the data storage unit 10
Performs matching with the edited duplicate information stored in
The search results are displayed at the bottom of the screen (304) in the applicable order (steps S61 and S62). As each search result, for example, a title, a question sentence, and a part of the answer sentence are displayed.

Here, when the user selects one from the displayed search results, the search result is displayed, for example, as shown in FIG.
The selected and edited duplicate information is displayed in the middle part of the screen (305), and the edited complementary information extracted from the same group is displayed in the lower part of the screen (306). Here, a search sentence for searching for the edited complementary information is input in a frame (302) at the top of the screen, and a “squeeze” button (30)
When 7) is pressed, a search is performed only from the edited complementary information in the displayed group, and the result is displayed (steps S63 and S64). As another search interface example, as shown in FIG. At the top of the screen, there may be provided both a search sentence input unit for duplicate information (question frame in the figure) and a search sentence input unit for complementary information (fine frame in the figure). In this case, if the search is executed by inputting both of the search sentence input units, it is possible to search for information in which both conditions of the duplicate information and the complementary information match as shown in the lower part of FIG.

The above functions can be realized as software.

Further, the present embodiment is a computer in which a program for causing a computer to execute predetermined means (or for causing a computer to function as predetermined means, or for causing a computer to realize predetermined functions) is recorded. It can also be implemented as a readable recording medium.

The present invention is not limited to the above-described embodiments, but can be implemented with various modifications within the technical scope thereof.

[0134]

According to the present invention, management and reuse of data can be facilitated by supporting selection and editing of information for data reuse. Further, only the information on which the editing work has been performed can be a search target from the outside.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration example of a search system according to an embodiment of the present invention.

FIG. 2 is an exemplary flowchart illustrating an example of a processing procedure in the embodiment.

FIG. 3 is a view showing an example of data in the embodiment.

FIG. 4 is an exemplary view showing an example of a form in the embodiment.

FIG. 5 shows an example of display of overlapping information along a form in the embodiment.

FIG. 6 is an exemplary view showing an example of a GUI screen according to the embodiment.

FIG. 7 shows an example of a GUI screen according to the embodiment.

FIG. 8 is a view showing an example of a GUI screen according to the embodiment.

FIG. 9 is an exemplary flowchart illustrating another example of the processing procedure in the embodiment.

FIG. 10 is an exemplary flowchart illustrating still another example of the processing procedure in the embodiment.

FIG. 11 is a view showing an example of a GUI screen according to the embodiment.

FIG. 12 is a view showing an example of a GUI screen according to the embodiment.

FIG. 13 is a diagram showing another configuration example of the search system according to the embodiment of the present invention.

FIG. 14 is an exemplary flowchart illustrating still another example of the processing procedure in the embodiment.

FIG. 15 is a view showing an example of a GUI screen according to the embodiment.

FIG. 16 is a diagram showing still another configuration example of the search system according to the embodiment of the present invention;

FIG. 17 is an exemplary flowchart illustrating still another example of the processing procedure in the embodiment.

FIG. 18 is a view showing an example of a GUI screen according to the embodiment.

FIG. 19 is a view showing an example of a GUI screen according to the embodiment.

FIG. 20 is an exemplary view showing a GUI screen example in the embodiment.

FIG. 21 is an exemplary view showing an example of a GUI screen according to the embodiment.

FIG. 22 is an exemplary view illustrating an example of a GUI screen according to the embodiment.

FIG. 23 is a flowchart showing still another example of the processing procedure in the embodiment;

FIG. 24 is a view showing an example of a GUI screen according to the embodiment.

FIG. 25 is a view showing an example of a GUI screen according to the embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 ... Data storage part 20 ... Data search part 30 ... Grouping part 40 ... Information extraction part 41 ... Complementary information extraction part 50 ... Information editing part 60 ... Information search part

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Hideki Tsutsui 1st Toshiba R & D Center, Komukai-ku, Kawasaki-shi, Kanagawa Pref. No. 1 town Toshiba R & D Center F-term (reference) 5B075 ND03 ND06 ND14 ND16 ND20 NR03 NR12 NS10 PP24 PQ02 PQ34 PQ46 PQ74 PR06 QM08

Claims (10)

[Claims] 1. A grouping means for grouping data stored in a predetermined storage device, and data grouped by the grouping means included in a plurality of data belonging to the same group in a redundant manner. A retrieval system comprising: extraction means for extracting duplicate information to be extracted; and editing means for editing the duplicate information extracted by the extraction means. 2. The retrieval system according to claim 1, wherein said editing means includes means for presenting an original data group from which duplicate information to be edited has been extracted. 3. The grouping means includes: means for detecting, from data having a structure, core data serving as a nucleus of a group based on the structure; means for extracting a feature from the core data; Means for performing grouping by including data having nuclear data including the same in the same group, and means for assigning a predetermined range including the feature of the nuclear data as a label of the group. The search system according to claim 1, which performs the search. 4. Each of the data stored in the predetermined storage device is provided with a unique identifier, and the grouping means is configured to output the other data described as the content in a certain data. 2. The method according to claim 1, wherein the certain data and the other data are in the same group based on an identifier assigned to the data or information capable of specifying the data.
Search system described in. 5. The extracting means includes means for converting each data in the grouped data group into a predefined form, and an item of the form overlapping at least two data and its contents. 2. The retrieval system according to claim 1, wherein the information is extracted as the duplicate information. 6. The apparatus according to claim 1, further comprising a search unit that searches for only information extracted by the extraction unit and edited by the editing unit as a search target. Search system described in. 7. The editing means includes an instructing means for instructing whether or not the information extracted by the extracting means can be disclosed, and the retrieving means retrieves only the information permitted to be disclosed by the instructing means as a retrieval target. The search system according to claim 6, wherein the search is performed. 8. The apparatus according to claim 1, further comprising supplementary information extracting means for extracting information other than the overlapping information and satisfying a predetermined criterion as supplementary information. Search system described. 9. Each of the data in the grouped data group is converted into a predefined form, and at least one form item that does not overlap with other data and its contents are stored in the group. Further comprising complementary information extracting means for extracting the complementary information as the complementary information, wherein the editing means can also edit the complementary information, and the search means can also search only the complementary information. The search system according to claim 6 or 7, wherein 10. A method for grouping data stored in a predetermined storage device, and extracting, from the grouped data, duplication information that is evaluated to be included in a plurality of data belonging to the same group. A search method characterized by editing extracted duplicate information.