JPH0744573A - Electronic filling device - Google Patents

Abstract

PURPOSE:To attain the automatic registration of document and image information without being affected by the knowledge or understanding of a registration person by automatically adding optimal management information, and operating automatic classification. CONSTITUTION:This device is equipped with an image inputting means 1 which inputs a document and image A, file device 2 which preserves image data F and management information G, and registering means 3. Also, the device is equipped with a character recognizing means 4 which converts character information in image data B into code data C, knowledge base 6, management data extracting means 5 which extracts management data D from the code data C by using the knowledge base 6, and classifying means 7 which collates classification data in the management data D to a concept network 8, checks whether or not the data are related to an already existing node, and applies attribute data E of the node when the data are related, and applies the attribute data E of the newly prepared node when the data are not related.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an electronic filing apparatus for filing a document / image as image data. There is a demand for an electronic filing apparatus to appropriately register a document / image information as image data and perform an efficient search.

[0002]

2. Description of the Related Art FIG. 5 is a diagram showing the concept of registration in a conventional electronic filing apparatus. In the conventional electronic filing device, when registering a document / image, the document / image A
Is input by the image input means 1 and written in the file device 2 by the registration means 3 together with the management data D. The management data D is composed of titles, keywords, etc. to be used at the time of search, and is added and managed for each case. It is troublesome to enter the management information for each case from the keyboard, so you can enter the management information for multiple documents / images all at once from a floppy disk, etc., or enter the management information on the management sheet and enter the information. A method of inputting a character using a character recognition means, a method of inputting management information using a mark sheet, and the like have been developed. Furthermore, the registrant classifies as necessary, creates cabinets, drawers, guides, etc., and puts them in appropriate places.

However, these methods still require the registrant to fully understand the contents of the document / image to be registered and appropriately extract necessary information.

[0004]

Therefore, in the case of a registrant who has insufficient knowledge and understanding about the document / image to be registered,
It is impossible to assign appropriate keywords. Since appropriate keywords are not assigned, there will be a problem in the efficiency of search.

Further, for the convenience of retrieval, it is also necessary to classify and register each related field according to the content of the document / image to be registered, and in combination with the above-mentioned appropriate keyword addition, Registration efficiency is also a problem.

The present invention automatically adds optimum management information and performs automatic classification to automatically register documents and images without being affected by the knowledge and comprehension of the registrant. The purpose is to realize a filing device.

[0007]

FIG. 1 is a block diagram showing the principle of the present invention. In an electronic filing apparatus having an image input means 1 for inputting a document / image A, a file device 2 for storing image data F and management information G, and a registration means 3, a character recognition means 4, a management data An extraction unit 5, a knowledge base 6, a classification unit 7, and a concept network 8 are provided.

The knowledge base 6 is a database in which rules / rules necessary for extracting the management data D are accumulated. The concept network 8 is a database in which individual data of the classification data are used as nodes and their relations are accumulated as attribute data E.

The character recognition means 4 recognizes the character information in the image data B input by the image input means 1 and converts it into code data C. The management data extracting means 5 extracts the management data D from the recognized code data C using the knowledge base 6.

The classification means 7 checks the classification data of the extracted management data D with the concept network 8 to check whether or not it is related to an already existing node. A node is newly created and the attribute data E of the node is added.

The registration means 3 stores the filed image data F in association with the management information G consisting of management data and attribute data in the file device 2. Here, a feature extracting means 9 for the image data B is further provided,
The characteristic data H extracted thereby can be made a part of the management data D.

[0012]

As shown in FIG. 2, the concept network is a database composed of concept nodes and relational links, each node and link having a semantic weight, and constituting a network. That is, the knowledge representation method called "conceptual relation model", which is a simplified representation of knowledge focusing on "concepts" and "relationships", introduces the concept of the entity relation model. A concept node is a "concept" itself classified by a knowledge base, and a relational link is "~".
Included in ~ "," is in ~ "," based on ~ "
, Which defines each "relationship".
Here, the “concept” (node) is individual classification data, and their “relationship” (relationship link) is attribute data.

First, an appropriate concept network (database) is constructed. To construct the concept network, information (rules) such as knowledge of a related field handled by a document / image to be registered and a document file system, which is constructed in the knowledge base 6, is used. For example, a representative document / image is analyzed and classified by the knowledge base 6. Build nodes according to this classification.

It is assumed that such a conceptual network is constructed. FIG. 3 shows an example of a conceptual network used in the embodiment of the present invention. Each node corresponds to a drawer of the filing system. When registering a document / image, the document / image to be registered is analyzed / classified by the knowledge base 6, it is checked whether the classification classification of the registered document / image is related to the node being constructed, and if it is related, The attribute information of that node is added. When it is possible to further classify, a new node is constructed in a form linked by a relational link. In this way, which node the document / image to be registered is related to is checked, attribute information of the related node is added, and a new node is constructed.

With this configuration, the registrant can give the optimum management information without requiring knowledge or experience, and without being aware of the management information at the time of registration. Also, classification is automatically performed by the concept network.

[0016]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 4 is a block diagram of an embodiment of the present invention. Figure 1
Elements having the same functions as those are indicated by the same reference numerals.

The document / image A is converted into image data B by the image scanner 1 which is an input means, and written in the memory of this apparatus. Note that image data recorded on a floppy disk or the like generated by another device may be read. In this case, a floppy disk drive (not shown) is used as the input means, but there is no essential difference.

The character portion of the image data on this memory is converted into code data C by the character recognition program 4. Code data C is management data extraction program 5
By using the knowledge base 6, lexical analysis,
Title as the management data, creator name, creation date,
Extract keywords and classification data.

An example of knowledge for extracting management data is shown below. Title: [Exists at the beginning of the document. ] [There are labels such as "name", "title", and "title". ] Creator: [Labels such as "author", "creator", etc.] [Next to face photo] [Specific business card] Keywords: [Word with subtitle, item number] [Same word] Many) [Underlined, shaded, framed] [Noted] Classification data is automatically analyzed by analyzing the meaning of words from titles, keywords, etc. (Number etc.) In this case, it defines whether or not attach any number to the advance which classification.

As the characteristics of the document / image to be registered, the ratio of the area occupied by the characters and the drawing, the layout characteristics (where the drawing is located vertically, horizontally, etc.) can be extracted and used as a kind of keyword. . Feature extraction program 9
Such feature data H is extracted from the image data by.

The classification program 7 matches the classification data in the management data D and the characteristic data H with the concept network, and updates the concept network as necessary,
Extract attribute data.

The registration program 3 stores the management data and the like thus extracted as management information G in the magneto-optical disk 2 together with the image data F which is the original document / image data.

When retrieving a document / image stored on the magneto-optical disk 2, the retrieval program 10 uses a keyboard
The management information is searched in accordance with the operator's instruction data (keyword etc.) input from the display 12.

Further, at the time of registration, the classification is performed based on the concept node of the concept network, so that the target document can be searched based on this concept node, and the related document can be easily obtained by the concept link. It will be possible to search.

For example, suppose that the user wishes to retrieve the material relating to the electronic filing system in the present apparatus in which the conceptual network shown in FIG. At this time, in this conceptual network, the "meeting material" and the "new product" are linked by the relationship "study", and the "electronic filing" is linked by "development". This indicates that the agenda at the time of the meeting was the electronic filing system, which was related to new development. Furthermore, "maker" connecting "company A" and "electronic filing"
Indicates that a manufacturer called Company A develops an electronic filing device. Documents can be searched based on such nodes, and related documents can be searched by following links.

[0026]

As described above, according to the present invention,
Appropriate management information is automatically extracted and added by the knowledge base and concept network when registering document / image information, so you can enter management information without being affected by the level of knowledge and experience of the registrant. It is possible to register efficiently without doing. Therefore, it is possible to realize an electronic filing apparatus having a high efficiency in searching. Furthermore, by using this concept network, a browsing function for browsing stored information and knowledge can be realized. Therefore, it becomes possible to "remember and recognize memory".

[Brief description of drawings]

FIG. 1 is a principle block diagram of the present invention.

FIG. 2 is an explanatory diagram of a conceptual network.

FIG. 3 is an example of a conceptual network.

FIG. 4 is a block diagram of an embodiment.

FIG. 5 is a conceptual diagram of registration in a conventional electronic filing device.

[Explanation of symbols]

1 image input means (image scanner) 2 image file device (magneto-optical disk) 3 registration means (registration program) 4 character recognition means (character recognition program) 5 management data extraction means (management data extraction program) 6 knowledge base 7 classification means (Classification program) 8 Concept network 9 Feature extraction means (feature extraction program) 10 Search program 11 Management information input means 12 Keyboard / display device A Document / image B Image data C
Code data D Management data E Attribute data F Image data G Management information H Characteristic data

Claims (2)

[Claims] 1. An image input means (1) for inputting a document / image (A), a file device (2) for storing image data (F) and management information (G), and a registration means (3). ), An electronic filing device including: a character recognition means (4); a management data extraction means (5);
A knowledge base (6), a classification means (7) and a concept network (8) are provided, and the knowledge base (6) is a database in which rules necessary for extracting the management data (D) are accumulated, The concept network (8) is a database in which individual data of the classification data are used as nodes and their relations are accumulated as attribute data (E). The character recognition means (4) is input by the image input means (1). The character information in the image data (B) is recognized and converted into code data (C), and the management data extracting means (5) uses the knowledge base (6) from the recognized code data (C). Then, the management data (D) is extracted, and the classification means (7) compares the classification data of the extracted management data (D) with the concept network (8) to identify the already existing nodes. If it is related, the node is generated, if not, a new node is generated and the attribute data (E) of the node is added, and the registration means (3) sets the image data (F ) And management information (G) consisting of management data and attribute data are associated with each other and stored in the file device (2). 2. The feature data extraction means (9) for image data (B) is provided, and the feature data (H) extracted thereby is used as a part of the management data (D). Electronic filing equipment.