JP2518000B2 - Electronic document retrieval device - Google Patents

Description

TECHNICAL FIELD The present invention relates to a search device for electronic documents.

(Prior Art) In searching for documents, images, etc., there are cases where a search based on ambiguous images is required. For example, a designer might search for something that fits his image, or a memory of what he saw before.

As an indexing method for such a search, a method is used in which a plurality of attributes having an interval scale property are regarded as coordinate axes, and the image of each document is indexed as points in a multidimensional coordinate space. In this index-based search, search conditions are described as values on some axes.

(Problems to be solved by the invention) This method describes an image by decomposing an image into elements represented by several axes, but it is possible to decompose the image in this way to some extent. . Therefore, a searcher can clearly specify values for a small number of axes, but cannot specify values for all other axes, which results in a problem that sufficient search accuracy cannot be obtained. For example,
When retrieving a facial photograph based on memory, a method of decomposing facial features into elements such as "eye size", "eyebrow depth", and "dull feeling" and configuring an interval scale for each element is possible. Are known. However, since the memory of the face is retained as an overall impression, it is only a few features that the searcher can clearly indicate the value of the factorized scale. Therefore, it has been pointed out that a sufficient search accuracy cannot be obtained.

To solve the above problems, the image is not decomposed into its elements, divided into multiple groups according to their overall similarity, a representative example of each group is presented, and which example is most similar to the one searched. There has been proposed a method of narrowing down by allowing a searcher to specify whether or not. However, this method has a drawback in that accurate information cannot be searched because the characteristic information that the searcher can clearly point out cannot be effectively used.

For example, if the characteristics that the searcher can clearly point out are not reflected in the grouping, there may be no difference between the representative examples regarding the characteristics that the searcher can clearly point out. In such a case, the searcher cannot specify a corresponding group or specifies a group that does not include the one to be searched, and thus a great deal of labor is required for the search.

I can't find the document I want.

Problems such as occur.

The first object of the present invention is to solve such problems,
It is to improve the search efficiency by utilizing the information of the features that the searcher can clearly specify, analyzing the part that cannot be clearly specified, and presenting the result. More specifically, the present invention, in a search using an index based on an interval scale, performs a search by a scale with which a searcher can clearly specify a value, and performs a principal component analysis on another scale for the searched document, By presenting the analysis results, the search efficiency is improved.

A second object of the present invention is to provide a method of using a principal component analysis result regarding a scale of a retrieved document, which is not used as a search condition, in a browsing direction instruction in a search using an index based on an interval scale. To do.

A third object of the present invention is to provide a method for using as a new search clue the result of principal component analysis regarding the scale of a searched document that was not used as a search condition in the search using the index based on the interval scale. Especially.

(Means for Solving Problems) According to the present invention, there is provided a retrieval device for an apparatus for storing and displaying an electronic document, the input means for inputting information of an instruction regarding retrieval of the electronic document, and the electronic computer. A document storage means for storing a document, a multidimensional spatial structure storage means for storing a multidimensional spatial structure for organizing an electronic document by a plurality of axes, and a search condition input by the input means, in the document storage means. A search unit that searches the stored document for the corresponding document, a search result analysis unit that performs a principal component analysis on the axes that are not included in the search condition for the document searched by the search unit, and the search result analysis unit. Principal component selecting means for selecting one or more principal components from the obtained principal components, example selecting means for selecting an example document of the principal component selected by the principal component selecting means, Analysis result presentation information creating means for creating display information of the main component selected by the main component selecting means; display means for displaying data from the document storing means and display output from the analysis result presentation information creating means; An electronic device comprising control means for controlling the data from the document storage means and the data from the search result analysis means according to the output of the input means, and supplying the data to the display means via the analysis result presentation information creating means. An apparatus for retrieving a document is obtained.

Further, according to the present invention, there is provided a search device for a device for storing and displaying a digitized document, the input means for inputting information of an instruction regarding search / selection of the digitized document, and a document for storing the digitized document. Storage means, multidimensional spatial structure storage means for storing a multidimensional spatial structure for organizing an electronic document by a plurality of axes, and a document stored in the document storage means according to a search condition input by the input means. Retrieval means for retrieving the relevant document, retrieving result analyzing means for subjecting the documents retrieved by the retrieving means to principal component analysis regarding axes not included in the retrieval conditions, and principal component obtained by the retrieving result analyzing means. A main component selecting means for selecting one or a plurality of main components from the above, an example selecting means for selecting an exemplary document of the main component selected by the main component selecting means, and the main component selecting hand. Analysis result presentation information creating means for creating display information of the main component selected by, and document selecting means for selecting a relevant document from the documents searched by the searching means in accordance with the document selection information input by the input means. Display means for displaying the data from the document storage means and the display output from the analysis result presentation information creating means, and the data from the document storage means and the data from the search result analysis means according to the output of the input means. An electronic document retrieving apparatus including control means for controlling and supplying to the display means via the analysis result presentation information creating means is obtained.

Further, according to the present invention, there is provided a search device for a device for storing and displaying an electronic document, the input means for inputting information of an instruction regarding the search / extraction of the electronic document, and a document storing the electronic document. Storage means, multidimensional spatial structure storage means for storing a multidimensional spatial structure for organizing an electronic document by a plurality of axes, and a document stored in the document storage means according to a search condition input by the input means. Retrieval means for retrieving the relevant document, retrieving result analyzing means for subjecting the documents retrieved by the retrieving means to principal component analysis regarding axes not included in the retrieval conditions, and principal component obtained by the retrieving result analyzing means. Principal component selecting means for selecting one or a plurality of principal components from the above, example selecting means for selecting an example document of the principal component selected by the principal component selecting means, and the principal component selecting A document extraction for selecting a relevant document from the documents retrieved by the retrieving means by the analysis result presentation information producing means for producing the display information of the main component selected by the step and the document extraction information inputted by the input means. Means, display means for displaying the data from the document storage means and the display output from the analysis result presentation information creating means, and output of the data from the document storage means and the data from the search result analysis means to the input means. An electronic document retrieving apparatus is provided which is controlled according to the above, and which includes control means for supplying the information on the display means via the analysis result presentation information creating means.

(Operation) In the present invention, in the search using the index based on the interval scale, the searcher performs a search by a scale whose value can be clearly specified, and the searched document is subjected to a principal component analysis on another scale to determine the main component. Visual representation of typical examples and their positions. An example of display information on the display means is shown in FIG. In this example, the first to third main components are displayed. As a typical example of each principal component, a document in which the maximum value, the median value, and the minimum value of the principal component scores are closest to each other is presented.

Further, these display information can be used for inputting the document selection instruction information. For example, in FIG. 4, when the position information of the star on the axis indicating the first principal component is input from the input means, it is closest to the position indicated by the input principal component score in the search document set. The document is displayed.

Further, these display information can be used for inputting search conditions. For example, in the state of FIG. 4, when the rectangular information indicated by the solid line on the axis indicating the second principal component and the third principal component is input, the second and third principal component scores fall within the designated range. A document is extracted.

The principal component obtained here is a comprehensive scale that does not allow the searcher to specify the value clearly, and at the same time, is a scale with the largest variance of the search result documents along it, and It is the best indicator of the characteristics. Therefore, even if the searcher cannot specify the decomposed values on each scale, it can be expected that it is possible to specify the principal component score that is the comprehensive characteristic value. Further, since the principal components are orthogonal to each other, using the principal component score as the search condition is the most efficient method.

Also, since the main component is a composite of multiple scales, it is a scale that has no name by itself. Usually, it is often the case that the principal component is interpreted from the contribution rate of the original scale and the name is given to itself. However, the interpretation and naming of the principal components is done subjectively and is often unreasonable. Therefore, in the present invention, the scale of each principal component is shown using an example. In this method, no interpretation is performed on the obtained main component, and abstract information such as words is not used as a medium, so problems such as individual differences in interpretation and naming do not occur. Furthermore, since specific information such as an example document can be used, it is easy to understand and it is possible to absorb individual differences and time variations in the interpretation of the divided scale.

(Example) Hereinafter, the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing a first embodiment of the present invention. The input means 1 of this embodiment is, for example, a pointing device such as a mouse or a trackball. If the operability is sacrificed a little, an input means such as a keyboard is also possible. As a result, it is possible to specify or change the document search / extraction conditions and input the document selection information. The document storage means 2 is a means for storing a plurality of digitized document data, and is composed of, for example, an optical disk device or a magnetic disk device.

In addition, the multidimensional space structure storage unit 3 stores the axis number, axis name,
And means for storing the digitized document number information corresponding to each position on each axis for a plurality of axes. Search means 4
Is controlled by the control means 10 described later, and the input means 1
It is a means for searching the document stored in the document storage means 2 for a corresponding document according to the search condition input by.

The search result analysis means 5 performs a principal component analysis on the document searched by the search means 4 regarding an axis not included in the search condition. FIG. 5 is a functional block diagram showing an embodiment of the search result analysis means 5. The correlation matrix creating means 51 of the present embodiment creates a correlation matrix for an axis not included in the search conditions. The principal component calculating means 52 is a correlation matrix creating means 51.
It is a means for obtaining the principal component from the correlation matrix created by. The contribution rate calculating means 53 calculates the calculated contribution rate of the main component.

The principal component selecting means 6 is a means for selecting a principal component to be presented as an analysis result. As a specific example of the method of selecting a principal component, a method of selecting a principal component having a contribution rate equal to or higher than a predetermined value, a method of selecting a principal component until the cumulative contribution rate reaches a predetermined value, or the like, Alternatively, there is a method for the user to manually select.

The example selection unit 7 is a unit that selects the example document of the main component selected by the main component selection unit 6 based on the search condition information output from the input unit 1 and the output of the search result analysis unit 5. As a specific example of the selection method of the example document, the range of each principal component score is divided into several ranges, and the document closest to the center of each divided range is selected. , The median, and the method of selecting the document closest to the minimum value.

FIG. 6 is a flowchart showing an example of detailed processing of the example selecting means 7 in the case where the range of each principal component score is divided into several ranges and the document closest to the center of the divided range is selected. Is expressed in. First, with the first principal component as the relevant principal component, in step 72, the relevant principal component score of the search result document is calculated. Next, in step 73, the principal component scores of the retrieved document set are sorted in ascending or descending order and set in the document number / principal component score table. Next range division step
At 74, the range is divided for the relevant principal component. In step 75, the search result document set is divided according to which of the divided ranges is included using the sorting result of step 73. Next, the center of each range is calculated at step 76. In the selection step 77, the division information created in step 75 is used to select the document closest to the center of gravity for each range, and this is set as the example document of the corresponding principal component. Step
At 78, the division information of the corresponding principal component and the example document information are set in the example document table. In step 79, if there is a next principal component, that principal component is regarded as the corresponding principal component and the process returns to step 72.
Next, if there is no main component, a processing end signal is output.

The analysis result presentation information creating unit 8 is a unit that creates display information about the main component selected by the main component selecting unit 6 based on the output of the example selecting unit 7. That is, the analysis result presentation information creating means 8 is a means for setting the main component number for displaying the headline, the scale information, and the display position information of the example document, and the actual operation is controlled by the control means 10 described later.

The display unit 9 displays the data from the document storage unit 2 and the display output from the analysis result presentation information creating unit 8. The display unit 9 is composed of a normal CRT display or the like.

Further, the control means 10 is means for controlling a series of search processing, controls the data from the document storage means 2 and the data from the search result analysis means 5 in accordance with the output of the input means 1, and creates analysis result presentation information creation means. Display means 9 via 8
Is a means to supply to.

FIG. 7 is a flow chart showing an example of a detailed procedure of the control means 10. First, when a data input signal is given from the input means, an operation instruction is given to the search means 4. When the processing end signal is given from the search means 4, an operation instruction signal is given to the search result analysis means 5. When the processing result signal is given from the search result analysis means 5, the operation instruction signal is given to the main component selection means 6. Similarly, after the operation of the main component selecting means 6 is completed,
The example selecting means 7 is operated. When the operation of the example selection unit 7 is completed, the first principal component of the exemplary document table set by the unit is set as the relevant principal component. Axis display step 106
Display the relevant main component number, axis line and scale information.
Next, in step 107, the first example document of the relevant principal component is set as the relevant document. In step 108, the document storage means 2
Read the information of the relevant document from. If it is determined in step 109 that the size of the example document is too large for the displayable area, step 110 reduces the size to an appropriate size. Example In the display step 111, the document information or the reduced information is displayed at a predetermined position on the display means 9. If there is a water example document, the process proceeds to step 112, and in step 113, it is set as the corresponding document, and the process returns to step 108. If there is no next example document, the process proceeds to step 114, and it is checked whether or not there is the next principal component. If there is no next main component, the process proceeds to step 116, the current relevant document is displayed on the display means 9,
The position information of the relevant document is set, and the process ends.

Next, FIG. 2 shows a second embodiment of the present invention. Here, the same number as the means in FIG.
It is a means having the same function. The document selection unit 11 is a unit that selects a corresponding document from the search result documents in accordance with the document selection instruction information input from the input unit 1. That is, the document selecting means 11 uses the document number / principal component score table created by the example selecting means 7 to set the score of the main component designated by the document selection instruction information to a value corresponding to the designated position. Select the closest document.

FIG. 8 is a flow chart showing an example of detailed processing of the control means 10 when the input means 1 is realized by a mask. Steps 100 to 117 are the same processes as in FIG. Mouse click detection In step 118, it is detected whether or not the mouse is clicked.
Continue to 119. If step 119 does not determine the end, step 120 detects the coordinates corresponding to the mouse cursor. If the position information of the currently displayed document set in step 117 is compared with the position information indicated by the detected coordinates, and it is determined that another document is displayed, the document selection means 11 , Select the relevant document. Then, the process returns to the corresponding document display step 116. Also,
If, for example, the end key is pressed or the end mouse button is clicked in step 119 and the end is considered, the operation is completed.

Next, FIG. 3 shows a third embodiment of the present invention. Here, the same number as the means in FIG.
It is a means having the same function. The document extraction unit 12 is a unit that extracts the relevant document from the retrieved document set based on the document extraction information input from the input unit 1. That is,
The document extracting means 12 uses the document number / principal component score table created by the example selecting means 7 to extract a document in which the score of the principal component designated by the document extraction information is included in the designated range.

FIG. 9 is a flowchart showing an example of detailed processing of the document extracting means 12. First, in step 81, the first principal component included in the document extraction condition is the relevant principal component, and the first column of the intermediate table is the relevant column. Next, with reference to the document number / principal component score table, the number of the document included in the extraction range of the relevant principal component is set in the relevant column of the intermediate table. In step 84, it is determined whether or not there is a next principal component, and if there is a corresponding principal component, the next column of the intermediate table is the relevant column, and the process returns to step 83. If there is no next principal component, go to step 87. If no document number is set in the intermediate table, an error message is displayed on the display means 9. When the document number is set in only one column of the intermediate table, the content is set in the extraction result table. AND if the document numbers are set only in multiple columns in the intermediate table
The obtained result is set in the extraction result table. When the data set in the extraction result table or the display of the error message is completed, the processing ends.

FIG. 10 is a flowchart showing an example of detailed processing of the control means 10 when the input means 1 is realized by a mouse and the extraction range is designated by a rectangle. Steps 100 to 119 are the same processes as in FIG.

In step 118, it is detected whether or not the mouse is clicked, and if it is clicked, the process proceeds to step 119. If it is not regarded as the end in step 119, the process proceeds to step 123, and it is determined whether or not the document extraction condition is set. If the document extraction condition settings are also determined, step 12
Enter the extraction conditions in step 4. Depending on the input extraction conditions,
The relevant document is extracted by using the document extracting means 12, and the process ends. Also, if in step 119, for example,
The operation is completed when the end key is pressed or the end mouse button is clicked and the end is considered.

(Effects of the Invention) As described above, according to the present invention, in a search system that uses an index based on an attribute axis having the property of an interval scale, information that can be clearly specified by a searcher is utilized, and search results An efficient search clue that is easy for the searcher to specify according to the characteristics and is efficient is automatically configured, and the search efficiency is improved. In addition, because the headline is displayed visually using a specific example, it is possible to understand the characteristics of the search results.
New search clues can be easily used and the user interface is improved.

[Brief description of drawings]

FIG. 1 is a functional block diagram showing a first embodiment of the computerized document retrieval device of the present invention. 2 and 3 are block diagrams showing second and third embodiments of the present invention, respectively. Further, in order to facilitate understanding of the method of the present invention, FIG. 4 is a diagram showing a specific example of display of the principal component analysis result and its use for document selection instruction and search condition input, and FIG. 1 is a flow chart showing an example of detailed processing of the search result analyzing means 5 in the embodiment, FIG. 6 is a flow chart showing an example of detailed processing of the example selecting means 7 in the embodiment of FIG. 1, and FIG. 6 is a flowchart showing an example of a detailed procedure of the control means 10 in the illustrated embodiment. FIG. 8 is a flow chart showing an example of a detailed procedure of the control means 10 in the second embodiment.
FIG. 9 is a flow chart showing an example of detailed processing of the document extracting means 12 in the third embodiment, and FIG. 10 is a flow chart showing an example of detailed procedures of the control means 10 in the third embodiment. In the figure, 1 ... Input means, 2 ... Document storage means, 3 ... Multidimensional space structure storage means, 4 ... Search means, 5 ... Search result analysis means, 6 ... Principal component selection means, 7 ... ... Example selection means, 8 ...
… Analysis result presentation information creation means, 9 …… Display means, 10 ……
Control means, 11 ... Document selection means, 12 ... Document extraction means.

Claims (3)

(57) [Claims] 1. A retrieval device for storing and displaying a digitized document, comprising: input means for inputting information on an instruction relating to the retrieval of the digitized document; and document storage means for storing the digitized document. A multi-dimensional spatial structure storage means for storing a multi-dimensional spatial structure for organizing an electronic document by a plurality of axes and a search condition input by the input means are searched for a corresponding document from the documents stored in the document storage means. Or one of the principal components obtained by the retrieval result analyzing unit, and the retrieval result analyzing unit for performing principal component analysis on the axes not included in the retrieval condition for the document retrieved by the retrieval unit. By using the principal component selecting unit that selects a plurality of principal components, the example selecting unit that selects the exemplary document of the principal component selected by the principal component selecting unit, and the output of the example selecting unit, Analysis result presentation information creating means for creating display information of the main component selected by the main component selecting means; display means for displaying data from the document storing means and display output from the analysis result presentation information creating means; Control means for controlling the data from the document storage means and the data from the search result analysis means according to the output of the input means, and supplying the data to the display means via the analysis result presentation information creating means. Electronic document search device. 2. A retrieval device for storing and displaying a digitized document, comprising input means for inputting information on instructions for searching and selecting the digitized document, and document storage means for storing the digitized document. A multi-dimensional spatial structure storing means for storing a multi-dimensional spatial structure for organizing an electronic document by a plurality of axes; and a document stored in the document storing means according to a search condition input by the input means. One of the search means for searching for, the search analysis means for performing the principal component analysis on the axis not included in the search condition for the document searched by the search means, and the main component obtained by the search result analysis means. Alternatively, a principal component selecting unit that selects a plurality of principal components, an example selecting unit that selects an exemplary document of the principal component selected by the principal component selecting unit, and a component selecting unit that selects the principal component by the principal component selecting unit. According to the analysis result presentation information creating means for creating the display information of the main component, and the output of the example selecting means in accordance with the document selection instruction information input by the input means A document selecting means for selecting a document to be stored, a display means for displaying data from the document storing means and a display output from the analysis result presentation information creating means, and a data from the document storing means and a search result analyzing means. An electronic document retrieving apparatus comprising: a control unit that controls data according to the output of the input unit and supplies the data to the display unit through the analysis result presentation information creating unit. 3. A retrieval device for storing and displaying a digitized document, comprising: input means for inputting information on instructions for searching / extracting the digitized document; and document storage means for storing the digitized document. A multi-dimensional spatial structure storing means for storing a multi-dimensional spatial structure for organizing an electronic document by a plurality of axes; and a document stored in the document storing means according to a search condition input by the input means. And a search result analyzing unit for performing a principal component analysis on an axis not included in the search condition for the document searched by the searching unit, and a main component obtained from the search result analyzing unit. Principal component selecting means for selecting one or a plurality of principal components, example selecting means for selecting an example document of the principal component selected by the principal component selecting means, and selecting by the principal component selecting means. From the document searched by the searching means using the output of the example selecting means, by the analysis result presentation information creating means for creating the display information of the main component and the document extraction information input by the input means. Document extracting means for selecting a document, display means for displaying data from the document storing means and display output from the analysis result presentation information creating means, data from the document storing means and data from the search result analyzing means Is controlled according to the output of the input means, and is supplied to the display means via the analysis result presentation information creating means.