JP2695784B2 - How to search and display sentences - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a digitized document information reading terminal, and in particular, retrieves a description related to the document from a text based on a chart and displays it side by side. About the system. [Prior Art] According to an operator's interest, a device for displaying information such as a CRT display and a device such as a cursor, a mouse, a keyboard, and a tablet for selecting one of the displayed items are used. When one of the displayed items is selected, a more detailed description of the selected item is drawn out, and a system has already been put to practical use in a captain system or the like. However, in these systems, the information to be displayed corresponding to the selected item is programmed in advance, so that the operator prepares the information prepared in advance by the system side only for the items that are allowed to be selected on the screen. It can be seen by tracing sequentially. For this purpose, the system provider prepares information to be displayed corresponding to all selected trees, creates data relating the location of the information corresponding to the selection, and registers it in the system. It is necessary. On the other hand, key word retrieval of document information stored in an electronic database has been performed, but search and display of mutually related parts in one document have not yet been performed. Not. [Problems to be Solved by the Invention] In the above-mentioned prior art, the creation and registration of the above-mentioned data to be associated is necessary in advance in order to display the relevant portion according to the selection from the objects displayed on the screen. Yes,
Making such preparations when the volume of literature is large can be an enormous task. SUMMARY OF THE INVENTION An object of the present invention is to provide a method for realizing a function of searching and displaying a related part in a document, which has not been realized conventionally because of a large amount of work. In particular, it is an object of the present invention to provide a method of displaying a related description portion in a text based on an instruction of a display position of an item in a chart. [Means for Solving the Problems] In order to achieve the above object, the present invention comprises the following element steps. (1) A step of causing the operator to designate a position in the display screen and specifying a designated word of the designated object. Here, the descriptive term is the number of a chart, the number of a block unit element of a chart, etc.,
Means the name. (2) A step of searching for the above-mentioned descriptive term from the text. (3) A step of cutting out, shaping, and displaying an appropriate part before and after from the found part. [Operation] FIG. 5 exemplifies a screen of the display CRT 6 of the terminal when the present invention is implemented. First, the drawing is displayed in the frame 51. Here, when the operator indicates a position on the screen, for example, an instruction word 52 of an element in a certain figure is indicated, and the system recognizes the instruction word from the position and the data on which the display is based. And identify. This is the first step. Next, a word that matches the specified descriptive word is searched for in the body text. This is the second step. From the matched part, the appropriate text before and after that is cut out in a range that can be displayed on the screen, placed on the same CRT 6 screen, and displayed as the frame 53 in the figure. This is the third step. It includes, of course, a display 54 of the instruction word. This can also be highlighted. Since each of the above-described steps has a problem, an example will be described below. Embodiment An embodiment of the present invention will be described below with reference to the flowchart of FIG. The description will be made according to the order in the figure. Step 101 is to display a portion of a page or a page of a chart or a document containing the chart.
This is displayed based on the instruction of the operator. The step 102 is to detect the operator looking at the displayed content and pointing to the whole of a certain chart or the approximate part of the chart to ask for an explanation of the element, or in other words, pointing. It is. Various tools can be used for this. For example, there are direct movement using a transparent tablet or a separate tablet superimposed on the display, such as moving the cursor and pressing a predetermined key, moving the mouse and pressing a button. The step 103 specifies an object to be pointed by the operator based on the input value of the position on the screen where the step of 102 was specified, and further obtains a descriptive word of the target. This may be, for example, the number of the entire figure or table, the number for a citation given to an element in the drawing of the patent, or the name written in its place. In step 104, the descriptive word obtained in step 103 is scanned and matched with the text to search for a matching location. In step 105, a sentence including the part before and after is extracted from the place in the text obtained in step 104, arranged, shaped, and displayed. After this, the process returns to step 102 and waits for the next operation. The method described above with reference to FIG. 2 does not require a preparation process in advance, and can be immediately applied to a document retrieved from a database. Also, no separate storage area is required. However, the search step of 104 generally requires significant processing time and is likely to delay response to the operator. FIG. 3 is a flowchart showing another embodiment for shortening the response time. Step 110 is a step of processing a document as a preparation step first when a document is selected, scanning a text or a chart to find all parts that may be selected as descriptive words, and calling its position. At the same time, it is recorded in the referential word positional relation table. Sort the descriptive words as necessary. Steps 101 to 103 and 105 are the same as the processing in FIG. 2 and the description is omitted. Step 114 replaces 104 in FIG. 2,
The position in the text corresponding to the instruction designated by the operator is obtained from the table using the instruction-position relation table created in 110. Therefore, the processing time is very short. The embodiment shown in FIG. 3 is characterized in that the response time is shorter than that of FIG. Here, a concrete implementation method of step 103 will be described. It is easy to convert input information from an indicator such as a mouse or a tablet into a coordinate system in units of pixels for display by a simple calculation. There are two types of indexing methods according to the display method. That is, a first case in which a display image is generated from code data such as characters and lines, and a second case in which the display image is reproduced based on data originally input as an image. In the first case, a method used to determine an element pointed by CAD (Computer Aided Design System) or the like can be adopted. An example is shown in FIG. FIG. 1 shows a system configuration of an embodiment of the present invention. There are code data 2 of a chart of a certain document read from the document database 1 and code data 3 of a sentence,
The code data of at least one chart is sent to the drawing processing unit 4 and written into the frame memory 5 based on the code data. The contents of the frame memory 5 are immediately displayed on the CRT 6. A transparent tablet 7 is set on the display surface 6. When the operator designates a certain element of the chart displayed by the touch, the screen input 81 is obtained by the tablet input circuit 8, and the touch operation is reported to the control unit 9. Here, the control unit 9 performs drawing processing of the chart again to search for the element in the designated chart, and generates a write address to the frame memory or the screen coordinates 41. The proximity search unit 10 determines that the writing screen coordinates 41 are the instruction screen coordinates 81.
Is detected within a certain distance of. When this is detected, the controller 9 is informed and the drawn element is determined as the element being drawn by the drawing processor, and the specified element is obtained. In addition, in the case of a system that uses a random scanning type CRT, does not have the frame memory 5, and performs repetitive drawing processing, the specified element is used as a drawing element at the moment when the drawing beam is detected by the light pen. There is a way to ask. Also, instead of drawing all elements, the instruction screen coordinates 81
Therefore, a method of shortening the search time by performing a proximity search by limiting a target element has been proposed. When the element obtained as described above is a character string, and when it is a graphic element, a descriptive word is obtained as a name stored correspondingly. As a second case, a method for obtaining a target word designated by an operator when a diagram or table is originally image data and the data is displayed is as follows. An example is shown in FIG. FIG. 4 shows a system configuration of an embodiment of the present invention. Image data 2 'of a chart of a certain document read from the document database 1'
And textual image data 3 ', and at least
Two charts are written to the frame memory 5 via the writing section 4 ', and are immediately displayed on the CRT 6. When the transparent tablet indicates an object having a chart, the coordinate input unit 8 sends the image coordinates to the control unit 20. The control unit 20 starts the following series of processes. First, a region near a point designated by image coordinates is extracted by the extraction unit 21, and a segment considered to be a character is extracted from the extracted region. In addition, not only the segment closest to the designated point, but also the segments that are closer to the designated point within a certain distance are sequentially extracted and sent to the next character recognition unit 22.
In character recognition, each of these segments is recognized as a character candidate, and an instruction is obtained as a quote number / symbol or word. The concrete implementation method of the step 103 has been described above together with the configuration example of the system. Next, a specific method for searching for the instruction word obtained in step 103 from step 104, that is, the text, will be described. Also at this time, there are a first case where the text is stored as code data and a second case where the text is stored only as an image. In the first case, the same character string as the descriptive word may be searched for from the code string of the sentence, and this part is a process often performed as a string search. The matching search unit 11 in FIG. 1 performs this. The document data 3 stored in the code is scanned to search for a part that matches the code string of the instruction word specified by the control unit 9. In the second case, image processing is required. Part of FIG. 4 is an example of the implementation. First, the character generation unit 23 generates an image pattern corresponding to the character of the instruction word according to the font and size of the character used in the sentence data 3 '. This pattern is passed to the pattern search unit 24.
The pattern search unit scans the image of the sentence data 3 ', sequentially extracts local patterns, compares the extracted local patterns with the designated image pattern, and searches for a local part that substantially matches.
The difference from the character reading device is that the character reading device needs to recognize each character candidate in the image from thousands of character categories, whereas this device has a local pattern that matches the specified pattern. You just need to search for the pattern. For this reason, the processing amount is gradually reduced. In order to further reduce the processing amount of this search, when the designated word is composed of a plurality of characters, the designated word is separated into individual characters,
First, only one character in the character string is searched, and when a match is detected, it is preferable to check whether the next character exists in an adjacent local area. By doing so, it is possible to cope with the case where the indicator is present separately at the end of the line and at the beginning of the next line. First, as in a frameless character reading device, a line position is extracted and a character pitch is extracted.
After cutting out one character at a time, a collation method can be adopted. On the other hand, if a dedicated circuit for high-speed matching is used, a method of searching the entire surface from the beginning is also possible. In order to further reduce the search time, there is a method using knowledge specific to the target document. For example, in the case of a patent publication, since the elements in the figures are given numerical designations, it is considered that what the operator indicates is also a numeral. On the other hand, in a sentence, these numerical descriptive words appear in ascending order of numerical values. Even if it appears several times, the first appearing place is in order, and it is considered that the element is explained at that place. Accordingly, a correspondence table between the number of the descriptive word and the position in the sentence can be created while performing this search several times. When searching for a new descriptive word, it is possible to estimate which part to search for by referring to this correspondence table, and it can be expected that the search can be performed in a shorter time than searching the entire sentence. . Not only examples of patent publications,
The same is true for table numbers. The choice between the two cases for 103 steps and the two cases for 104 steps depends on the type of data supplied. The embodiment of FIG. 1 corresponds to the case where both are in the form of a code, and the embodiment of FIG. 4 corresponds to the case where both are in the form of an image.
The text may be stored in a code, and in this case, the text can be configured in combination. Create the predicate positional relationship table shown in FIG. 3 in advance 110
In the step (1), in the case of a patent document, it is sufficient to create a figure / table number and a designation by the number. Also, if a general document is also prepared using only the figure / table number, the range of the estimated appearance position in the text of the element can be narrowed from the figure / table number including the element designated by the operator. The display of the text of the 105 steps may be performed as follows. When the appearance position of the descriptive word is found, several lines before and after it are cut out and displayed. Alternatively, it is conceivable to detect and output a paragraph including the position. However, since the paragraph may be long, it may be limited to several lines before and after that. In any case, the display range may be shifted by a scroll instruction. As for the display arrangement, the position of the sentence display may be set on the screen excluding the entire chart or the vicinity instructed by the operator or the screen excluding the vicinity instructed. When the screen is small, the original chart may be covered, but it is sufficient to easily return to the display of the chart again. [Effects of the Invention] According to the present invention, it is possible to automatically search for a corresponding sentence and display it side by side only by pointing to an element in a diagram,
The operator does not need to search in the text, and the efficiency for understanding the document is improved. This is a new function that was not possible with the conventional method of viewing documents bound on paper.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram of one embodiment of the present invention, FIG. 2 is a diagram showing a flow chart of the operation according to the present invention, and FIG. 3 is another flow chart of the operation according to the present invention. FIG. 4 is a block diagram showing another embodiment of the present invention. FIG. 5 is a diagram showing a display example according to the present invention. 2 ... Chart table data, 3 ... Text data, 4 ... Drawing processing unit, 8 ... Coordinate input unit, 11 ... Match search unit, 24 ... Pattern search unit.

   ────────────────────────────────────────────────── ─── Continuation of front page    (72) Inventor Yasuaki Nakano               1-280 Higashi Koigabo, Kokubunji-shi               Central Research Laboratory, Hitachi, Ltd.                (56) References JP-A-63-62033 (JP, A)

Claims (1)

(57) [Claims] Searches for text that can be displayed sequentially or side-by-side according to the operator's request from text data including text data and chart data stored in the system.
In the displaying method, a command word for specifying the displayed chart data or its element according to the screen position designated by the operator is extracted from the displayed chart data of the chart data, and the extracted instruction is If the word is stored as an image, the extracted descriptive word is converted into character image data of the font and size of the character used in the sentence data, and the converted character image data is used as described above. A sentence characterized by searching the sentence data for the above-mentioned descriptive term, finding the location of the sentence data corresponding to the searched descriptive term, and displaying the corresponding document data based on the found location. How to search and display. 2. 2. The method according to claim 1, wherein the operator extracts a nearby area within a predetermined distance from the screen position, and extracts the first area from the extracted nearby area as a character. Are extracted, a second segment included at a predetermined distance from the extracted first segment is extracted, and each of the extracted second segments is recognized as a character candidate, and is recognized as a character. A method for retrieving and displaying a sentence characterized by extracting a sentence as the above-mentioned descriptive word. 3. In the method for retrieving and displaying a sentence according to claim 2, a graphic element and a name of the graphic element are stored in advance in association with each other, and the first and second segments are graphic elements. In some cases, a method for searching and displaying a sentence characterized by extracting a stored name as a descriptive word. 4. A method for retrieving and displaying a sentence according to any one of claims 1 to 3, wherein at least one of the sentence data and the chart data is scanned, and the occurrence of the descriptive word and the descriptive word is performed. The position is determined, the relationship between the determined descriptive term and the appearance position is stored as a descriptive term data positional relation table, and the location of the sentence data corresponding to the retrieved descriptive term is determined using the descriptive term data positional relation table. A method for searching and displaying sentences that are characterized by finding the position. 5. Searches for text that can be displayed sequentially or side-by-side according to the operator's request from text data including text data and chart data stored in the system.
In the displaying method, the displayed chart data or an element thereof is specified from the displayed chart data of the chart data according to the screen position specified by the operator, and an instruction including a numeral in at least a part thereof is specified. A word is extracted, the above-mentioned descriptive term is searched from the above-mentioned sentence data, the location of the sentence data corresponding to the searched descriptive term is obtained, and a correspondence table showing the correspondence between the obtained location and the above-mentioned descriptive term A method for searching and displaying a sentence characterized by searching a predetermined area in the sentence data using the correspondence table when performing the next search.