JP2021002278A - Information processing apparatus, control method, and program - Google Patents

Abstract

To provide an arrangement for properly analyzing an object to be analyzed.SOLUTION: According to the present invention, a bubble surrounding a word acquired as a result of morphological analysis is displayed. If an operation for fusion of the bubble is carried out with use of a mouse operation, then bubbles are fused and displayed as a single bubble. Alternatively, a word forming a word displayed with the operated bubble is fused to another bubble and displayed as a single bubble, and if an operation for separating a word forming the word indicated by the bubble is carried out, then the separated word is displayed in a new bubble.SELECTED DRAWING: Figure 5

Description

The present invention relates to an analysis method to be analyzed.

Recently, companies have been conducting activities that lead to sales promotion by analyzing customer feedback regarding their services and products.

In analyzing customer feedback, it leads to sales promotion by inferring interests and preferences from the comments of posters regarding services and products for services such as websites and SNS operated by the company.

In order to make such a guess, it is necessary to analyze the keywords included in the comments of the poster, but in general, morphological analysis is used as an analysis technique.

A keyword is extracted using such a technique, and the meaning of this keyword is inferred by specifying it using a term dictionary (see, for example, Patent Document 1).

JP-A-2007-226568

However, when such a mechanical technique is used, there are cases where keywords cannot be properly extracted from comments.

For example, if you analyze a comment that you purchased because it is a cashback campaign, it will be analyzed as cashback / campaign / because / purchased, but depending on how you think of keywords, cash / back / campaign / Therefore, there are cases where it is analyzed, such as / purchased.

Therefore, an object of the present invention is to provide a mechanism for obtaining an appropriate analysis result for an analysis target.

The present invention for achieving the above object is a display for displaying an analysis means for analyzing an analysis target located in a predetermined order and a set showing an arbitrary meaning in the predetermined order based on the analysis result by the analysis means. It is characterized by comprising means and an operation moving means for moving a portion included in the set so as to be included in another set adjacent to the set by operating the set displayed by the display means.

The present invention has the effect that an appropriate analysis result can be obtained for the analysis target.

It is a system block diagram which shows an example of the structure of an information processing system. It is a block diagram which shows the hardware configuration of an information processing apparatus. This is a sample corpus used in morphological analysis. It is a flowchart which renders the automatic analysis result from text information on a screen. It is a flowchart which identifies the operation type of a user. It is a flowchart which identifies a user's operation and reflects it on a screen. It is a flowchart which shows the method of saving the corrected automatic analysis result. This is an example of a screen that corrects the automatic analysis result on a character-by-character basis. This is an example of a screen that corrects the automatic analysis result word by word. This is an example of a screen for modifying the automatic analysis result by a special operation. This is an example of a screen that notifies a candidate for another word based on the automatic analysis result.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a system configuration diagram showing an example of the configuration of the information processing system according to the present invention. The detailed functions of each part will be described together with the flowchart described later.

The information processing device 100 includes a rendering unit 102, an operation identification unit 103, an operation reflection unit 104, and a data registration unit 105.

The morphological analyzer 101 indicates a morphological analyzer, and the result of automatic analysis by performing morphological analysis based on the information related to the corpus stored in the corpus 106 with respect to the document to be analyzed is transferred to the automatic analysis result DB 107. Remember.

The rendering unit 102 has a function of acquiring the automatic analysis result of the morphological analysis of the document from the automatic analysis result DB 107 and displaying it on the UI (user interface) screen. The operation identification unit 103 allows the user to display the UI screen. It has a function to identify the operation to be performed on the user and determine the intention of the user.

The operation reflection unit 104 functions to modify the automatic analysis result of the morphological analysis according to the user's intention recognized by the operation identification unit 103 and reflect it on the UI screen.

The data registration unit 105 functions to add the analysis result corrected by the operation reflection unit 104 to the original corpus 106 (correctly analyzed text data) or to the automatic analysis result DB 107 that saves the automatic analysis result. There is.

The morphological analyzer 101 can reuse the expanded corpus to encourage learning of new morphological analysis with higher accuracy.

Although the information processing device 100, the morphological analyzer 101, the corpus 106, and the automatic analysis result DB 107 are shown to be included in separate devices, the information processing is not limited to such a configuration and all configurations are processed. The configuration may be included in the device 100, or any function may be included in the information processing apparatus 100.

Hereinafter, the hardware configuration applicable to the information processing apparatus 100 according to the embodiment of the present invention will be described with reference to FIG.

FIG. 2 is a block diagram showing a hardware configuration of the information processing apparatus 100 shown in FIG.

In FIG. 2, 201 is a CPU that comprehensively controls each device and controller connected to the system bus 204.

Further, the ROM 202 or the external memory 211 is necessary to realize a function executed by a BIOS (Basic Input / Output System) or an operating system program (hereinafter, OS) which is a control program of the CPU 201, and each server or each PC. Various programs described later are stored.

Reference numeral 203 denotes a RAM, which functions as a main memory, a work area, and the like of the CPU 201. The CPU 201 realizes various operations by loading a program or the like necessary for executing a process from the ROM 202 or the external memory 211 into the RAM 203 and executing the loaded program.

Reference numeral 205 denotes an input controller, which controls input from a keyboard (KB) 209, a pointing device such as a mouse (not shown), or the like.

Reference numeral 206 denotes a video controller, which controls the display on a display such as a CRT display (CRT) 210. Although it is described as CRT210 in FIG. 2, the display may be not only the CRT but also another display such as a liquid crystal display. These are used by the administrator as needed.

The 207 is a memory controller that can be used in an external storage device (hard disk (HD)), flexible disk (FD), or PCMCIA card slot that stores boot programs, various applications, font data, user files, edit files, various data, etc. It controls access to an external memory 211 such as a compact flash (registered trademark) memory connected via an adapter.

Reference numeral 208 denotes a communication I / F controller, which connects and communicates with an external device via a network, and executes communication control processing on the network. For example, communication using TCP / IP is possible.

The CPU 201 can display the outline font on the CRT 210 by executing the outline font expansion (rasterization) process in the display information area in the RAM 203, for example. Further, the CPU 201 enables a user instruction with a mouse cursor or the like (not shown) on the CRT 210.

Various programs described later for realizing the present invention are recorded in the external memory 211, and are executed by the CPU 201 by being loaded into the RAM 203 as needed.

Further, a definition file and various information tables used when executing the above program are also stored in the external memory 211, and detailed description of these will be described later.

Hereinafter, the corpus used in the morphological analyzer 101 according to the present embodiment will be described with reference to the example shown in FIG.

FIG. 3 shows an example of a corpus for learning the morphological analyzer 101. The first column shows a word string that constitutes an example sentence such as "integrate by digital processing.", And the second column shows part of speech information (for example, noun, adjective, verb) for each word.

The initial morphological analysis model can be learned by using the corpus data as shown in FIG. 3, and a new model can be learned by adding a new analysis result.

Hereinafter, the process of registering the result of performing the morphological analysis of the document according to the operation of the user will be described with reference to the flowcharts shown in FIGS. 4, 5, and 6.

The process of reflecting (rendering) the result of automatic analysis of the example sentence on the UI screen by the morphological analyzer 101 will be described with reference to the flowchart shown in FIG.

In step S401, the rendering unit 102 reads the analysis result obtained by automatically performing the morphological analysis of the document from the automatic analysis result DB 107 in a text format, and acquires the information of the divided word string.

In step S402, the rendering unit 102 acquires the part-speech information of the word corresponding to each of the divided words from the automatic analysis result.

In step S403, the rendering unit 102 determines the number of bubbles surrounding the words according to the number of divided words.

In step S404, the rendering unit 102 determines the range surrounded by each bubble according to the length of each word.

In step S405, the rendering unit 102 determines the color of each bubble according to the part of speech information of each word.

In step S406, the parameters determined in each process from step S401 to step S405 are used to render the word surrounded by colored bubbles on the UI screen.

Using the flowchart shown in FIG. 5, a process of receiving a user's operation and recognizing a target to be modified and a method of modifying the word surrounded by bubbles will be described.

In step S501, the operation identification unit 103 determines whether the cursor that the user moves by operating the mouse is hovering over the lower part of the word surrounded by each bubble, and in the state of hovering over. If it is determined that there is, the process proceeds to step S505, and if it is not determined that the mouse is over, the process proceeds to step S502.

In step S502, the operation identification unit 103 determines whether the mouse is over the position where the character with the cursor is displayed, and if it is determined that the mouse is over, the operation identification unit 103 determines that the mouse is over. If it is not determined that the mouse is over, the process proceeds to step S503.

In step S503, the operation identification unit 103 determines whether the cursor is in the "drawing a line" operation, and if it is determined that the cursor is in the "drawing a line" operation, proceeds to step S504 and "draws a line". If it is not determined that the operation is performed, the process proceeds to step S513.

In step S504, the operation reflection unit 104 calculates the line segment drawn by the user from the position where the user presses the mouse button and the position where the mouse button is released, and further, the center of the two characters in each bubble. Calculate each line segment connecting.

By calculating a two-character line segment having an intersection with the line segment drawn by the user, it is determined that the line segment is divided between the two characters.

In step S505, the operation identification unit 103 recognizes that the operation target is the entire word surrounded by bubbles.

In step S506, the operation identification unit 103 recognizes that the operation target is the corresponding character alone.

In step S507, the operation identification unit 103 determines whether the operation target is a mouse drag operation, and if it is determined that the mouse is a drag operation, proceeds to step S508 and determines that the operation is a drag operation. If not, the process proceeds to step S510.

In step S508, the operation identification unit 103 determines whether the intention is to join by approaching another bubble and dragging and dropping, and if it is determined that the intention is to join, the process proceeds to step S509. If it is not determined that the intention is to join, the process proceeds to step S512.

In step S509, the operation identification unit 103 determines that the operation target and the bubble in contact with the operation target are fused.

In step S510, the operation identification unit 103 determines whether it is a "double-click operation", and if it determines that it is a "double-click operation", proceeds to step S511 and determines that it is a "double-click operation". If no determination is made, the process proceeds to step S513.

In step S511, the operation identification unit 103 determines that the operation is to add the part of speech information of the word in the bubble.

In step S512, the operation identification unit 103 determines that the moved target bubble is returned to its original position.

Using the flowchart shown in FIG. 6, a method of recognizing the target to be corrected by the user and the correction method in the process of FIG. 5 and then reflecting the intention on the UI screen will be described.

In step S601, the operation reflection unit 104 determines whether the operation target is an operation to be fused with another bubble, and if it is determined that the operation is to be fused, the process proceeds to step S602 and the fusion is performed. If it is not determined, the process proceeds to step S603.

In step S602, the operation reflection unit 104 determines whether the operation target is a part of the bubble, and if it determines that the operation target is a part, proceeds to step S604, and if it does not determine that it is a part, proceeds to the process. , Proceed to step S605.

In step S603, the operation reflection unit 104 divides the bubbles according to the line text drawn by the user.

In step S604, the operation reflection unit 104 separates the operation target from the original bubble and configures it into a unique bubble.

In step S605, the operation reflection unit 104 fuses with the bubble in contact with the bubble to be operated.

Using the flowchart shown in FIG. 7, a method of saving the automatic analysis result modified according to the user's instruction will be described.

In step S701, the data registration unit 105 determines whether to register the automatic analysis result in the corpus according to the user's instruction, proceeds to step S702 if it determines to register, and steps if it does not determine to register. Proceed to process to S703.

In step S702, the data registration unit 105 registers the corrected automatic analysis result in the corpus 106 as a correct answer.

In step S703, the data registration unit 105 registers the automatic analysis result in the automatic analysis result DB 107 in an unfinished state.

In FIG. 8, one method of correcting the automatic analysis result by operating the mouse is displayed. As an example, suppose that "discussing the right of foreigners to vote" is given as a document to be analyzed by morphological analysis, and "foreign", "carrot", "government", "o", and "discuss" are obtained as a result of automatic analysis. ..

And, as shown in the top row, automatic analysis results should be performed such as "foreign", "people", "suffrage", "o", "discuss", but the wrong analysis result is Since it was obtained, the automatic analysis result is corrected.

First, among the words "foreign", "carrot", "regime", "o", and "discuss" surrounded by bubbles, there is an error in "carrot" and "regime", so the user corrects it by operating the mouse. Do.

As a result, the analysis result can be presented to the user in an easy-to-read display method for the sentence for which the morphological analysis has been performed, and the analysis is performed by enclosing a character string in the range of each word using a bubble expression. It is possible to clarify which word the sentence was divided into.

The user moves the cursor over the center of the character "San" and changes the mouse pointer to the "Palm" icon, which indicates that the character "San" is selected. ing.

Furthermore, by dragging the selected target, the character "Ginseng" is split from the original "Ginseng" bubble to form a single bubble.

As a result, words can be divided by separating bubbles of undivided words with a simple mouse operation.

Then, by contacting the "suffrage" bubble and the "regime" bubble, it merges into the "suffrage" bubble.

For bubbles that are split or newly constructed, it is assumed that there is no part of speech information by default, and it is surrounded by gray bubbles.

Part of speech information can be selected or entered by double-clicking on the gray bubble.

As a result, the background color of the bubble is changed to the color corresponding to various part of speech, and the bubble of the word that has not been specified is made gray, so that the same type of word can be referred to, the part of speech information of the unspecified word, etc. The operation of is convenient.

Furthermore, regarding the new composition and the part-of-speech information of the divided words, by presenting the possible part-speech information options at the moment when the new word is generated, the user's work is supported by forgetting to select the user or selecting by a series of operations. can do.

Next, FIG. 9 displays one of the other methods for correcting the automatic analysis result by operating the mouse. As an example, it is assumed that a morphological analysis is performed on a document similar to the document described in FIG. 8 and a similar automatic analysis result is obtained, and this result is corrected.

First, place the cursor at the bottom of the word "carrot" and mouse over it, the mouse pointer changes to a "palm" icon, and an underscore appears at the bottom of the word bubble to select the word "carrot". Indicates that it is a target.

By dragging the bubble "Ginseng" to be selected and contacting the "Government" bubble, it merges with the "Ginseng Government" bubble.

This makes it possible to correct overdivided words correctly and easily by selecting the whole word surrounded by bubbles and fusing it with other bubbles.

In FIG. 10, one of the other methods for correcting the automatic analysis result by operating the mouse is displayed. As an example, for words decomposed as "foreign", "ginseng government", "o", "discuss", draw a line with the mouse between the letters "person" and the letter "san" in the bubble "ginseng government". When the pulling action is performed, the original bubble is split from the corresponding position, and the bubble is split into a bubble "person" and a bubble "suffrage".

As a result, words can be divided by separating bubbles of undivided words with a simple mouse operation.

It should be noted that the cost information between the words and the word displayed between the adjacent bubbles are different from each other by using the cost information indicating the connection cost between the latest words of the adjacent bubbles and the occurrence cost of each word. When the cost information between words is compared and there is a high possibility that the different words are connected (for example, the cost difference is within a predetermined threshold), information for notifying the user may be displayed.

For example, as shown in FIG. 11, as a result of morphological analysis, "carrot" and "administration" are analyzed and displayed as different words, and the cost information of "carrot" and "suffrage" are displayed. It may be compared with the cost information of "right", and if it may be "suffrage", it may be displayed to notify that fact.

Further, in the embodiment of the present invention, the word is displayed as a bubble, but the display method is not limited to such a display method, such as highlighting a word, displaying a rectangle so as to surround the word, or displaying an underline on the word. The method may be used, and it is sufficient that the word is displayed so that the user can recognize that the word is divided.

Further, in the embodiment of the present invention, the document is the analysis target, but the analysis target is not limited to the document, but a numerical value, an image, a symbol, etc., which are located in some order, and an arbitrary meaning is given based on the analysis result. Any object may be used as long as it can be classified and displayed (set) as shown.

As described above, according to the present invention, an appropriate analysis result can be obtained for the analysis target.

Further, according to the present invention, when statistical learning is performed using the results of morphological analysis, correct answer data as a result of correct analysis is required, but a large amount of correct answer data is required for such learning. You will need it.

Creating (annotating) this correct answer data was often done manually, but it was also a work load to correct the results that were not analyzed well, but the present invention This makes it possible to reduce such a workload.

In addition, the results of sentence analysis are displayed in lattice format, for example, each word divided by the analysis is displayed in one box, and adjacent words are connected by a line in the displayed word box. The numbers and the numbers at the bottom of the box represent the cost of the word and the cost of the connection between the words, respectively, and click the box in a format that displays ambiguous word divisions and part-of-speech in multiple paths. A system that allows you to select the correct path is also conceivable.

However, in such a system, although it is possible to use multiple analysis results as a reference for the user, it is difficult to annotate a word division method that is not included in the analysis results, and the lattice structure is a word division of a sentence. Although it is not intuitive to represent information or the like, according to the present invention, annotation work can be easily performed, and words that are intuitively divided can be easily grasped.

100 Information processing device 101 Morphological analyzer 102 Rendering unit 103 Operation identification unit 104 Operation reflection unit 105 Data registration unit 106 Corpus 107 Automatic analysis result DB

Claims (10)

An analysis means for analyzing analysis targets located in a predetermined order,
A display means for displaying a set showing an arbitrary meaning in the predetermined order based on the analysis result by the analysis means, and a display means.
An operation moving means for moving a part included in the set so as to be included in another set adjacent to the set by an operation on the set displayed by the display means.
An information processing device characterized by being equipped with. The information processing apparatus according to claim 1, wherein the operation moving means moves the set so as to be included in the other set by operating the set. The information processing apparatus according to claim 1 or 2, wherein the operation moving means is moved so as to include the part of the set to another set by operating the part of the set. The information processing apparatus according to any one of claims 1 to 3, wherein the operation moving means is moved so as to include the part of the set in a new set by operating the part of the set. The analysis means analyzes the document as the analysis target, and analyzes the document.
The display means displays the words of the document as the set.
The information processing apparatus according to any one of claims 1 to 4, wherein the operation moving means moves a word included in the word to an adjacent word. The information processing device according to claim 5, wherein the display means separates and displays words in the document. A registration means for registering each set based on the movement by the operation movement means, and
A learning means for learning to perform an analysis by the analysis means using a set registered by the registration means, and a learning means for performing analysis by the analysis means.
The information processing apparatus according to any one of claims 1 to 6, wherein the information processing device is provided. The information processing apparatus according to claim 7, wherein the analysis means analyzes the analysis target based on correct answer data obtained by learning the learning means. It is a control method for information processing equipment.
The information processing device
An analysis step that analyzes the analysis target located in a predetermined order, and
A display step that displays a set showing an arbitrary meaning in the predetermined order based on the analysis result of the analysis step, and a display step.
An operation moving step of moving a part included in the set so as to be included in another set adjacent to the set by an operation on the set displayed by the display step.
An information processing device characterized by executing. Computer,
An analysis means for analyzing analysis targets located in a predetermined order,
A display means for displaying a set showing an arbitrary meaning in the predetermined order based on the analysis result by the analysis means, and a display means.
An operation moving means for moving a part included in the set so as to be included in another set adjacent to the set by an operation on the set displayed by the display means.
A program to make it work.

Images