JP5171087B2 - Input information analyzer - Google Patents

Description

  The present invention relates to an input information analysis apparatus that can easily grasp the correlation between input information and classification.

Sentences (input information) such as customer opinions and questionnaires sent by e-mail and postcards contain useful information such as customer opinions and market needs. For this reason, by acquiring useful information included in such text, it is possible to respond quickly and accurately to customer opinions, market needs, and the like. Here, in the case of a sentence (standard input information) that selects the corresponding one from the choices, it is easy to analyze the sentence, but a free description (atypical) opinion or questionnaire etc. In the case of (atypical input information), it is difficult to analyze the sentence. For this reason, development of analysis technology (input information analysis technology) for obtaining useful information from an enormous amount of atypical sentences (input information) is being promoted.
Conventionally, an input information analysis technique (text mining) as described in Patent Document 1 is known. The input information analysis technique described in Patent Document 1 divides input information (sentence) to extract words constituting the input information, and keywords corresponding to each word and each of a plurality of categories (classifications) in the input information. , Count the number of matches as the frequency by category, calculate the correlation coefficient of the frequency by category, and summarize the large amount of text appropriately by calculating the factor and factor loading from the correlation It is.
JP 2004-86350 A

Here, in order to quickly and accurately obtain useful information such as customer opinions and market needs contained in the input information entered as sentences, the results of analyzing the input information can be easily grasped visually. There is a demand for the development of input information analysis technology that can do this.
As a method of outputting the result of classifying input information by category (classification) so that it can be visually confirmed, a method of outputting by a list for each category or statistical data is used. However, such an output method cannot be used as an output method capable of visually confirming the result of classification into a plurality of categories (classifications). Further, the input information analysis technique described in Patent Document 1 uses a technique that compares word information constituting input information with keywords corresponding to each of a plurality of categories (classifications) and counts them as a frequency for each category. The output method such as frequency by category is not considered.
The present invention was devised in view of the above points, and is an input information analysis technique capable of easily and visually outputting the correlation between input information input as sentences and a plurality of classifications. The purpose is to provide.

1st invention of this application is an input information analysis apparatus provided with the structure of Claim 1. FIG.
The present invention includes input means, storage means, management means, word information extraction means, word importance parameter determination means, similarity determination means, coordinate information calculation means, and output means. Various input means and output means can be used as the input means and output means. The present invention is suitable when the display means is used as the output means.
The storage unit stores classification word information corresponding to each of a plurality of classifications. The classification (category) is appropriately selected according to the field / use to which the present invention is applied. Further, as the classification word information, various word information representing the classification is used. As the classification word information, a noun is preferably used, but it is not limited to a noun.
The word information extraction unit divides a plurality of pieces of input information input from the input unit, and extracts word information constituting each of the plurality of pieces of input information. As a method for extracting word information constituting input information, typically, a morpheme analysis technique for dividing input information into morphemes (words) is used. Various known techniques can be used as the morphological analysis technique. As the input information for extracting the word information, the input information input from the input unit and stored in the storage unit can be used. The input information stored in the storage means is included in the “input information input from the input means” of the present invention.
In addition, “a plurality of classifications stored in the storage unit” means all the classifications stored in the storage unit or the designated classification among the classifications stored in the storage unit. The instructed classification includes a predetermined classification or a classification input via an input device whenever necessary. In other words, it may be at least a classification necessary for analyzing input information.
The word importance parameter discriminating means discriminates the word importance parameter of each classified word information corresponding to each of a plurality of classifications for each of a plurality of input information. The word importance parameter of each classification word information corresponding to each of a plurality of classifications with respect to a plurality of input information is respectively included in the word information constituting each of the plurality of input information and each of the plurality of classifications stored in the storage unit. The determination is made based on the corresponding classified word information. Classification words corresponding to a plurality of classifications for a plurality of input information based on word information constituting each of the plurality of input information and classification word information corresponding to each of the plurality of classifications stored in the storage means Various methods can be used as a method of discriminating the word importance parameter of each piece of information.
The similarity determination unit determines the similarity between the plurality of input information and each of the plurality of classifications based on the word importance parameters of the plurality of input information and the respective classification word information corresponding to the plurality of classifications. An appropriate method can be used as a method of discriminating the degree of similarity between a plurality of input information and a plurality of classifications.
The coordinate information calculation means arranges a classification symbol indicating each of a plurality of classifications along the outer circumference of a circle or an ellipse on a two-dimensional coordinate plane, and a classification having a high degree of similarity between each of the plurality of input information Classification symbol coordinate information indicating the arrangement position of each classification symbol on the two-dimensional coordinate plane is calculated. In addition, the input information symbol indicating each of the plurality of input information corresponds to the similarity between the input information indicated by each input information symbol and the classification indicated by each classification symbol in the arrangement relationship between each input information symbol and each classification symbol. The input information symbol coordinate information indicating the arrangement position of each input information symbol on the two-dimensional coordinate plane is calculated. In a two-dimensional coordinate plane having a plurality of classifications as coordinate axes, coordinate axes corresponding to the plurality of classifications extend from the origin. The position along the outer periphery is appropriately selected. For example, it is set to be arranged at equal intervals along the outer periphery. The “position along the outer circumference of the circle or ellipse” includes the position along the outer circumference of the circle or ellipse. Various known methods can be used as a method for determining the correlation of the similarity pattern. The classification symbol coordinate information is arranged at a position where classification symbols indicating a classification having a high degree of similarity between each input information, that is, a classification having a similar similarity pattern, are close to each other along a circle or an ellipse. Is calculated as follows.
The management means outputs the classification symbol and the input information symbol from the output means based on the classification symbol coordinate information and the input information symbol coordinate information.
In the present invention, for the plurality of input information, the similarity between the plurality of input information and the plurality of classifications is determined based on the word importance parameter of each of the classification word information corresponding to each of the plurality of classifications, and the plurality of classifications are coordinate axes. The classification information indicating a plurality of classifications is arranged on the two-dimensional coordinate plane and the input information so that the arrangement relationship between the classification symbols and the input information symbols corresponds to the similarity between the input information and the classification. The symbol is arranged. This makes it easy to visually grasp the correlation between input information and classification, the correlation between input information, the tendency of the entire input information classified into multiple classifications, etc. from the positional relationship between the input information symbols and the classification symbols. Can do. In particular, in the present invention, each classification symbol is arranged on a two-dimensional coordinate plane along a circle or an ellipse in which the distance from the origin to each classification symbol does not change greatly. It can be easily grasped. In addition, since a classification symbol indicating a classification whose similarity pattern with the input information is similar is arranged at a position close to the circle or ellipse on the two-dimensional coordinate plane, the correlation between the input information and the classification is It can be grasped more easily.

According to a second aspect of the present invention, there is provided an input information analyzing apparatus having the configuration according to the second aspect.
In the present invention, the word importance parameter determination means determines the word importance parameter of each classification word information corresponding to each of a plurality of classifications for each of a plurality of input information by the following method.
First, the classification word information is obtained by comparing the word information constituting each of the plurality of input information extracted by the word information extraction means with the classification word information corresponding to each of the plurality of classifications stored in the storage device. The number of occurrences of word information indicating the number of input word information in the input information is determined for each of the classification word information corresponding to each of the plurality of classifications for each of the plurality of input information. In addition, the number of times of appearance of input information indicating the number of input information including word information corresponding to the classification word information is determined for each of the classification word information corresponding to each of the plurality of classifications. In this case, the word information appearance count is determined for each of the plurality of input information, for each of the classification word information of each of the plurality of classifications, the input information appearance count is common to the plurality of classifications, each of the plurality of classifications Are determined for each classification word information.
For each of the plurality of input information, the number of occurrences of word information determined for each of the classification word information corresponding to each of the plurality of classifications and the number of input information appearances determined for each of the classification word information corresponding to each of the plurality of classifications Based on each of the plurality of input information, the word importance parameter of each of the classification word information corresponding to each of the plurality of classifications is determined. For example, the word importance parameter between arbitrary input information and arbitrary classification word information corresponding to an arbitrary classification includes the word information appearance count of the classification word information corresponding to the classification for the input information and the input information. The classification is performed using the number of appearances of input information of the classification word information corresponding to the classification. Of course, the method of determining the word importance parameter based on the word information appearance count and the input information appearance count is not limited to this.
In the present invention, the word importance parameter is determined using the word information appearance count and the input information appearance count regarding the classification word information corresponding to the classification. Thereby, it is possible to easily determine the word importance parameter with high accuracy.

According to a third aspect of the present invention, there is provided an input information analyzing apparatus having the configuration according to the third aspect .
In the present invention, when the input information symbol output request signal indicating the classification symbol is input in a state where the classification symbol and the input information symbol are being output, the management means is instructed by the input information symbol output request signal. An input information symbol indicating input information having a similarity to the classification indicated by the classification symbol is output in a different output mode from other input information symbols.
The input information symbol output request signal designating the classification symbol is input by, for example, placing the mouse pointer at a position corresponding to the classification symbol or clicking the mouse in the arranged state. As a method of outputting in different output modes, for example, a method of outputting in different colors, a method of changing the shape of the input information symbol, or the like can be used. Moreover, it can also be comprised so that the input information symbol output request signal which supports a some classification | category symbol can be input. In this case, input information symbols related to the designated classification symbol (input information symbols indicating input information having similarity to the classification indicated by the classification symbol) can be output in a different manner for each classification symbol. It can also be configured.
In the present invention, when an input information symbol output signal indicating a classification symbol is input, the input information symbol of the input information related to the classification indicated by the indicated classification symbol is displayed, so that an input correlated with a specific classification Information can be easily grasped.

According to a fourth aspect of the present invention, there is provided the input information analyzing apparatus according to the fourth aspect .
In the present invention, when an input information output request signal indicating a classification symbol is input in a state in which the classification symbol and the input information symbol are output, the management means outputs the classification symbol indicated by the input information output request signal The input information having the similarity with the classification indicated by is output.
An appropriate method can be used as a method of outputting input information having a similarity to the classification. For example, it is possible to use a method in which a main display unit and a sub display unit are provided in the display unit of the display means, a classification symbol and an input information symbol are displayed on the main display unit, and input information is displayed on the sub display unit.
When outputting the input information having the similarity with the classification indicated by the classification symbol, the input information symbol indicating the input information may be output in a manner different from the other input information symbols.
In the present invention, when an input information output request signal indicating a classification symbol is input, input information related to the classification indicated by the specified classification symbol is output. Therefore, input information correlated with a specific classification can be easily obtained. Can be confirmed.

According to a fifth aspect of the present invention, there is provided an input information analyzing apparatus having the configuration according to the fifth aspect .
In the present invention, when the input information output request signal indicating the input information symbol is input in a state where the classification symbol and the input information symbol are output, the management means inputs the input indicated by the input information output request signal Output the input information corresponding to the information symbol.
As a method for outputting the input information, an appropriate method can be used. For example, when a classification symbol is instructed while a classification symbol and an input information symbol are displayed on the display section of the display means, a display window is displayed on the display section in the vicinity of the classification symbol, and a window section is formed to form a main display. A sub-display unit, a classification symbol and an input information symbol are displayed on the main display unit, and input information is displayed on the sub-display unit.
In the present invention, when an input information output request signal indicating an input information symbol is input, the input information indicated by the instructed input information symbol is output, so that the input information can be easily confirmed.

According to a sixth aspect of the present invention, there is provided an input information analyzing apparatus having the configuration according to the sixth aspect .
In the present invention, when the classification symbol and the position of the classification symbol are instructed in a state where the classification symbol and the input information symbol are output, the coordinate information calculation means determines the coordinate information of the instructed position. Then, when the determined coordinate information is the classification symbol coordinate information of the designated classification symbol, the input information symbol coordinate information of the input information symbol indicating the input information having the classification and the similarity indicated by the designated classification symbol Is calculated. In this case, as described above, the input information symbol coordinate information is such that the arrangement relationship between the designated classification symbol and the input information symbol is similar to the classification indicated by the designated classification and the input information indicated by the input information symbol. It is calculated to correspond to the degree.
The classification symbol and the position indicated with respect to the classification symbol are detected by, for example, dragging the classification symbol with the mouse pointer to detect the classification symbol indication signal indicating the classification symbol at the location corresponding to the mouse pointer and the position of the mouse pointer. It can be determined by inputting a signal from the input device.
In the present invention, when the classification symbol and the position of the classification symbol are instructed in a state where the classification symbol and the input information symbol are output, the designated classification symbol is linked with the movement of the designated classification symbol. The input information symbol coordinate information of the input information symbol indicating the input information having the classification and similarity shown is calculated, and the input information symbol is moved. Thereby, the correlation between classification and input information can be grasped in more detail.

A seventh aspect of the present invention is an input information analyzing apparatus having the configuration according to the seventh aspect .
In the present invention, the management means outputs the classification symbol and the input information symbol in different output modes.
As a method of outputting the classification symbol and the input information symbol in different output modes, for example, a method of outputting in different colors and different shapes can be used.
In the present invention, since the classification symbol and the input information symbol are output in different output modes, the classification and the input information can be easily distinguished, and the correlation between the classification and the input information can be easily grasped.

An eighth aspect of the present invention is the storage medium according to the eighth aspect .
The present invention is a program for causing a computer to execute processing of any of the management means, word information extraction means, word importance parameter determination means, similarity determination means, and coordinate information calculation means of any of the first to seventh inventions. It is.
By using the program of the present invention, the same effects as those of the first to ninth inventions can be obtained.

A ninth aspect of the present invention is the storage medium according to the ninth aspect .
The present invention is a program for causing a computer to execute processing of any of the management means, word information extraction means, word importance parameter determination means, similarity determination means, and coordinate information calculation means of any of the first to seventh inventions. Is a storage medium on which is recorded.
By using the storage medium of the present invention, the same effects as those of the first to ninth inventions can be obtained.

By using the input information analysis apparatus according to any one of claims 1 to 7 , the program according to claim 8 , and the storage medium according to claim 9 , the correlation between the input information and the plurality of classifications can be easily made visually. I can grasp it.

Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 shows a schematic block diagram of an embodiment of an input information analyzing apparatus of the present invention. In the present embodiment, word information that constitutes input information by inputting, as input information, an atypical sentence such as a free description opinion or a questionnaire transmitted (submitted) from a sender such as a customer It is comprised as an input information analysis apparatus which analyzes the classification | category (category) represented by these.

The present embodiment includes a processing device 10, a storage device 20, an input device 30, a display device 40, and the like.
As the storage device 20, various storage devices such as a ROM and a RAM can be used. The storage device 20 includes a classification word information database 20a, an input information database 20b, and the like. The storage device 20 can be constituted by one storage device having the classified word information database 20a and the input information database 20b, or can be constituted by individual storage devices. Further, the arrangement position of the storage device 20 can be set as appropriate as long as the processing device 10 (each unit of the processing device 10) can access.
In the classified word information database 20a, classified word information representing a classification (category) is stored corresponding to the classification. An example of the classification single information word database 20a is shown in FIG. The classification word database 20a shown in FIG. 2 includes classification identification information (classification ID) indicating classification, classification name, classification word information corresponding to the classification, classification word information identification information (classification word information ID), weight, and the like. Is remembered. For example, corresponding to the classification ID [C1], classification word information representing the classification name [stock] and classification [stock], classification word information [stock] of the classification word information ID [T1], classification word information ID [T2] ] Classified word information [stock price] and classified word information [shareholder] of classified word information ID [T3] are stored. Further, a weight is stored corresponding to each classification word information. The weight is used when determining each component of the classification vector. For example, [stock], [stock price], and [shareholder] are used as the classification word information corresponding to the classification ID [C1] [stock], and the weights of the classification word information are [0.8] and [1.0], respectively. , [1.0] is set, the classification vector [C1 (stock, stock price, shareholder)] of the classification ID [C1] [stock] is (0.8, 1.0, 1.0). expressed. It is used as appropriate when discriminating word importance parameters and the like. In FIG. 2, classification word information such as [E-mail], [ADSL], [Homepage], classification name [all electrification] (classification ID [C3], corresponding to classification name [Internet] (classification ID [C2]). ]) Corresponding to classification word information such as [cooking heater], [floor heating], [utilization cost], and classification name [new energy] (classification ID [C4]) [photovoltaic power generation], [wind power generation] Classification word information such as “power generation” and “fuel cell” is stored.
In the classification word information database 20a shown in FIG. 2, noun information is used as the classification word information, but word information of part of speech other than the noun information can be used as the classification word information. Moreover, since the same word information may represent different classifications (categories), the same classified word information may be stored corresponding to different classifications.
Information stored in the classified word information database 20a can be set as appropriate. Moreover, the aspect which memorize | stores classification word information in the memory | storage device 20 can be changed suitably.
The input information database 20b stores input information and the similarity between the input information and each classification. An example of the input information database 20b is shown in FIG. In the input information database 20b shown in FIG. 3, the input information such as text transmitted from the sender, and the similarity between the input information and each classification are stored corresponding to the input information identification information (input information ID). Has been. A method for determining the similarity between the display information and each classification will be described later.
Information stored in the input information database 20b can be set as appropriate. For example, when the similarity between the input information and each classification is determined each time, the similarity between the input information and each classification can be omitted. Moreover, the aspect which memorize | stores input information etc. in the memory | storage device 20 can be changed suitably.
The storage device 20 corresponds to the “storage means” of the present invention.

As the input device 30, a keyboard capable of inputting input information and various instruction signals, a touch panel provided in the display unit of the display device 40, a mouse, an information reading device for reading information stored in a storage medium, and the like Various input devices can be used.
The input device 30 corresponds to the “input unit” of the present invention.
As the display device 40, various display devices such as a liquid crystal display device capable of displaying analysis results can be used. Note that, when configured to be communicable with a remote terminal device, the input device and output device of the remote terminal device correspond to the input device and output device of the present invention.
The display device 40 corresponds to the “output unit” of the present invention.

The processing apparatus 10 includes a management unit 10a, a word information extraction unit 10b, a word importance parameter determination unit 10c, a similarity determination unit 10d, and a coordinate information calculation unit 10e. The processing apparatus 10 can be configured by a single CPU that executes the processes of the units 10a to 10e, or can be configured by an individual CPU that executes the processes of a plurality of processing units.
The management means 10a corresponds to the “management means” of the present invention, the word information extraction means 10b corresponds to the “word information extraction means” of the present invention, and the word importance parameter determination means 10c corresponds to the “word importance parameter” of the present invention. Corresponding to the “discriminating means”, the similarity classification means 10d corresponds to the “similarity determining means” of the present invention, and the coordinate information calculating means 10e corresponds to the “coordinate information calculating means” of the present invention.

  The management unit 10a manages the processing of the entire input information analysis apparatus. For example, it manages input processing of input information and instruction signals from the input device 30, output processing of analysis results and the like to the display device 40, execution of processing by each means 10b to 10e, and the like. Note that the processing of the management unit 10a may be executed by the word information extraction unit 10b, the word importance parameter determination unit 10c, the similarity determination unit 10d, and the coordinate information calculation unit 10e. The output mode of the analysis result by the management unit 10a will be described later.

The word information extraction means 10b divides the input information input from the input device 30 or the input information input from the input device 30 and stored in the input information database 20b of the storage device 20 to constitute input information. Extract word information. As a technique for extracting word information constituting input information, for example, a morpheme analysis technique for decomposing input information into morphemes is used. Various methods are known as morphological analysis methods. For example, the longest match method, the minimum number of division method, and the like are known. A dictionary used when decomposing input information into morphemes is stored in the storage device 20.
The input information stored in the input information database 20b is input from the input device 30 and stored. Therefore, the input information stored in the input information database 20b is included in the input information input from the input device 30.

The word importance parameter discriminating means 10c includes the word information constituting the input information extracted by the word information extracting means 10b and the classified words stored in the classified word information database 20a provided in the storage device 20. Based on the information, the word importance parameter of each classification word information corresponding to each classification for each input information is determined.
As a method for discriminating the word importance parameter of each classification word information corresponding to each classification for each input information, for example, the word information appearance frequency TF indicating the number in the input information of the classification word information and the classification word information. The input information appearance frequency DF indicating the number of input information included is determined, and the input information and the classification are corresponded based on the word information appearance frequency TF and the input information appearance frequency DF for each classification word information corresponding to the classification for the input information. A method of calculating the word importance parameter W with the classified word information is used.
In the input information Di, the number (number of appearances of word information) of the classification word information Fj corresponding to the classification Cn in the input information Di (number of appearances of word information) TFij and the number of input information (number of occurrences of input information) including the classification word information Fj corresponding to the classification Cn In order to determine the number (number of appearances of sentence information) DFij, the word information constituting the input information Di is compared with the classified word information Fj, and the number of the word information corresponding to the classified word information Fj in the input information is determined. Determine. Whether or not the word information corresponds to the classified word information Fj is determined by considering the grammatical variation of at least one of the classified word information and the word information, and whether or not the classified word information and the word information match. To do. Then, the number of word information corresponding to the classification word information Fj in the word information constituting the input information Cn is set as the number (word information appearance frequency) TFij in the input information Ci of the classification word information Fj. By performing this process for each of the classification word information Fj corresponding to the classification Cn, the word information appearance frequency TFij for each classification word information Fj corresponding to the classification Cn with respect to the input information Ci is determined.
The number of input information Dj including the classification word information Fj corresponding to the classification Cn, that is, the number of input information whose word information appearance frequency DFij for the classification word information Fj is 1 or more (DFij ≧ 1) is included in the classification Cn. This is the number of input information Di (number of appearances of input information) DFij including the corresponding classified word information Fj.
As described above, the word information appearance count TFij and the input information limit count DFij for each classification word information Fj corresponding to the classification Cn with respect to the input information Ci are determined.
Next, based on the word information appearance count TFij and the input information limit count DFij of the classification word information Fj corresponding to the classification Cn for the input information Di, the word of the classification word information Fj corresponding to the classification Cn for the input information Di The importance parameter Wij is determined. The word importance parameter Wij can be determined using, for example, the following expression.
Wij = TFij × IDFij = TFij × log (N / DFij)
Here, N is the total number of input information.
This processing is performed on the input information Di for each of the classification word information Fi corresponding to the classification Cn, thereby determining the word importance parameter Wij between the input information Ci and each of the classification word information Fj corresponding to the classification Cn. To do.
The word importance parameter Wij represents the weight (importance) of the classified word information Fj in the input information Di.

The similarity determination unit 10d determines the similarity between the input information and the classification based on the word importance parameter between the input information and each of the classification word information corresponding to the classification.
As a method for determining the similarity between the input information and the classification, for example, the following method is used.
The classification Cn is represented as a classification vector << Cn >> using classification word information Fj corresponding to the classification Cn. The symbol <<>> represents a vector.
<< Cn >> = (F1, F2,...)
For example, the classification vector << Cn >> (stock) in the case of the classification name [stock] shown in FIG. 2 is expressed as follows.
<< Cn >> (stock) = (F1 [stock], F2 [stock price], F3 [shareholder])
Each component of the classification vector << Cn >> (stock), F1 [stock], F2 [stock price], F3 [shareholder] values are set corresponding to the classification word information [stock], [stock price], [shareholder] It is determined based on the weight.
The input information Di is represented as an input information vector << Di >> using the word importance parameter Wij of the input information Di and each of the classification word information Fj corresponding to the classification Cn. For example, the input information vector << Di >> for the classification [stock] of the input information Di is the word importance of the input word Di and the classification word information [stock], [stock price], and [shareholder] corresponding to the classification [stock]. It is expressed as follows using the parameters Wi1 [stock], Wi2 [stock price], and Wi3 [shareholder].
<< Di >> = (Wi1 [stock], Wi2 [stock price], Wi3 [shareholder])
The classification vector << Cn >> and the input information vector << Di >> are expressed in, for example, a multidimensional space based on F1, F2, and F3 shown in FIG.
In the present embodiment, the similarity SIMin of the input information Di with respect to the classification Cn is expressed using the classification vector << Cn >> and the word importance parameters of the classification information corresponding to the input information Di and the classification Cn. Using the angle θ between the input information vector << Di >>, it is determined by the following equation.
Similarity SIMin = cos θ = << Cn >> · << Di >> / (| << Cn >> | × | << Di >> |)
The operation symbol • indicates an inner product of vectors, and the operation symbol || indicates the length (norm) of the vector.
Here, since the input information vector of the input information Di for the classification Cn is represented by using the similarity between the input information Di and the classification word information corresponding to the classification Cn, the classification vector << Cn >> is normalized. Can be treated as a vector.

The coordinate information calculation unit 10e calculates classification symbol coordinate information and input information symbol coordinate information for arranging a classification symbol indicating classification and an input information symbol indicating input information on a two-dimensional coordinate plane.
Here, as described above, the classification and input information are represented as a classification vector and an input information vector having components corresponding to the number of classification word information corresponding to the classification. When classification vectors and input information vectors having different numbers (dimensions) of these components are projected as they are onto the two-dimensional coordinate plane, the classification symbols and input information symbols projected onto the two-dimensional coordinate plane are used for classification. And the correlation between the input information cannot be grasped visually.
In the present embodiment, the similarity determination unit 10d classifies the classification symbol indicating classification and the input information symbol indicating input information based on the similarity between the input information and the classification determined by the similarity determination unit 10d. Classification symbol coordinate information and input information symbol coordinate information for placement on a two-dimensional coordinate plane with the coordinate axis as the coordinate axis are calculated. Thereby, by taking into consideration the arrangement position of the classification symbol indicating the classification, the correlation between the input information and the classification value can be easily grasped visually from the classification symbol and the input information symbol projected onto the two-dimensional coordinates. .
In addition, although classification symbols indicating all classifications can be arranged on the two-dimensional coordinate plane, there is a case where it is not necessary to grasp the correlation between all classifications and input information. For this reason, in the present embodiment, the coordinate information calculation unit 10e arranges the classification symbol indicating the instructed classification and the input information symbol indicating the input information on a two-dimensional coordinate plane having the instructed classification as a coordinate axis. Classification symbol coordinate information and input information symbol coordinate information are calculated.
The “instructed classification” includes a predetermined classification and a classification designated by the input device 30. When a predetermined classification is used as the “instructed classification”, when an analysis result is output, for example, when an analysis result output instruction signal for instructing the output of the analysis result is input from the input device 30 The classification symbol and the input information symbol are output based on the similarity between the predetermined classification and the input information. When the classification designated by the input device 30 is used as the “instructed classification”, a classification instruction signal for instructing classification is input from the input device 30 when outputting the analysis result.

In consideration of the above points, in the present embodiment, the coordinate information calculation unit is configured so that the classification symbol indicating the designated classification is arranged along the outer periphery of the circle on the two-dimensional coordinate plane. Information is calculated. That is, a coordinate plane is formed in which the center of the circle is the origin and the coordinate axis corresponding to the designated classification extends from the origin. At this time, the arrangement position of the designated classification symbol along the outer circumference of the circle can be selected as appropriate. For example, it can be arranged evenly along the outer periphery of the circle, or it can be arranged hard. In the present embodiment, the classification symbol coordinate information is calculated so that the classification symbols are evenly arranged along the circumference of the circle.
The classification symbols can be arranged along the outer periphery of the ellipse. In addition, a mode in which the classification symbol is arranged along the outer periphery of the circle or ellipse includes a mode in which the classification symbol is roughly arranged along the outer periphery of the circle or ellipse.
Then, the coordinate information calculation unit 10e is an input information symbol in which the relationship between the input information symbol and the classification symbol on the two-dimensional coordinate plane is determined by the similarity determination unit 10d based on the calculated coordinate information of the classification symbol. Input information symbol coordinate information for arranging the input information symbol on the two-dimensional coordinate plane is calculated so as to correspond to the similarity between the input information shown and the classification indicated by the classification symbol. As a relationship between the input information symbol and the classification symbol on the two-dimensional coordinate plane, for example, a value (distance from the origin along the coordinate axis) corresponding to each coordinate axis of the input information symbol on the two-dimensional coordinate plane is used. Can do.
In addition, when the coordinate axes of classification A and classification B having similarity to the input information are arranged at opposite positions across the origin, the input information symbol position on the two-dimensional coordinate plane And the correlation between classification A and classification B cannot be accurately grasped. In such a case, it is necessary to move the classification coordinate axis (the arrangement position of the classification symbol along the circumference of the circle) to an appropriate position.
In addition, since the display area of the display device 40 on which the classification symbols and the input information symbols are arranged is limited, it is preferable to arrange the input information symbols effectively. Therefore, in the present embodiment, the coordinate information calculation unit 10e is configured to set the similarity between the display information and the classification to [0] when the similarity between the display information and the classification is less than the threshold. Has been. The threshold is appropriately selected according to display information and classification. For example, when the similarity between the input information D1 and the classifications C1, C2, and C3 is [0.1], [0.2], and [0.1], and the threshold is set to [0.3] All of the similarities of the input information D1 with the classifications C1, C2, and C3 are less than [0.3]. In this case, the input information symbol indicating the input information D1 is not displayed on the two-dimensional coordinate plane having the classifications C1, C2, and C3 as coordinate axes.
The management means 10a is arranged so that the classification symbol and the input information symbol are arranged at the arrangement position on the two-dimensional coordinate plane indicated by the classification symbol coordinate information and the input information coordinate information calculated by the coordinate information calculation means 10e. The input information symbol is displayed on the display device 40.
At this time, it is preferable to output the classification symbol and the input information symbol in different output modes. The different output modes include, for example, a mode of outputting using different colors, a mode of outputting using different characters and symbols, and the like.

Next, the operation of the present embodiment will be described with reference to the flowcharts shown in FIGS. Note that the flowchart of FIG. 4 shows a similarity determination process for determining the similarity between input information and classification. The flowchart of FIG. 5 shows an input information classification diagram output process for outputting an input information classification diagram representing the similarity between input information and classification.
The process illustrated in FIG. 4 is started, for example, when a similarity determination process start signal instructing the start of the similarity determination process is input from the input device 30.

In step A1, word information is extracted from each of N pieces of input information Di (i = 1... N) input from the input device 30. The word information constituting the input information Di is extracted by the method described above by the word information extracting means 10b. The input information input from the input device 30 includes input information input from the input device 30 and stored in the input information database 20b.
In step A2, for each input information Di, each classification word information Fj (j = 1... The number of classification word information corresponding to the classification Cn) corresponding to each classification Cn (n = 1... The number of classifications). The word importance parameter Wij is determined. The word importance parameter Wij is determined by the word importance parameter determination means 10c by the method described above.
For example, the word information constituting the input information Di extracted in step A1 and the jth classified word information corresponding to the nth classification Cn stored in the classified word information database 20a of the storage device 20 Compare Fj. Then, the number (word information appearance count) TFij in the input information Di of the word information corresponding to the classified word information Fj is determined. Further, the number (input information appearance frequency) DFij in the total input information of the input information Di including the word information corresponding to the classified word information Fj is determined.
Next, based on the word information appearance frequency TFij and the input information appearance frequency DFij, the word importance parameter Wij of each classification word information Fj corresponding to the classification Cn with respect to the input information Di is determined. The word importance parameter Wij is calculated by the following equation, for example.
Wij = TFij × IDFij = TFij × log (N / DFij)
In step A3, the similarity SIMin between the input information Di and the classification Cn is determined based on the word importance parameter Wij and the classification word information corresponding to the classification Cn stored in the classification word information database 20b. The similarity SIMin is determined by the similarity determination means 10d by the method described above.
For example, based on the classification vector << Cn >> and the input information vector << Di >>, the following formula is used.
SIMin = cos θ = << Cn >> · << Di >> / (| << Cn >> | × | << Di >> |)
The input information Di and the similarity SIMin between the input information Di and the classification Cn are stored in the input information database 20b of the storage device 20.

Next, an input information classification diagram output process for outputting an input information classification diagram representing the similarity between input information and classification will be described with reference to the flowchart of FIG. The flowchart of FIG. 5 shows an operation in the case of outputting an input information classification diagram in which an input information symbol indicating input information and a classification symbol indicating classification are arranged on a two-dimensional coordinate plane having the classification as a coordinate axis.
The process of FIG. 5 is started at an appropriate time after the similarity determination process shown in FIG. 4, for example, or an input information classification diagram output process start signal instructing the start of the output process of the input information classification diagram Is input from the input device 30.

In step B1, it is determined whether or not a classification for displaying the input information classification diagram has been instructed. As a classification instruction method, an appropriate method can be used. For example, in the case of using a pre-instructed (pre-set) classification, it is confirmed that the classification is instructed by the input of the input information classification diagram display instruction signal for instructing the display of the input information classification diagram. to decide. Further, when the method of instructing classification using the input device 30 is used when displaying the input information classification diagram, the classification instruction is completed after the classification instruction signal instructing the classification is input from the input device. When a classification instruction completion signal (or input information classification diagram display instruction signal) indicating this is input, it is determined that the classification is instructed. When the classification is instructed, the process proceeds to Step B2, and when the classification is not instructed, the process proceeds to Step B5.
The process of step B5 is executed by the management means 10a.
In step B2, classification symbol coordinate information for arranging the classification symbol indicating the classification designated in step B1 on a two-dimensional coordinate plane having the designated classification as a coordinate axis is calculated. In the present embodiment, the classification symbols are evenly arranged along the outer periphery of a circle centered on the origin on the two-dimensional coordinate plane. Therefore, in step B2, the classification symbol coordinate information of each classification symbol is calculated so that the classification symbols are evenly arranged along the circumference of the circle. At this time, the reference position on the circumference for arranging the classification symbol, the classification arranged at the reference position, and the like are appropriately set. For example, in a state where a circle is displayed on the display unit of the display device 40, an upper position (12 o'clock position) is a reference point along the outer periphery, and a classification symbol indicating the classification is clockwise from the reference point. It is set to be arranged in the order of the classification ID.
In step B3, the classification symbol coordinate information indicating the arrangement position of the classification symbol on the two-dimensional coordinate plane calculated in step B2 and the similarity between the input information and the classification specified in step B1 (step A3 in FIG. 4). The input information symbol coordinate information indicating the arrangement position of the input information symbol on the two-dimensional coordinate plane is calculated. At this time, the input information symbol coordinate information is calculated so that the relationship between the input information symbol and the classification symbol in the two-dimensional coordinate plane corresponds to the similarity between the input information indicated by the input information symbol and the classification indicated by the classification symbol. To do. Note that the calculation of the input information symbol coordinate information can be omitted for input information whose similarity between the input information and each classification is equal to or less than a threshold value.
Steps B2 and B3 are executed by the coordinate information calculation means 10e.
In step B4, based on the classification symbol coordinate information calculated in step B2 and the input information symbol coordinate information calculated in step B3, the classification symbol and the input information symbol are displayed on the two-dimensional coordinate plane of the display unit of the display device 40. .
The process of step B4 is executed by the management unit 10a.

Here, in this embodiment, as described above, the classification coordinate information is calculated so that the classification symbols are arranged in the order of the classification IDs indicated by the classification symbols along the outer circumference of the circle on the two-dimensional coordinate plane. doing. However, when the classification coordinate information is calculated using this method, the classification in which the state of similarity of display information with respect to the classification (similarity pattern with respect to display information) is similar depending on the input information and the designated classification. May be arranged at positions separated along the outer circumference of the circle (for example, positions facing each other across the origin). In this case, there is a possibility that the correlation between the input information and the classification cannot be accurately displayed on the two-dimensional coordinate plane. For this reason, it is preferable to arrange | position the classification | category with which the similarity pattern with respect to the display information of a classification | category is close in the position.
For example, as shown in FIG. 8, in the order of the classification IDs, (classification ID [Ca] classification) → (classification ID [Cb] classification) → (classification ID [Cd] Classification) → (Classification of classification ID [Ce]). Further, the correlation between the classifications Ca to Ce and the classification state (classification pattern) of the input information Da to De is determined. As a method for determining this correlation, for example, correspondence analysis, which is one method of multivariate analysis, can be used. Considering the correlation between each classification and the classification pattern of the input information by correspondence analysis, the arrangement order of the classification symbols (coordinate axes corresponding to the classification) is (classification of classification ID [Ca]) → Sorted in the order of (classification of classification ID [Cd]) → (classification of classification ID [Cc]) → (classification of classification ID [Ce]). As a method for determining the correlation between the classification and the classification pattern in the input information, various known methods other than the correspondence analysis can be used.
Then, as shown in FIG. 9, on the two-dimensional coordinate plane having the classification IDs [Ca] to [Ce] as the coordinate axes (a axis to d axis), they are rearranged clockwise along the outer circumference of the circle. (Classification symbol of classification ID [Ca]) → (Classification symbol of classification ID [Cd]) → (Classification symbol of classification ID [Cc]) → (Classification symbol of classification ID [Cb]) in equal order Placed in. As shown in FIG. 10, the input information symbols of the input information D1 to D4 are input to the input information D1 to D4 on the two-dimensional coordinate plane having the classification of the classification IDs [Ca] to [Ce] as coordinate axes. The relationship between the information symbols and the classification symbols of the classification IDs [Ca] to [Cd] is arranged so as to correspond to the similarity between the input information D1 to D4 and the classification IDs [Ca] to [Cd].

In step B5, it is determined whether or not an input information symbol is designated in a state where the input information classification diagram in which the classification symbol and the input information symbol are arranged on the two-dimensional coordinate plane is displayed on the display device 40. Whether or not an input information symbol has been specified is, for example, an input information output indicating the input information symbol when the mouse pointer is placed at a position corresponding to the input information symbol or when the mouse is further clicked The determination can be made based on whether or not the request signal is input from the input device 30. If an input information symbol is instructed, the process proceeds to step B6, and if not, the process proceeds to step B7.
In step B6, the input information indicated by the input information symbol indicated by the input information output request signal input from the input device 30 is read from the input information database 20b and output. For example, the input information is displayed in a display window formed in the vicinity of the instructed input information symbol.
When the input information output request signal is input by placing the mouse pointer at a position corresponding to the input information symbol, the mouse pointer moves from a position corresponding to the input information symbol. The display of input information corresponding to the input information symbol can be stopped.
If the input information output request signal is input when the mouse pointer is placed at a position corresponding to the input information symbol and the mouse is clicked, the input information corresponding to the input information symbol is set. When the mouse pointer is placed at a position corresponding to the input information symbol while the input information symbol corresponding to the input information symbol is displayed or the mouse is clicked again. Thus, the display of the input information corresponding to the input information symbol can be stopped.

In step B7, it is determined whether or not a classification symbol is instructed in a state where the input information classification diagram in which the classification symbol and the input information symbol are arranged on the two-dimensional coordinate plane is displayed on the display device 40. Whether or not the classification symbol is instructed is determined, for example, by the input information symbol output request signal indicating the classification symbol when the mouse pointer is placed at a position corresponding to the classification symbol or further clicked. It can be determined by whether or not the input from 30. When the classification symbol is instructed, the process proceeds to Step B8, and when the classification symbol is not instructed, the process proceeds to Step B9.
In step B8, an input information symbol indicating input information having similarity with the classification indicated by the classification symbol indicated by the input information symbol output request signal is displayed in a distinguishable manner.
For example, consider the case where the input information classification diagram shown in FIG. In the input information classification diagram shown in FIG. 11, classification symbols of classification IDs [Cp] to [Ct] are arranged on a two-dimensional coordinate plane with classifications of classification IDs [Cp] to [Ct] as coordinate axes (p-axis to t-axis). And the input information symbol is displayed. In this state, when the mouse pointer is placed at a position corresponding to the classification symbol of the classification ID [Cq], an input information symbol output request signal indicating the classification symbol of the classification ID [Cq] is input from the input device 30. . In this case, an input information symbol (indicated by hatching in FIG. 12) indicating input information having similarity to the classification indicated by the classification symbol of the classification ID [Cq] indicated by the input information symbol output request signal. Input information symbols) are displayed in an output mode different from other input information symbols. The different output modes include a mode of outputting using different colors, a mode of outputting using different characters and figures, and the like.

Note that a step of executing the following processing may be provided instead of or in addition to step B10.
That is, an input information output request signal indicating a classification symbol is input from the input device 30 in a state where the input information classification diagram in which the classification symbol and the input information symbol are arranged on the two-dimensional coordinate plane is displayed on the display device 40. It is judged whether it was done. Whether or not the input information output request signal for instructing the classification symbol is input from the input device 30 is determined by, for example, the classification symbol being displayed when the mouse pointer is placed at a position corresponding to the classification symbol or further clicked. It can be determined by whether or not an input information output request signal for instructing is input from the input device 30. When an input information output request signal indicating a classification symbol is input, input information having a similarity to the classification indicated by the classification symbol indicated by the input information output request signal is read from the input information database 20b. Output. In this case, for example, a main display unit and a sub display unit that display the input information classification diagram may be provided on the display unit of the display device 40, and the input information may be displayed on the sub display unit. Regarding the display information display stop control, for example, the method described above can be used.

In step B9, it is determined whether or not the classification symbol has moved. As a method for determining whether or not the classification symbol has moved, for example, whether or not the classification symbol has been dragged by the mouse pointer (whether or not the position detection signal indicating the position of the mouse pointer that has acquired the classification symbol has moved). ) Can be used. The instructed classification symbol can be determined based on the position detection signal of the mouse pointer and the classification symbol coordinate information. If the classification symbol has moved, the process proceeds to step B10. If the classification symbol has not moved, the process proceeds to step B12.
In step B10, based on the mouse pointer position detection signal, the classification symbol designated by the mouse pointer and the position of the mouse pointer are determined. Then, the position of the mouse pointer is set as the arrangement position of the classification symbol designated by the mouse pointer (classification symbol coordinate information is calculated), and the classification symbol has a similarity with the classification indicated by the classification symbol designated by the mouse pointer. The input information symbol coordinate information of the input information symbol indicating the input information is calculated. The method for calculating the input information symbol coordinate information is the same as the input information symbol coordinate information calculation process in step B3. That is, based on the classification symbol coordinate information, the relationship between the classification symbol and the input information symbol on the two-dimensional coordinate plane corresponds to the similarity between the classification indicated by the classification symbol and the input information indicated by the input information symbol. Then, input information symbol coordinate information is calculated.

In step B11, the classification symbol and the input information symbol are displayed based on the classification symbol coordinate information and the input information symbol coordinate information calculated in step B10 (the display positions of the classification symbol and the input information symbol are moved).
For example, in a state where the input information classification diagram shown in FIG. 13 is displayed on the display device, the mouse pointer is arranged at a position corresponding to the classification symbol of the classification ID [Cq] and dragged as indicated by a broken line arrow And In this case, it can be determined from the position detection signal of the mouse pointer that the classification symbol of the classification ID [Cq] is indicated by the mouse pointer. Further, based on the position detection signal of the mouse pointer, the movement destination position of the classification symbol of the designated classification ID [Cq] is used as the classification symbol coordinate information of the classification symbol (classification coordinate information is calculated). Further, it is assumed that the classification symbol of the designated classification ID [Cq] is arranged at the position indicated by the position detection signal of the mouse pointer, and has a similarity to the classification indicated by the classification symbol of the classification ID [Cq]. The input information symbol coordinate information of the input information symbol indicating the input information is calculated. Then, the classification symbol of the classification ID [Cq] and the input information symbol of the input information having similarity to the classification of the classification ID [Cq] are indicated by the calculated classification symbol coordinate information and input information symbol coordinate information. Place in position. For example, in FIG. 13, the classification symbol indicating the classification of the classification ID [Cq] moves to the position where the drag operation has been performed. Further, the input information symbol of the input information having the similarity with the classification of the classification ID [Cq] (the input information symbol indicated by hatching in FIG. 13) is the movement of the classification symbol of the classification ID [Cq]. (In FIG. 13, it moves to the position of the input information symbol shown in black).
In step B12, it is determined whether an instruction to end the process has been issued. If the end of the process is instructed, the process ends. If the end of the process is not instructed, the process returns to step B1.
The processes of steps B5 to B9 and B11 are executed by the management unit 10a, and the process of step B10 is executed by the coordinate information calculation unit 10e.
The input information classification diagrams shown in FIGS. 11 to 13 are provided with a classification list section. With this classification list section, it is possible to instruct classification in step B1.

Next, word information extraction processing, word importance parameter determination processing, and similarity determination processing in the present embodiment will be specifically described with reference to FIG.
Now, in the input information database 20b, as input information Di, [D1] “We received a compliment from the shareholders that the management attitude such as the deletion of officers is well evaluated and that it also appears in the stock price.” [D2] “I thought I would buy it again when it was in the 2400 yen range, and I was asked about the future trend of the stock price.”, [D3] “I bought the stock several years ago, but I received a calendar. I don't think it's good as a shareholder service because it doesn't come in. Last year, it was finally sent a few times, and I have other companies' stocks. I would like you to consider sending a calendar to shareholders as well. ”[D4]“ Because your stock price rose this spring, you sold all your stock. "D5]" I participated in the general meeting of shareholders. I knew the enthusiasm about management well. Even though it was a large corporation, I would like to cherish individual investors. Is stored (total number of input information N = 5), and the classification word information database 20a includes F1 [stock], F2 [as classification word information Fj corresponding to the classification C1 [stock]. It is assumed that a classification [stock] including “stock price” and F3 [shareholder] is stored.

First, the word information extraction means 10b divides each of the input information [D1] to [D5] into sentence information, and extracts word information constituting each sentence information (step A1).
Next, the word importance parameter Wij (i = 1 to 5, j = 1) of the classification word information Fj (j = 1 to 3) corresponding to the classification C1 [stock] for the input information Di (i = 1 to 5). To 3) are discriminated by the following method, for example (step A2).
Classification word information Fj (j = 1 to 3) (F1 [stock], F2 [stock price], F3 [) corresponding to the word information constituting the input information Di (i = 1 to 5) and the classification C1 [stock] Shareholder]), the number of word information corresponding to the classification word information Fj in the input information Di (number of word information appearances) TFij (i = 1 to 5, j = 1 to 3) and the classification word information Fj The number (sentence information appearance frequency) DFij (i = 1 to 5, j = 1 to 3) of all the input information including the word information to be determined is determined. In the example shown in FIG. 6, for input information D1, the number of word information corresponding to F1 [stock] TF11 = [0], the number of word information corresponding to F2 [stock price] TF12 = [1], F3 The number TF13 = [1] of word information corresponding to [shareholder] is determined. For the input information D1 to D5, the number DFi1 (i = 1 to 5) = [3] in all input information of the input information corresponding to the classification word information F1 [stock] is determined.
Then, for each input information Di (i = 1 to 5), the number of word information appearances TFij (i = 1 to 5, j = 1) of each classification word information Fj (J = 1 to 3) corresponding to the classification C1 [stock]. 1 to 3) and the sentence information appearance frequency DFij (i = 1 to 5, j = 1 to 3), each classification corresponding to the classification C1 [stock] for each input information Di (i = 1 to 5) The word importance parameter Wij (i = 1 to 5, j = 1 to 3) of the word information Fj (j = 1 to 3) is determined. For example, as the word importance parameter W12 of the input information D1 and the word information F2 [stock price], [0.22] is determined by the following equation.
[W12 = TF12 × IDF12 = TF12 × log (5 / DF12)] = 0.22
Next, the word importance parameter Wij (I = 1 to 5, J) of each classification word information Fj (J = 1 to 3) corresponding to the classification C1 [stock] for each input information Di (i = 1 to 5). = 1 to 3), the similarity SIMi1 between each input information Di (i = 1 to 5) and the classification C1 [stock] is determined. For example, the similarity SIMin between the input information D1 and the classification C1 [stock] is determined as follows. Note that the input information vector << D1 >> of the input information D1 = (0.00, 0.22, 0.40), and the classification vector << C1 >> of the classification C1 = (0.8, 1.0, 1.0). ). Also,
SIMin = cos θ = << C1 >> · << D1 >> / (| << C1 >> | × | << D1 >> |)
= (0.8 × 0.00 + 1.0 × 0.22 + 1.0 × 0.40) / [SQRT (0.8 × 0.8 + 1.0 × 1.0 + 1.0 × 1.0)] × [SQRT (0.00 × 0.00 + 0.22 × 0.22 + 0.40 × 0.40)] = 0.84
SQRT (X) is an operation symbol for obtaining the square root of X.

As described above, in the present embodiment, the similarity between each of the plurality of input information and each of the plurality of classifications is determined, and the plurality of classifications are based on the similarity between each of the plurality of input information and each of the plurality of classifications. And an input information symbol indicating a plurality of input information are arranged on a two-dimensional coordinate plane having a plurality of classifications as coordinate axes. Thereby, the correlation between the plurality of input information and the plurality of classifications, the correlation between the input information, the trend of the entire plurality of input information, and the like can be easily grasped visually.
In addition, classification symbols indicating a plurality of classifications are arranged along the circumference of a circle or an ellipse, whereby the correlation between the input information and the classification can be grasped more easily.
In addition, since the plurality of classification symbols are arranged at positions close to the classification symbols having similar similarity pattern states with respect to the input information, the correlation between the input information and the classification can be accurately grasped.
In addition, in conjunction with the movement of the classification symbol, the input information symbol indicating the input information having similarity to the classification indicated by the classification symbol is also moved, so that the correlation between the input information and the classification can be grasped more appropriately. be able to.

The present invention is not limited to the configuration described in the embodiment, and various changes, additions, and deletions are possible.
For example, the configuration of the present invention is not limited to the configuration described in the embodiment.
The processing of each unit is not limited to the processing described in the embodiment. For example, a method of extracting word information from input information, a method of determining a word importance parameter indicating importance (weight) of classified word information in the input information, a method of determining similarity between input information and classification, classification The method for arranging the symbols and the input information symbols on the two-dimensional coordinate plane is not limited to the method described in the embodiment, and various methods can be used.
The classification word information corresponding to the classification and classification stored in the storage device 20 is not limited to the classification and classification word information described in the embodiment, and various classifications and classification word information can be selected.
The operation of each means can be changed as appropriate.
Each configuration described in the embodiment can be used individually, or a plurality of appropriately selected configurations can be used in combination.

It is a schematic block diagram of one embodiment of the present invention. It is a figure which shows an example of a classification word information database. It is a figure which shows an example of an input information database. It is a flowchart which shows an example of a similarity determination process. It is a flowchart which shows an example of an input information classification diagram output process. It is a figure explaining the operation | movement of one embodiment of this invention concretely. It is a figure which shows the relationship between a classification vector and an input information vector. It is a figure explaining the process which adjusts the arrangement position of a classification vector. It is the figure which has arrange | positioned the classification vector along the circle on the two-dimensional coordinate plane. It is a figure which shows the example which has arrange | positioned the classification vector and input information vector shown in FIG. 8 on the two-dimensional coordinate plane. It is a figure which shows an example of the input information classification | category diagram which is displaying the result of having analyzed input information using this invention. FIG. 12 is a diagram illustrating a state in which a classification symbol is designated in the input information classification diagram illustrated in FIG. 11. FIG. 12 is a diagram showing a state in which the designated classification symbol is moved in the input information classification diagram shown in FIG. 11.

10. Processing device (processing means)
10a management means 10b word information extraction means 10c word importance parameter determination means 10d similarity determination means 10e coordinate information calculation means 20 storage device 20a classified word information database 20b input information database 30 input device (input means)
40 Display device (output means)

Claims (9)

An input information analysis device for analyzing input information,
Input means, storage means, management means, word information extraction means, word importance parameter determination means, similarity determination means, coordinate information calculation means, and output means,
The storage means stores classification word information corresponding to each of a plurality of classifications,
The word information extracting means extracts word information constituting each of the plurality of input information input from the input means,
The word importance parameter determining means includes word information constituting each of a plurality of input information extracted by the word information extracting means, and classified word information corresponding to each of a plurality of classifications stored in the storage device And determining the word importance parameter of each of the plurality of input information and each of the classification word information corresponding to each of the plurality of classifications,
The similarity determination unit is configured to determine the plurality of the plurality of input information and the word importance parameter of each of the classification word information corresponding to each of the plurality of classifications determined by the word importance parameter determination unit. Determining the similarity between the input information and the plurality of classifications;
The coordinate information calculation means includes
The classification symbol indicating each of the plurality of classifications is arranged along the outer circumference of a circle or an ellipse on a two-dimensional coordinate plane, and the classification symbol indicates a classification having a high level of similarity with each of the plurality of input information Class symbol coordinate information indicating an arrangement position of each of the classification symbols on the two-dimensional coordinate plane for arranging the symbols at close positions ;
In addition, the input information symbol indicating each of the plurality of input information, the input information indicated by each input information symbol and the classification indicated by each classification symbol, the arrangement relationship between each input information symbol and each classification symbol, Calculating input information symbol coordinate information indicating an arrangement position of each input information symbol on the two-dimensional coordinate plane to be arranged at a position corresponding to the similarity of
The input information analysis apparatus characterized in that the management means outputs the classification symbol and the input information symbol from the output means based on the classification symbol coordinate information and the input information symbol coordinate information calculated by the coordinate information calculation means. The input information analysis apparatus according to claim 1,
The word importance parameter determining means includes word information constituting each of a plurality of input information extracted by the word information extracting means, and classified word information corresponding to each of a plurality of classifications stored in the storage device The number of appearances of word information indicating the number of pieces of word information corresponding to the classification word information in the input information is determined for each of the classification word information corresponding to each of the plurality of classifications with respect to each of the plurality of input information. And determining, for each of the classification word information corresponding to each of the plurality of classifications, the number of input information appearances indicating the number of input information including the word information corresponding to the classification word information, and further, the plurality of input information For each, the word information appearance times determined for each of the classification word information corresponding to each of the plurality of classifications. And the word importance parameter of each of the plurality of input information and each of the classification word information corresponding to each of the plurality of classifications based on the number of times of appearance of the input information determined for each of the classification word information corresponding to each of the plurality of classifications An input information analyzing apparatus characterized by discriminating between. The input information analysis device according to claim 1 or 2 ,
The management means receives the input information symbol output request signal for instructing the classification symbol when the classification means and the input information symbol are output from the output means. An input information symbol indicating input information having a similarity with the classification indicated by the classification symbol indicated by the input information symbol output request signal is output from the output means in a different output mode from the other input information symbols. An input information analyzer characterized by the above. The input information analysis apparatus according to any one of claims 1 to 3 ,
When the input means output request signal indicating the classification symbol is input from the input means in a state where the management means outputs the classification symbol and the input information symbol from the output means, the input information output that has been input An input information analyzing apparatus characterized in that input information having a classification and similarity indicated by a classification symbol indicated by a request signal is output from the output means. The input information analysis device according to any one of claims 1 to 4 ,
When the input information output request signal indicating the input information symbol is input from the input means in a state where the management means outputs the classification symbol and the input information symbol from the output means, the input information input An input information analyzing apparatus, wherein input information corresponding to an input information symbol indicated by an output request signal is output from the output means. The input information analysis apparatus according to any one of claims 1 to 5 ,
The coordinate information calculation means, when outputting the classification symbol and the input information symbol from the output means, when the input means indicates the classification symbol and the position of the classification symbol, the coordinate information of the indicated position And the positional relationship between the input information symbol indicating the input information having similarity with the classification indicated by the indicated classification symbol and the indicated classification symbol is indicated by the indicated classification symbol. The input information having the classification and the similarity indicated by the indicated classification symbol so as to correspond to the similarity between the input information having the classification and the similarity and the classification indicated by the indicated classification symbol. An input information analysis apparatus for calculating input information symbol coordinate information for arranging an input information symbol on the two-dimensional coordinate plane coordinates . The input information analysis apparatus according to any one of claims 1 to 6 ,
The management means outputs the classification symbol and the input information symbol in different output modes. A program for causing a computer to execute processing of a management unit, a word information extraction unit, a word importance parameter determination unit, a similarity determination unit, and a coordinate information calculation unit according to any one of claims 1 to 7 . A storage in which a program for causing a computer to execute the processing of the management unit, the word information extraction unit, the word importance parameter determination unit, the similarity determination unit, and the coordinate information calculation unit according to any one of claims 1 to 7 is recorded. Medium.

Images