JP7704401B2 - Image generating device, image generating method, and program - Google Patents

Description

The present invention relates to an image generation device, an image generation method, and a program.

Infographics are known as a method of visualizing information. Infographics are a visual representation of information that combines images. Combined with the appeal of the visual effect, infographics make it easy to remember the information you want to convey, so they are used in many situations, including news, signs, and meeting materials.

Approaches to automatically create infographics from text are being considered (Non-Patent Document 1). In Non-Patent Document 1, terms related to infographics, such as "n%", "m in n", "m out of n", and "half of", are extracted from text. Infographics are generated from text by applying the extracted terms to templates.

Weiwei Cui, Xiaoyu Zhang, Yun Wang, He Huang, Bei Chen, Lei Fang, Haidong Zhang, Jian-Guan Lou, Dong-mei Zhang, “Text-to-Viz: Automatic Generation of Infographics from Proportion-Related Natural Language Statements”, IEEE Transactions on Visualization and Computer Graphics, Vol. 26, pp. 906-916 (2020)

Generally, color has a large impact on visual effects, but Non-Patent Document 1 does not disclose or suggest anything about color. It is thought that by generating infographics with a focus on color, information can be more appropriately visually represented.

The present invention has been made in consideration of the above circumstances, and an object of the present invention is to provide a technology that can visually represent information more appropriately.

The image generating device of one aspect of the present invention includes a storage device that stores template data that defines the position at which a numerical expression representing a number is drawn in image data, the position at which the numerical name of the numerical expression is drawn, and the position of a polarity color, and hue definition data that defines a positive hue value and a negative hue value in a color space specified by a plurality of elements including hue, an acquisition unit that acquires input text and either a positive or negative polarity for the input text, an extraction unit that extracts a numerical expression and a numerical name from the input text, and an output unit that draws the numerical expression and the numerical name extracted by the extraction unit at the position of the numerical expression and the numerical name defined in the template data, acquires the polarity hue value acquired by the acquisition unit from the hue definition data, changes the hue of the color corresponding to the numerical name to the acquired hue value, and sets the color at the position of the polarity color to output image data.

The image generating method of one aspect of the present invention includes a step of storing template data that defines the position at which a numerical expression representing a number is drawn in image data, the position at which the numerical name of the numerical expression is drawn, and the position of the polarity color, and hue definition data that defines the positive hue value and the negative value in a color space specified by a plurality of elements including hue, a step by the computer acquiring an input text and either a positive or negative polarity for the input text, a step by the computer extracting a numerical expression and a numerical name from the input text, and a step by the computer drawing the numerical expression and the numerical name extracted in the extracting step at the position of the numerical expression and the numerical name defined in the template data, acquiring the polarity hue value acquired in the acquiring step from the hue definition data, setting the hue of the color corresponding to the numerical name to the acquired hue value at the position of the polarity color, and outputting the image data.

One aspect of the present invention is a program that causes a computer to function as the image generating device.

The present invention provides a technology that can visually represent information more appropriately.

FIG. 1 is a diagram illustrating functional blocks of an image generating apparatus according to an embodiment of the present invention. FIG. 2 is a diagram for explaining an example of the data structure and data of the hue definition data. FIG. 3 is a diagram for explaining an example of the data structure and data of the numeric color data. FIG. 4 is a diagram for explaining an example of the data structure and data of the numerical value change definition data. FIG. 5 is a diagram illustrating an example of the data structure and data of the template definition data. FIG. 6 is a diagram illustrating an example of template data. FIG. 7 is a diagram for explaining an example of image data generated from the template data shown in FIG. FIG. 8(a) is a diagram illustrating an example of the results of dividing an input text into words and assigning numerical expressions and numerical change labels, and FIG. 8(b) is a diagram illustrating the results of dependency syntax analysis of the input text. FIG. 9 is a diagram illustrating the colors acquired by the color acquisition unit. FIG. 10 is a flowchart illustrating an example of processing by the image processing device. FIG. 11 is a diagram for explaining the hardware configuration of a computer used in the image generating apparatus.

Below, an embodiment of the present invention will be described with reference to the drawings. In the description of the drawings, the same parts are given the same reference numerals and the description will be omitted.

(Image Generation Device)
The image generating device 1 according to the embodiment of the present invention generates image data 26, which is an infographic related to a numerical expression, from an input text including the numerical expression. The input text is data in which one or more sentences are expressed in a computer-processable data format, and is typically data in a text format.

In general, input text that contains numerical expressions can have either positive or negative sentiment depending on the attributes of the person presenting it. For example, if there is input text about rising land prices, sellers will feel positive about it, while buyers will feel negative about it.

The image generating device 1 then obtains the polarity along with the input text, and generates image data 26 in a color corresponding to the polarity. The polarity is set to a binary value of positive or negative. The polarity indicates whether the impression of the input text is positive or negative. In this way, the image generating device 1 can more appropriately visually express information by expressing the impression the viewer has in the image data 26 with a color.

1, the image generating device 1 includes the following data: hue definition data 11, numerical value name color data 12, numerical value change definition data 13, template definition data 14, template group data 15, input text 21, polarity 22, numerical value name 23, numerical value expression 24, template data 25, and image data 26, and the functions of an acquisition unit 31, an extraction unit 32, a selection unit 33, and an output unit 34. Each data is stored in a storage device such as a memory 902 or a storage 903. The hue definition data 11, numerical value name color data 12, numerical value change definition data 13, template definition data 14, and template group data 15 are stored in the image generating device 1 in advance before the image generating device 1 generates image data 26 from the input text 21. The input text 21, polarity 22, numerical value name 23, numerical value expression 24, template data 25, and image data 26 are stored in the image generating device 1 as the image generating device 1 generates image data 26 from the input text 21. Each function is implemented in the CPU 901.

Hue definition data 11 defines positive and negative hue values in a color space specified by multiple elements including hue. Examples of color spaces specified by multiple elements including hue include the HSB system (H: hue, S: saturation, B: brightness) and the HSV system (H: hue, S: saturation, B: value). When expressing color in the HSB system, the hue is expressed as a value from 0 to 360. As shown in Figure 2, the hue definition data 11 associates a range of hue values with each polarity, positive and negative.

In an embodiment of the present invention, the hue definition data 11 is generated with reference to Plutchik's two-dimensional emotion wheel to determine the hue values of each polarity. The hue values for positivity are 0 to 135, which correspond to anger, anticipation, joy, trust, etc. on Plutchik's two-dimensional emotion wheel. The hue values for positivity are 136 to 360, which correspond to trust, fear, surprise, sadness, disgust, and anger on Plutchik's two-dimensional emotion wheel. In the hue definition data 11, the hue values may be defined in any way as long as they correspond to the impressions people receive for each positive and negative polarity and the polarity.

Numeric name color data 12 is data that defines the color corresponding to a numeric name. A numeric name is the name of the object of a numeric expression contained in the input text. When the numeric expression is a price such as "200 yen," the numeric name is "price."

The numeric name color data 12 may correspond to a numeric name, an icon, and a representative color, as shown in FIG. 3. One icon and one representative color are associated with one numeric name. Multiple numeric names may be associated with one icon. An icon is an image that recalls a numeric name. An icon may be set in image data 26. A representative color is a color recalled by a numeric name 23. In the example shown in FIG. 3, the representative color is expressed by two hexadecimal digits for each RGB value. The representative color may be expressed by a color name or each HSB value.

The representative color is determined by any method. It may be determined from the color of the icon shown in Figure 3. Alternatively, the representative color may be determined from the colors of multiple icons that are commonly used as numerical names. Also, since white and black are commonly used, a condition may be set that if the color determined from the icon, etc. is "r > 200, g > 200 and b > 200" or "r < 55, g < 55 and b < 55", it will not be adopted as the representative color.

The numerical change definition data 13 associates terms related to numerical changes with numerical change labels that indicate either an increase or a decrease in the numerical value that the term indicates. As shown in FIG. 4, the numerical change definition data 13 associates a numerical change label with a term and a part of speech that corresponds to the numerical change label. Any numerical change label may be used as long as it indicates either an increase or a decrease in the numerical value. In another embodiment, a numerical change label that indicates no change in the numerical value may also be used.

In the example shown in FIG. 4, the numeric change labels are "UP" which indicates an increase in a numeric value and "DOWN" which indicates a decrease in a numeric value. In the example shown in FIG. 4, the term "surpass" evokes an increase in a numeric value, so it is associated with the numeric change label "UP." The term "slump" evokes a decrease in a numeric value, so it is associated with the numeric change label "DOWN."

The template definition data 14 defines the attributes of multiple template data held in the template group data 15 of the image generating device 1. As shown in FIG. 5, the template definition data 14 associates the identifier of the template data with the number of numerical expressions expressed in the template data and the number of changes in the numerical expressions. The changes in the numerical expressions are changes relative to the numerical expressions to be compared, and are expressed as percentages or magnifications. This is used when the input text contains not only numerical values represented by numerical expressions, but also changes in the numerical expressions, and the changes in the numerical expressions are expressed in the image data 26. The number of changes in the numerical expressions corresponds to the number of figures expressing percentages or magnifications included in the template data.

The template group data 15 is data that identifies multiple template data that are referenced when generating the image data 26. The template data defines the position at which each part included in the image data 26 is drawn, such as the position at which a numerical expression representing a number is drawn, the position at which the numerical name of the numerical expression is drawn, the position of the polarity color, the position at which an icon is drawn, and the position at which a change in the numerical expression is drawn. The templates included in the template group data 15 are associated with the template identifiers defined in the template definition data 14.

The template data will be described with reference to Figs. 6(a) and (b). Figs. 7(a) and (b) are examples of image data 26 generated by the image generating device 1 using the template data shown in Figs. 6(a) and (b), respectively.

The template in FIG. 6(a) has a numerical expression display section P1 for drawing numerical expressions, a numerical name display section P2 for drawing numerical names, and an icon display section P3 for displaying icons. As shown in FIG. 7(a), the image data 26 shows specific data in the respective positions of the numerical expression display section P1, the numerical name display section P2, and the icon display section P3.

The template in FIG. 6(b) includes a percentage display section P4 that displays the numerical value of the change in the numerical expression such as a percentage format in addition to the numerical expression display section P1, numerical name display section P2, and icon display section P3 included in the template in FIG. 6(a). As shown in FIG. 7(b), the image data 26 shows specific data at the respective positions of the numerical expression display section P1, numerical name display section P2, icon display section P3, and percentage display section P4. The shape of the partially missing circle in the percentage display section P4 in FIG. 7(b) may differ depending on the value represented by the percentage display section P4. As shown in FIG. 7(b), the percentage display section P4 shows "90%". If this numerical value is even lower, the circle represented by the percentage display section P4 will be even more missing. For example, if the percentage display section P4 represents "50%," the circle represented by the percentage display section P4 will be a semicircle.

6(a) and 6(b) each have a polarity color display section P5 in a frame. The polarity color display section P5 is a section that displays a color that is set according to the polarity 22 acquired by the acquisition section 31. Specifically, when the polarity 22 is positive, a color that evokes positivity is displayed in the polarity color display section P5. When the polarity 22 is negative, a color that evokes negativity is displayed in the polarity color display section P5. In the examples shown in FIGS. 6(a) and 6(b), the polarity color display section P5 is a frame that surrounds the numerical expression display section P1, etc., but is not limited to this. The polarity color display section P5 may be provided as a part that constitutes part of the template data, such as a pattern or background in the template data.

In addition, multiple polarity color display sections P5 may be defined in one template data. Each polarity color display section P5 may be displayed in the same color or in a different color. When each polarity color display section P5 is displayed in a different color, the template data 25 associates an identifier that specifies the color to be displayed with each polarity color display section P5. The template data 25 may associate each polarity color display section P5 with a shift amount of hue relative to a reference polarity color. Each polarity color display section P5 displays a color that differs only in hue but has the same saturation and brightness. The shift amount is the difference with respect to the hue value of the polarity color, and is expressed as +30, -50, etc.

The input text 21 is data input to the image generating device 1. The input text 21 specifies a sentence from which the image generating device 1 generates image data 26. The input text 21 includes a numerical expression representing a number and the numerical name of the numerical expression. Note that in the embodiment of the present invention, a "number" is a word that can be expressed in numbers, such as a quantity, an amount, or a percentage. Numbers include not only words that can be expressed in integers, but also words that can be expressed in any number, such as decimals and fractions.

Polarity 22 is data input to image generating device 1. Polarity 22 is positive or negative. Image generating device 1 changes the hue of polarity color display section P5 in image data 26 according to the value set in polarity 22. In the embodiment of the present invention, polarity is positive or negative, but three or more polarities may be set.

The numerical name 23 and the numerical expression 24 are each data extracted from the input text 21. The numerical name 23 and the numerical expression 24 are each extracted by the extraction unit 32. The input text 21 includes at least one numerical expression 24.

The template data 25 is one template data selected from the template group data 15. The template data 25 is selected by the selection unit 33.

The image data 26 is data generated by the image generating device 1 from the input text 21 and the polarity 22. The image data 26 is generated by the output unit 34. The image data 26 is an infographic generated from the input text 21 and the polarity 22. The image data 26 has a color corresponding to the polarity 22, making it possible to express an impression corresponding to the polarity 22 in color.

The acquisition unit 31 acquires the input text 21 and the polarity 22, which is either positive or negative, for the input text 21. The acquisition unit 31 acquires the input text 21 and the polarity 22, which are input by the user, for example.

The extraction unit 32 extracts the numerical name 23 and the numerical expression 24 from the input text 21. The extraction unit 32 extracts the subject in the input text 21 as the numerical name 23. When the input text 21 includes a plurality of numerical expressions, such as three or more, the extraction unit 32 extracts two numerical expressions 24 related to the subject in the input text 21. Of the plurality of numerical expressions included in the input text 21, the extraction unit 32 extracts the numerical expression 24 most related to the subject and a numerical expression 24 with the same label as that of this numerical expression. The label specifies the type of number, as described below. The numerical expression 24 extracted in this manner is considered to modify the numerical name that is the subject, and is appropriate as a numerical expression for the numerical name.

The extraction unit 32 performs morphological analysis and named entity extraction processing on the input text 21, and performs dependency syntactic analysis and part-of-speech tags on the input text 21. The extraction unit 32 determines subject-predicate pairs in the input text 21, and determines numerical names and numerical expressions from the determined subject-predicate pairs. In the embodiment of the present invention, a case will be described in which spaCy, Sudachi, and GiNZA, a Japanese Universal Dependencies model, are used as NPL (Natural Language Processing) libraries, but the present invention is not limited to this.

The extraction unit 32 performs morphological analysis on the input text 21 using Sudachi. The extraction unit 32 passes the analysis results to spaCy. The extraction unit 32 performs dependency structure analysis and part-of-speech tagging of the input text 21 using the numerical change definition data 13 and the Japanese Universal Dependencies model. As a result, the input text 21 is treated as a set of tokens divided into words.

The extraction unit 32 assigns a named entity, a numerical expression, and a numerical change label to each token. A named entity is a proper noun. A numerical expression is a word expressed as a number, such as a quantity, an amount, or a percentage. The extraction unit 32 defines, among the labels classified as numerical expressions, labels that represent numbers as numerical expressions. Labels representing numbers include, for example, "number of people (N_Person)", "number of organizations (N_Organization)", "number of locations (N_Location)", "number of locations_other (N_Location_Other)", "number of countries (N_Country)", "number of facilities (N_Facility)", "number of products (N_Product)", "number of events (N_Event)", "number of natural objects (N_Natural_Object)", "number of natural objects_other (N_Natural_Object_Other)", "number of animals (N_Animal)", "number of plants (N_Flora)", "money expression (Money)", "percentage expression (Percent)", "multiplication expression (Multiplication)", "frequency expression (Frequency)", and "age (Age)".

Figure 8 (a) shows the result of the extraction unit 32 dividing the sentence "The price of vegetables has fallen from 200 yen to 150 yen" into words and assigning numerical expressions and numerical change labels. The divisions in Figure 8 (a) are divisions of the divided words. The extraction unit 32 assigns labels of monetary expressions to each of "200 yen" and "150 yen". The extraction unit 32 assigns a numerical change label to "declined".

Figure 8(b) shows the result of dependency syntax analysis. Figure 8(b) associates each word separated from the sentence "The price of vegetables has dropped from 200 yen to 150 yen" with a part of speech, and further shows the dependency of each part of speech.

Next, the extraction unit 32 determines subject and predicate pairs in the input text 21. We will explain the case where the extraction unit 32 first extracts multiple subject-predicate pairs and then determines one subject-predicate pair.

First, the process of extracting the numerical name 23 contained in the input text 21 will be described. The extraction unit 32 searches for a token that can be a subject of the numerical name. The extraction unit 32 searches for a token that has a dependency tag of nsubj, which means a subject noun, or obj, which means an object, from the beginning of the sentence, and sets the searched token as a subject word candidate. In this case, if the subject word candidate token is obj, the extraction unit 32 determines whether or not the two conditions "the obj token does not contain a target numerical expression" and "an nmod token, which means a noun modifier, appears at the beginning of the sentence from the obj token" are satisfied. If both conditions are satisfied, the extraction unit 32 sets the obj token as a subject word candidate. On the other hand, if either condition is not satisfied, specifically, if "the obj token contains a target numerical expression" or "an nmod token does not appear", the extraction unit 32 removes the obj token from the subject word candidates. These conditions are in consideration of adopting the subject "infected" rather than "20 million" when the input text 21 is "The number of people infected with the new coronavirus in the United States has exceeded 20 million in total, reaching over 20 million, at over 20,007,000." Also, when a numerical expression appears in obj as a sentence construction, the word nmod, which can be another subject, often already appears in the text.

The extraction unit 32 extracts the subject in the input text 21 and identifies the predicate that corresponds to the subject.

When there are multiple extracted subject-predicate pairs, the extraction unit 32 narrows down the multiple pairs to one pair. The subject of the subject-predicate pair narrowed down here is a likely numerical name. The extraction unit 32 obtains the distance between the subject of each subject-predicate pair and a token in the input text 21 to which any of the labels of numerical expressions is attached. The distance here is the number of tokens existing between the two target tokens. In addition, when the subject token is one character, it is often not appropriate as a subject, so in the case of a one-character subject, the extraction unit 32 may add a weight so that the distance between the subject and the numerical expression is 1.5 times. For each subject-predicate pair, the extraction unit 32 calculates the average value of the distance between the subject and each numerical expression token in the input text 21, and adopts the one with the smallest average distance, that is, the one with the closest subject among the multiple numerical expressions, as the final subject-predicate pair. In addition, the subject of the final subject-predicate pair becomes a numerical name.

There are cases where no subject-predicate relationships are obtained from the input text 21. In such cases, the extraction unit 32 obtains multiple keywords from the input text 21 and determines the subject from these keywords. A keyword is composed of a noun and a word that describes the noun, and is, for example, a proper noun token with an obj tag. For each keyword, the extraction unit 32 counts the number of noun tags NOUN and the number of other tags for the part-of-speech tags of each word included in the keyword. The extraction unit 32 determines the keyword with the largest number of noun tag words as the subject among the multiple keywords. When there are multiple keywords with the largest number of noun tag words, the extraction unit 32 determines the keyword with the smallest number of non-noun part-of-speech tag words among the keywords as the subject. The extraction unit 32 obtains tokens related to the subject as predicates.

The extraction unit 32 can obtain the numerical name in the input text 21 through the above process, but the extended subject may also be the numerical name. The extended subject is expressed as, for example, "A's B" or "ABC." "A's B" is, for example, "Sales of 'Game'." Even if the subject is "Sales," if "Game" is set as a modifier for the sales, the extraction unit 32 may also use "Sales of the game" as the subject. "ABC" is, for example, "Virus, New, Infected." Even if the subject as a word is "Infected," if "New Virus" is set as a modifier for the infected, the extraction unit 32 may use "New Virus Infected" as the subject. In addition, when the subject is one character and the modifier of the subject is one character, a proper noun of two or more characters before the adopted modifier may be adopted as the extended subject, assuming that the subject at the beginning of the sentence plays a larger role as the subject.

Next, the process of extracting the numerical expression 24 contained in the input text 21 will be described. The extraction unit 32 determines the numerical expression 24 from the subject-predicate pair. When the input text 21 contains one numerical expression, the numerical expression is set as the numerical expression 24 of the input text.

When the input text 21 contains multiple numerical expressions, the extraction unit 32 determines the degree of association between the subject and each numerical expression, except for "Percent" and "Multiplication," which mean a percentage. First, when a word with a numerical expression tag and the subject have a dependency or modification relationship, the extraction unit 32 increments the degree of association between the subject and the corresponding numerical expression by 1. At this time, when the numerical expression token is the subject itself, the extraction unit 32 does not change the degree of association.

Finally, the extraction unit 32 selects the numerical expression token with the highest degree of association as a candidate numerical expression in the input text 21. Here, the numerical expression includes multiple types of expressions such as "number of products (N_Product)" and "money expression (Money)". If there are tokens to which the same type of numerical expression as the candidate numerical expression is assigned, these tokens are adopted as the numerical expression 24 in the input text 21. For example, a case will be described in which the input text 21 is "New product, priced at 220 yen instead of 200 yen for a pack of three." In this input text 21, a monetary expression is assigned to each of "200 yen" and "220 yen," and the number of products is assigned to "3 pieces." If the subject is "new product," the degree of association of "220 yen" with "new product" is higher than the degrees of association of "3 pieces" and "200 yen." Therefore, the extraction unit 32 selects "220 yen," which has the highest degree of relevance to the subject, and "200 yen," which is a numerical expression of the same type as this numerical expression, as numerical expression candidates, and excludes "3 pieces" from the numerical expression candidates.

Here, words labeled "Percent" and "Multiplication" are treated as changes in the adopted numerical expression. Note that if there are no changes in the numerical expression contained in the input text 21 due to the prior removal of "Percent" and "Multiplication," words labeled "Percent" or "Multiplication" may be adopted as changes in the numerical expression in the input text 21.

Next, the extraction unit 32 determines the numerical change in the input text 21. Here, tokens to which numerical change labels have been assigned in the numerical change definition data 13 are referenced.

First, when a numerical change label is assigned to a predicate, the extraction unit 32 takes the numerical change label assigned to the predicate as the numerical change in the input text 21. When a word to which a numerical change label is assigned is related to a predicate, the extraction unit 32 takes the numerical change label assigned to the word related to the predicate as the numerical change in the input text 21. When a word to which a numerical change label is assigned is a subject, the numerical change label assigned to the subject is taken as the numerical change in the input text 21. When a word to which a numerical change label is assigned, or a word in a dependency relationship with that word, is related to a numerical expression, the numerical change label assigned to that word is taken as the numerical change.

When the input text 21 includes multiple numerical change labels, the extraction unit 32 determines the numerical change label at the end of the sentence as the numerical change in the input text 21. When the input text 21 does not include a word to which a numerical change label is assigned, and the token "Percent" or "Multiplication" is related to a word related to UP or DOWN, the extraction unit 32 determines the numerical change based on this label classification. For example, when "20% increase" is included in two tokens, "20%" and "increase", "Percent", "20%" and "increase" related to UP have a consecutive relationship, and when "20% increase" is divided into one token, this token and the word "increase" related to UP have an inclusive relationship. Alternatively, when a token of "N_xxx" such as "number of people (N_Person)" includes a word related to UP or DOWN, the extraction unit 32 determines the numerical change based on this label classification. Here, words related to UP are "increase", "up", "over", etc. Words related to DOWN are "decrease", "down", etc.

For example, when a word with a token of "Percent" or "Multiplication" is immediately followed by a word indicating a numerical change, the values converted from these words are expressed as a graph or the like that represents image data 26. For example, "a 10% decrease" is converted to "90%", "a 30% increase" is converted to "130%", and "1.5 times" is converted to "150%". Numerical fluctuations may also be converted to be normalized and reflected in a graph or the like that represents an infographic. For example, when the phrase "a 10% decrease" is expressed as an infographic, "90%" that has been converted to normalize "10%" may be expressed as a graph or the like.

By the above process, the extraction unit 32 extracts the numerical name 23 and the numerical expression 24 from the input text 21. The extraction unit 32 also obtains the numerical value of the change in the numerical expression, which is a percentage or a multiplication factor, such as "Percent" or "Multiplication." Note that the extraction unit 32 is only required to be able to extract at least the numerical name 23 and the numerical expression 24, and may extract these by a process different from the above process.

The selection unit 33 extracts template data 25 from the template group data 15 based on the number of numerical expressions 24 and the number of changes in the numerical expressions extracted by the extraction unit 32. The selection unit 33 refers to the template definition data 14 to identify an identifier of a template that can express the number of numerical expressions 24 and the number of changes in the numerical expressions extracted by the extraction unit 32. The selection unit 33 selects template data 25 corresponding to the identified identifier from the template group data 15.

The output unit 34 outputs the image data 26 according to the template data 25 selected by the selection unit 33. The output unit 34 draws each piece of information on the numerical expression 24 and the numerical name 23 extracted by the extraction unit 32 at the positions of the numerical expression and the numerical name defined in the template data 25. The output unit 34 draws the numerical name 23 and the numerical expression 24 extracted by the extraction unit 32 in the numerical name display section P2 and the numerical expression display section P1 of the template data 25. The output unit 34 also extracts an icon corresponding to the numerical name 23 from the numerical name color data 12 and draws the extracted icon in the icon display section P3. If there is no record in the numerical name color data 12 that contains a numerical name that matches the numerical name 23, an icon may be extracted from a record that contains a numerical name that is a synonym of the numerical name 23, or a record that contains a numerical name that is highly similar to the numerical name 23. If the extraction unit 32 extracts multiple numerical names 23, the output unit 34 draws an icon corresponding to any one of the numerical names. If the template data 25 has a percentage display section P4 that displays changes in numerical expression, it is drawn according to the changes in numerical expression extracted by the extraction section 32.

The output unit 34 also renders the polarity color display section P5 of the template data 25 in a color acquired by the color acquisition unit 35 described below. The color acquisition unit 35 acquires the color of the polarity color display section P5 in the template data.

The color acquisition unit 35 acquires the polarity hue value acquired by the acquisition unit 31 from the hue definition data 11. The color acquisition unit 35 also acquires the representative color (color) corresponding to the numerical name 23 extracted by the extraction unit 32 from the numerical name color data 12. If there is no record in the numerical name color data 12 that includes a numerical name that matches the numerical name 23, the representative color may be extracted from a record that includes a numerical name that is a synonym of the numerical name 23, or a record that includes a numerical name that is highly similar to the numerical name 23. If the hue of the representative color is within the range of the hue value acquired from the hue definition data 11, the color acquisition unit 35 sets the representative color as the polarity color. If the hue of the representative color is not within the range of the hue value acquired from the hue definition data 11, the color acquisition unit 35 changes the hue of the representative color to the hue value acquired from the hue definition data 11, and sets the color having the saturation and brightness of the representative color as the polarity color. The polarity color display section P5 of the template data 25 is displayed in the polarity color acquired by the color acquisition section 35, and image data 26 is generated.

The hue of the polar color acquired by the color acquisition unit 35 may be within the range of the polar hue values acquired by the acquisition unit 31 in the hue definition data 11.

For example, when the template data 25 associates the polarity color display section P5 with an identifier that specifies the color to be displayed in each polarity color display section P5, the color acquisition section 35 acquires multiple colors used in the polarity color display section P5. One of the multiple colors is the polarity color, and the other colors are colors in which the hue of the polarity color is changed. The color acquisition section 35 may generate the image data 26 using a color in which the hue of the polarity color is further changed. When the template data 25 has multiple polarity color display sections P5, colors that differ only in hue may be displayed in each section. The color acquisition section 35 acquires the polarity color and a color that changes the hue of the polarity color by ±30, for example, so that the polarity color display sections P5 are displayed in similar colors.

An example of a method for determining the hue of a color acquired by the color acquisition unit 35 is described below. Here, a case is described in which three colors are used in the polarity color display unit P5 in the template data 25.

The color acquisition unit 35 first maps the representative color (A) corresponding to the numerical value 23 onto the HSB hue wheel. The color on the mapping is called A'. Using A' as a reference, while fixing S (saturation) and B (lightness) to the saturation and lightness of the representative color, it acquires the colors of a triad (B1, B2) where H (hue) is ±120, or the colors of a tetrad (C1, C2, C3) where H is ±90. Next, the color acquisition unit 35 determines one color from A', B1, B2, C1, C2, and C3 that is included in the range corresponding to the polarity 22 acquired by the acquisition unit 31 in the hue definition data 11.

Let D be the one color determined. If the saturation or brightness of D is low, the color selected based on D will also be dark. Therefore, the color acquisition unit 35 may adjust the color of D by setting the lower limit of the saturation value of D to 50 and the lower limit of the brightness value to 60. Let the adjusted color be D'1. The color acquisition unit 35 selects two colors based on D'1. For example, the color acquisition unit 35 obtains colors (D'2, D'3) with the hue of D'1 set to ±30. Here, the difference in hue between D'1 and D'2, and between D'1 and D'3 is set to 30, but the value of the difference may be set appropriately. The color acquisition unit 35 determines D'1, D'2, and D'3 as the colors to display the polarity color display unit P5. The output unit 34 displays the polarity color display unit P5 with D'1, D'2, and D'3.

Referring to FIG. 9, we will explain the three colors acquired by the color acquisition unit 35 when the color corresponding to the numerical name 23 is color code #6aaa30. Here, the polarity 22 is assumed to be positive.

The color acquisition unit 35 converts the color code #6aaa30 into the HSB system (92,72,67). Based on (92,72,67), the color acquisition unit 35 sets two colors (1,72,67), (211,72,67), and (311,72,67) from the triad method and three colors (181,72,67), (271,72,67), and (1,72,67) from the tetrad method as candidates for polarity colors. The hues of the two colors (1,72,67), (211,72,67), and (311,72,67) from the triad method are shown in Figure 9 (a). The hues of the three colors (181, 72, 67), (271, 72, 67), and (1, 72, 67) calculated using the tetrad method are shown in Figure 9 (b).

In the embodiment of the present invention, the hues that are positively associated in the hue definition data 11 are 0-135. The color acquisition unit 35 selects the polarity color from (92, 72, 67) and (1, 72, 67) whose hues are in the range of 0-135 among the six colors selected as candidates for the polarity color. When the color acquisition unit 35 selects (92, 72, 67) as the polarity color, it acquires similar colors (121, 72, 67) and (61, 72, 67) with H values of ±30. Therefore, the polarity color display unit P5 is colored with (92, 72, 67), (121, 72, 67), or 61, 72, 67) as shown in FIG. 9(c).

With reference to Figure 10, an image generation method using the image generation device 1 according to an embodiment of the present invention will be described.

First, in step S1, the image generating device 1 acquires the input text 21 and the polarity 22. In step S2, the image generating device 1 extracts the numerical name 23 and the numerical expression 24 from the input text 21.

In step S3, the image generating device 1 selects template data 25 from the numerical expression number extracted in step S2, etc. In step S4, the image generating device 1 changes the hue of the color corresponding to the numerical name 23 extracted in step S2 to the hue value corresponding to the polarity 22 obtained in step S1. Note that if the hue of the color corresponding to the numerical name 23 extracted in step S2 is within the range of hues associated with the polarity obtained in step S1, the image generating device 1 does not need to change the hue of the color corresponding to the numerical name 23.

In step S5, the image generating device 1 sets the numerical representation 24 and numerical name 23 extracted in step S2 to the template data 25 selected in step S3, and generates image data 26 using the color whose hue was changed in step S4.

The image generating device 1 according to an embodiment of the present invention acquires input text 21 necessary for generating image data 26 of an infographic, and also acquires polarity 22, which is set to positive or negative. Polarity 22 indicates whether a person who refers to the image data 26 has positive or negative feelings toward the input text 21. The image generating device 1 generates image data 26 using colors according to the polarity 22. As a result, a person who refers to the image data 26 is likely to have positive or negative feelings from the image data 26. Furthermore, the numerical value is positive or negative depending on the position of the person who refers to it. The image generating device 1 can generate image data 26 according to the position of the person who refers to the infographic.

In the embodiment of the present invention, a case where polarity 22 is input has been described, but this is not limiting. Instead of polarity 22, a subject referring to image data 26 may be input, and image generating device 1 may determine polarity 22. Image generating device 1 may refer to polarity dictionary data (not shown) that associates a numerical change of UP or DOWN with a polarity for a subject, identify polarity 22 from the input subject and the numerical change determined by extraction unit 32, and generate image data 26 according to the identified polarity 22.

In this way, the image generating device 1 can generate image data 26 that can visually represent information more appropriately.

The image generating device 1 of the present embodiment described above uses, for example, a general-purpose computer system including a CPU (Central Processing Unit, processor) 901, a memory 902, a storage 903 (HDD: Hard Disk Drive, SSD: Solid State Drive), a communication device 904, an input device 905, and an output device 906. In this computer system, the CPU 901 executes a program loaded on the memory 902, thereby realizing each function of the image generating device 1.

Note that the image generating device 1 may be implemented in one computer, or in multiple computers. The image generating device 1 may also be a virtual machine implemented in a computer.

The program for the image generating device 1 can be stored on a computer-readable recording medium such as a HDD, SSD, USB (Universal Serial Bus) memory, CD (Compact Disc), or DVD (Digital Versatile Disc), or can be distributed via a network.

The present invention is not limited to the above embodiment, and many variations are possible within the scope of the invention.

REFERENCE SIGNS LIST 1 Image generating device 11 Hue definition data 12 Numerical value name color data 13 Numerical value change definition data 14 Template definition data 15 Template group data 21 Input text 22 Polarity 23 Numerical value name 24 Numerical value expression 25 Template data 26 Image data 31 Acquisition unit 32 Extraction unit 33 Selection unit 34 Output unit 35 Color acquisition unit 901 CPU
902 Memory 903 Storage 904 Communication device 905 Input device 906 Output device P1 Numerical expression display section P2 Numerical name display section P3 Icon display section P4 Percentage display section P5 Polarity color display section

Claims (6)

template data defining, in the image data, a position at which a numerical expression representing a number is to be drawn, a position at which a numerical name of the numerical expression is to be drawn, and a position of a polarity color;
a storage device that stores hue definition data that defines positive and negative hue values in a color space that is specified by a plurality of elements including hue;
An acquisition unit that acquires an input text and an impression of the input text as either positive or negative;
an extraction unit for extracting numerical expressions and numerical names from the input text;
drawing the numerical expression and the numerical name extracted by the extraction unit at positions of the numerical expression and the numerical name defined in the template data;
obtaining a hue value of the polarity obtained by the obtaining unit from the hue definition data;
an output unit that changes the hue of the color corresponding to the numerical name to the acquired hue value, sets the resulting color in the position of the polarity color, and outputs image data. The image generating device according to claim 1 , wherein the output unit generates the image data by using a color obtained by changing a hue of the polarity color. The image generating device according to claim 1 , wherein the extraction unit extracts a subject in the input text as the numerical name. The image generating device according to claim 1 , wherein, when the input text includes a plurality of numerical expressions, the extraction unit extracts a numerical expression related to a subject in the input text. A computer generates template data that defines, in the image data, a position at which a numerical expression representing a number is to be drawn, a position at which a numerical name of the numerical expression is to be drawn, and a position of a polarity color;
A step of storing hue definition data that defines positive and negative hue values in a color space specified by a plurality of elements including hue;
The computer acquires an input text and an impression of the input text, the impression being either positive or negative;
extracting numerical expressions and numerical names from the input text;
the computer draws the numerical expressions and numerical names extracted in the extracting step at positions of the numerical expressions and numerical names defined in the template data;
Obtaining a hue value of the polarity obtained in the obtaining step from the hue definition data;
a step of changing the hue of the color corresponding to the numerical name to the obtained hue value, setting the color in a position of the polarity color, and outputting image data. A program for causing a computer to function as an image generating device according to any one of claims 1 to 4.

Images