JP7543339B2 - Document mapping display device, document mapping display method, and document mapping display program - Google Patents

Description

The present invention relates to a document mapping display device, a document mapping display method, and a document mapping display program, and in particular to a document mapping display device and method and program for visualizing in which fields documents such as papers and patent publications are increasing.

For published documents such as papers and patent publications, the location and number of documents are sometimes expressed as a distribution map on a two-dimensional plane, focusing on specific sentences present in the document. For example, a visual display (such as a so-called heat map) has been used in which the color scheme is darkened depending on the amount of information elements on the two-dimensional plane (see Patent Document 1).

However, with the method of arranging multiple documents on a two-dimensional plane, although it is possible to visually display the similarity of documents, it can be influenced by the experience and intuition of the person who understands the arrangement on the two-dimensional plane, and it cannot be said to be sufficiently objective. Furthermore, with the method of arranging multiple documents on a two-dimensional plane, when the collection of multiple documents changes, the arrangement of the documents on the two-dimensional plane also changes, making it difficult to determine the progress of the documents over time. Note that with previous display methods, multiple documents are visually organized on a two-dimensional plane with reference to certain sentences present in the documents, and this was used to investigate trends in documents, such as which fields are attracting attention.

Japanese Patent Application Publication No. 9-114859

The present invention has been made in consideration of the above points, and provides a document mapping display device that ensures objectivity when displaying published documents such as papers and patent publications in a two-dimensional distribution map focusing on specific sentences present in the document, and that can visualize the progress of documents over time, as well as a document mapping display method and a document mapping display program.

That is, the document mapping display device of the embodiment is characterized by comprising a document acquisition unit that acquires multiple documents, a sentence acquisition unit that acquires specified sentences from the multiple documents, a document arrangement unit that arranges the multiple documents on a two-dimensional plane according to similarities between specified sentences contained in each of the multiple documents, an extraction unit that extracts the amount of change in the number of documents over time in a specified area existing on the two-dimensional plane, a calculation unit that calculates a growth degree based on the amount of change in the number of documents over time in a specified area existing on the two-dimensional plane, and an output unit that outputs the growth degree.

In addition, the document mapping display device of the embodiment is characterized by comprising a document acquisition unit that acquires multiple documents, a sentence acquisition unit that acquires specified sentences from the multiple documents, a document arrangement unit that arranges the multiple documents on a two-dimensional plane according to the similarity of specified sentences contained in each of the multiple documents, an extraction unit that extracts the amount of change in the number of documents over time in a specified area existing on the two-dimensional plane, a calculation unit that calculates a growth degree based on the amount of change in the number of documents over time in a specified area existing on the two-dimensional plane, an accumulation detection unit that detects an accumulation area based on the multiple documents in the specified area existing on the two-dimensional plane, and a time series change output unit that displays the growth degree in the specified area and the time series change points in the accumulation area on the two-dimensional plane.

Furthermore, the document acquisition unit may be equipped with a crawling unit and acquire multiple documents via an Internet line.

Furthermore, tag information indicating characteristics of each of the multiple documents may be attached to each of the multiple documents, and the document acquisition unit may acquire the multiple documents based on the tag information.

Furthermore, the placement of specified sentences on a two-dimensional plane according to similarities between sentences in the document placement section may be based on natural language processing.

Furthermore, the document placement unit may include a vectorization unit that generates a feature vector for each of the multiple documents using natural language processing.

Furthermore, the extraction unit may be configured to extract the number of documents present within a given region, the given region being one of the sections that is generated by dividing a two-dimensional plane into a grid.

Furthermore, the extraction unit may extract the amount of change in the number of documents over time based on a comparison of the number of documents in a predetermined area existing on a two-dimensional plane.

Furthermore, the extraction unit may extract a change in a set of specified sentences on a two-dimensional plane, and the extraction unit may extract a difference in the change in the set of specified sentences on a two-dimensional plane based on the amount of change in the number of documents in a specified area.

Furthermore, the calculation unit may calculate the growth rate of a given section from the amount of change over time in the number of documents present in another section adjacent to the given section.

The output unit may further display the degree of growth on a two-dimensional plane using a type of arrow. The output unit may further display the degree of growth for each document as a numerical value.

Furthermore, the accumulation detection unit may detect an accumulation area based on the density of multiple documents in a specified area existing on a two-dimensional plane, and any number of accumulation areas may be specified for the accumulation area. Furthermore, the time series change output unit may display time series change points in the accumulation area of multiple documents on a two-dimensional plane as circles and lines connecting the circles. Furthermore, the time series change output unit may display the magnitude of the growth level on a two-dimensional plane using types of arrows.

The document mapping display device of the present invention includes a document acquisition unit that acquires a plurality of documents, a text acquisition unit that acquires a predetermined text from the plurality of documents, a document arrangement unit that arranges the plurality of documents on a two-dimensional plane according to the similarity of the predetermined text contained in each of the plurality of documents, an extraction unit that extracts the amount of change in the number of documents in a predetermined area on the two-dimensional plane over time , a calculation unit that calculates a growth rate based on the amount of change in the number of documents in a predetermined area on the two-dimensional plane over time, and an output unit that outputs the growth rate, so that it is possible to ensure objectivity when expressing documents such as published papers and patent publications on a distribution map on a two-dimensional plane by focusing on the predetermined text contained in the document, and to visualize the transition of documents over time. Note that the same effect can be obtained in the document mapping display method and the document mapping display program.

In addition, the document mapping display device of the present invention includes a document acquisition unit that acquires a plurality of documents, a text acquisition unit that acquires a predetermined text from the plurality of documents, a document arrangement unit that arranges the plurality of documents on a two-dimensional plane according to the similarity of the predetermined text contained in each of the plurality of documents, an extraction unit that extracts the amount of change in the number of documents in a predetermined area existing on the two-dimensional plane over time , a calculation unit that calculates the degree of growth based on the amount of change in the number of documents in a predetermined area existing on the two-dimensional plane over time, an accumulation detection unit that detects an accumulation area based on the plurality of documents in the predetermined area existing on the two-dimensional plane, and a time series change output unit that displays the degree of growth in the predetermined area and the time series change points in the accumulation area on the two-dimensional plane on the two-dimensional plane, so that it is possible to ensure objectivity when expressing documents such as published papers and patent publications on a distribution map on a two-dimensional plane by focusing on the predetermined text contained in the document, and to visualize the transition of documents over time. The same effect can be obtained in the document mapping display method and the document mapping display program.

1 is a schematic configuration diagram showing an overview of a document mapping display device common to first and second embodiments. FIG. 2 is a block diagram showing a configuration within a document mapping display device common to the first and second embodiments. 1 is a block diagram showing functional units in a document mapping display device according to a first embodiment; FIG. 2 is a schematic diagram of a two-dimensional plane displaying a plurality of documents according to the first embodiment. 1A is a schematic diagram of a two-dimensional plane, and FIG. 1B is a schematic diagram showing grid-like divisions. FIG. 11 is a first schematic diagram showing an example of calculation of growth degree in a grid-shaped section, where (A) is an example of calculation in the horizontal axis direction, (B) is an example of calculation in the vertical axis direction, and (C) is an example of calculation for determining the overall growth degree of a certain section. FIG. 2 is a second schematic diagram showing an example of calculating the growth rate in a grid-shaped section, where (A) is an example of the calculation in the horizontal axis direction, (B) is an example of the calculation in the vertical axis direction, and (C) is an example of the calculation of the overall growth rate of a certain section. 13 is an example of how the growth level is displayed, where (A) is an example of an arrow display, and (B) is an example of a numerical value display. 4 is a first flowchart showing main control of the document mapping display device of the first embodiment. 11 is a second flowchart showing main control of the document mapping display device of the first embodiment. 11 is a third flowchart showing main control of the document mapping display device of the first embodiment. FIG. 11 is a block diagram showing functional units in a document mapping display device according to a second embodiment. FIG. 11 is a schematic diagram of a two-dimensional plane displaying a plurality of documents according to the second embodiment. FIG. 14 is a schematic diagram showing circles of document accumulation areas and lines connecting the circles superimposed on the two-dimensional plane of FIG. 13 . FIG. 15 is a partially enlarged schematic diagram of FIG. 14 . 13 is a first flowchart showing main control of the document mapping display device of the second embodiment. 13 is a second flowchart showing main control of the document mapping display device of the second embodiment. 13 is a third flowchart showing main control of the document mapping display device of the second embodiment.

The document mapping display device of the first and second embodiments is a device that displays on a two-dimensional plane the degree of similarity between multiple documents based on specific sentences present in those documents, i.e., the closeness between documents, and extracts the amount of change in the number of documents on the two-dimensional plane, visually clarifying and displaying the direction in which the number of documents is increasing.

The multiple documents include, for example, papers (research papers, academic conference reports), technical reports, published patent bulletins, patent publications, etc. issued domestically or internationally. In addition, any type of document written in text, such as newspaper or magazine articles, laws, regulations, notices made public by legislative, administrative, or judicial bodies, or announcements from corporations, etc., is acceptable.

A specific sentence that exists in multiple documents is a sentence that indicates the specific meaning found in the documents. For example, the name of the document, a summary of the document, or an abstract of the document. Of course, the number of sentences is not limited to one per document, but can be any appropriate number.

In addition, tag information indicating the characteristics of each document is attached to each document. If the document is a paper, the tag information includes a summary of the document, keywords for the document, the year of publication of the document, and bibliographic information such as the author and affiliation. In the case of patent publications, bibliographic information such as the International Patent Classification (IPC), inventor, and applicant is also added to the tag information. Note that this tag information is also input by the user of the document mapping display device.

Figure 1 is a schematic diagram showing the configuration of a literature mapping display device 1 of an embodiment common to the first and second embodiments. Literature to be the subject of literature mapping is stored in fixed media 2 such as CD-ROM or DVD-ROM, or is available via an Internet line 3. Literature obtained via the fixed media 2 or the Internet line 3 is obtained by the literature mapping display device 1. The literature mapping display device 1 is any of various electronic computers (computing resources, computers) such as a personal computer (PC), tablet terminal, or smartphone. In addition, a server (not shown) for storing data is connected to the literature mapping display device 1.

Figure 2 is a block diagram showing the internal configuration of the document mapping display device 1 common to the first and second embodiments. As can be seen from this block diagram, in terms of hardware, it is composed of a CPU 11, RAM 12, ROM 13, storage unit 14, and I/O (input/output interface) 15. Other components include a main memory, an LSI, etc. Also, in terms of software, it is realized by a document mapping display program loaded into the main memory, etc.

When each functional part of the document mapping display device 1 is realized by software, the document mapping display device 1 is realized by executing the commands of a program, which is software that realizes each function. The recording medium that stores this program can be a "non-transient tangible medium," such as a CD, DVD, semiconductor memory, or programmable logic circuit. In addition, this program may be supplied to the document mapping display device 1 (computer) via any transmission medium (communications network, broadcast waves, etc.) that can transmit the program.

The various storage units in the document mapping display device 1 are RAM 12, ROM 13, and storage unit 14, which is a storage device such as an HDD or SSD. In addition, each functional unit that executes calculation processing is a calculation element such as a CPU 11. As shown in the block diagram of FIG. 3, the document mapping display device 1 has functional units such as a document acquisition unit 110, a text acquisition unit 120, a document placement unit 130, an extraction unit 140, a calculation unit 150, an output unit 160, a crawling unit 111, and a vectorization unit 131.

The I/O 15 is an interface, buffer, etc. for communication (transmission and reception). The I/O 15 is used for connection to an Internet line or for receiving input signals from a reading unit 4 (reader) such as a CD-ROM or DVD-ROM, and for sending output signals to a display unit 7, etc., and works in conjunction with the CPU 11. The display unit 7 is a known display (liquid crystal display device, organic EL display device, etc.). In addition, the display unit 7 may be a device with an image display function such as a tablet terminal or a smartphone. Furthermore, devices such as a keyboard 5 and a mouse 6 are connected to the I/O 15 as input devices.

The document mapping display device 1 (computer) common to the first and second embodiments has a function to visualize and display the degree of growth by plotting a distribution map on a two-dimensional plane based on multiple documents and a specified text, as described below.

First, the individual functional parts of the document mapping display device 1 (computer) of the first embodiment will be described in order with reference to the block diagram in FIG. 3.

The document acquisition unit 110 acquires multiple documents. When acquiring documents, if the documents are stored in fixed media 2 such as a CD-ROM or DVD-ROM, they can be acquired as data through the reading unit 4. Alternatively, if connected to an Internet line, the target documents can be received and acquired from an external server (not shown). After acquisition, the documents are temporarily stored in the memory unit 14.

The document acquisition unit 110 may include a crawling unit 111. The crawling unit 111 acquires information on websites existing on the Internet and creates a search database index. Then, the crawling unit 111 automatically accesses the website on which the target document exists and acquires the target document.
When crawling, information on the target document (the title of a paper, a publication such as a patent issued by a specific government agency) is input by a user of the document mapping display device 1. The document acquisition unit 110 then crawls the website based on the document information, acquires the HTML information of the relevant document, and acquires the relevant document.

The text acquisition unit 120 acquires specific text from multiple documents.

The document placement unit 130 places the specified sentences contained in each of the multiple documents on a two-dimensional plane according to their similarity. Furthermore, the placement on the two-dimensional plane in the document placement unit 130 according to the similarity between the specified sentences is based on natural language processing. The sentences present in the documents have fluctuations and inconsistencies in expression that are specific to the language. Therefore, it is desirable to use natural language processing to facilitate a comparison of the similarity between the specified sentences.

The similarity between given sentences here refers to the closeness of the meaning of the sentences.

The document placement unit 130 includes a vectorization unit 131. This vectorization unit 131 generates a feature vector for each of the multiple documents using natural language processing. That is, each document holds a feature vector. The feature vector can be displayed in two dimensions by reducing the number of dimensions. Therefore, each of the multiple documents is placed in a two-dimensional plane at a position corresponding to the feature vector that has been dimension-reduced from a reference point (not shown). The feature vector is generated using a series of models used to generate word embeddings such as Word2vec.

The feature vector generated by the document placement unit 130 has a high number of dimensions, up to several hundred. When the feature vector has such a high number of dimensions, it becomes very complicated to illustrate and visualize it. Therefore, the number of dimensions of the high-dimensional feature vector is reduced to two dimensions (dimensionality reduction). When reducing the number of dimensions, a known dimensionality reduction method such as pre-learning or UMAP processing that arranges the distribution based on the results of pre-learning is used. In other words, pre-learning is possible in reducing the number of dimensions, and when the learned results are used to execute each time, the distribution on the dimensionally reduced two-dimensional plane does not change significantly. This makes it possible to capture the amount of change over time. In addition, when pre-learning, it is desirable to comprehensively pre-learn predetermined sentences of multiple documents. If these are domestic patents, they may be pre-learned for all publications at a certain point in time. Furthermore, the pre-learning may be updated every few years, for example, like a United Nations survey.

The schematic diagram in Figure 4 shows an example of multiple documents in the first embodiment arranged on a two-dimensional plane 20. The documents that are the subject of document mapping are Japanese patents. In the diagram, one point corresponds to one document.

In the schematic diagram in Figure 4, multiple documents are displayed on a two-dimensional plane, visualizing the collection of documents. However, at the schematic diagram stage in Figure 4, the collection of documents is only displayed at a certain point in time.

The extraction unit 140 extracts the amount of change in the number of documents in a specified region on a two-dimensional plane. In addition, based on the amount of change in the number of documents in the specified region, the extraction unit 140 extracts the change in a specified set of sentences on the two-dimensional plane or the difference in the change in the specified set of sentences. The amount of change in the number of documents here may be the difference between the number of documents in a specified region in a certain year and the number of documents in the specified region in the previous year. The calculation unit 150 calculates the growth rate of the specified region based on the amount of change in the number of documents in the specified region on a two-dimensional plane. The growth rate refers to a feature (e.g., gradient, etc.) including a vector quantity calculated from the amount of change in documents.

In this embodiment, the extraction unit 140 divides the two-dimensional plane 20 (see FIG. 4) into a grid pattern, defines each section as a specific region, and extracts the amount of change in the number of documents based on the number of documents present in the specific region and the tag information of the documents. This is shown in the schematic diagram of FIG. 5.

The aforementioned specified area is a portion that is partitioned by defining a specified range to be analyzed from within the two-dimensional plane 20 shown in FIG. 5, etc., and is one (one) partition partitioned in a grid pattern (lattice pattern) as shown in FIG. 6 below.

In the diagram in Figure 4, each document is represented as a gray dot. The tag information mentioned above can be reflected in each document. For example, it is possible to color-code one international patent classification as tag information, "blue," another as "orange," and yet another as "green." This makes it possible to grasp trends in document distribution using the tag information as a clue.

As can be seen from FIG. 5(A), an area 21 to be analyzed is selected from a two-dimensional plane 20. The area 21 on the two-dimensional plane 20 is partitioned into a grid (lattice) shape by a grid 30 having a uniform, predetermined spacing 31. In this way, the area 21 to be analyzed is partitioned into a plurality of sections 32 (so-called squares) by the grid 30. Then, the amount of change in the number of documents present in each of the sections 32 (squares) is extracted.

In FIG. 5(B), the region 21 is represented as an x-y plane with the horizontal direction being the x-axis and the vertical direction being the y-axis. The numbers in the sections 32 in the figure represent the specific change in the number of documents. The predetermined interval 31 for the region 21 can be set arbitrarily. A wider predetermined interval 31 allows for a macroscopic understanding, and a narrower predetermined interval 31 allows for a microscopic understanding.

After the number of documents present in the multiple sections 32 (grids) generated by each grid 30 is extracted, the calculation unit 150 calculates the growth rate of the one section from the number of documents present in other sections adjacent to the one section. In other words, the characteristics of the one section are calculated based on the difference between the sections before, after, and above and below the one section, based on the relationship between the vertical and horizontal axes.

For example, when multiple documents from a certain year are arranged on a two-dimensional plane (see Figures 4 and 5(A) above), the specific number of documents for a certain year in a specified area is found (see Figure 5(B)). That is, the number of documents in a specified area is extracted based on the tag information for the year. Then, since multiple documents can be arranged on a two-dimensional plane for the previous year of the certain year, the specific number of documents for a certain area in the previous year of the certain year is found (see Figure 5(B)). That is, the amount of change in the number of documents is found for each year based on the tag information. Therefore, it is possible to compare (difference) the numerical values (so-called number of documents) for a certain year and the previous year for each section 32. This makes it easy to grasp the degree of change in the number of documents from the amount of difference.

The change in the number of documents from the difference amount described above is what is known as a grasp of the change in the number of documents from year to year (yearly unit). In addition to this, it is also possible to compare the difference amount between multiple years (average difference amount) and the difference amount between a certain year and the previous year. In addition to grasping the change in the number of documents from year to year, it is also possible to calculate the amount of change in the difference amount, such as upward and downward fluctuations, from the change in the number of documents over the past few years. This extraction is performed by the extraction unit 140. Then, the calculation unit 150 calculates the characteristics of a certain section based on the difference from the sections before and after and above and below the certain section, based on the relationship between the vertical and horizontal axes.

The schematic diagram in FIG. 6 shows how to calculate the gradient in one section. The numerical values in each section in the diagram are the amount of change in the number of documents. FIG. 6(A) shows the calculation of the x-axis, which is the horizontal direction. The amount of change in the number of documents in section 41 being calculated is "7", the amount of change in the number of documents in section 42 adjacent to the left is "5", and the amount of change in the number of documents in section 43 adjacent to the right is "9". In the case of the embodiment, the amount of change in the number of documents in section 42, "5", is subtracted from the amount of change in the number of documents in section 43, "9", to obtain the difference "4". The difference is multiplied by "1/2" to obtain "2". This "2" is the growth rate (dx) in the x-axis direction, which is the horizontal direction of section 41 being calculated. Note that both ends are forward or backward differences (not shown).

Figure 6 (B) shows the calculation of the y-axis, which is the vertical direction. The change in the number of documents in the section 41 being calculated is "7", the change in the number of documents in the adjacent section 44 above is "2", and the change in the number of documents in the adjacent section 45 below is "0". In the case of the embodiment, the change in the number of documents in section 45, "2", is subtracted from the change in the number of documents in section 44, "0", to obtain the difference "-2". The difference is multiplied by "1/2" to obtain "-1". This "-1" is the growth rate (dy) in the y-axis direction, which is the vertical direction of the section 41 being calculated. Note that both ends are forward or backward differences (not shown).

FIG. 6C is an example of a calculation for determining the overall growth rate of a certain section. This is a formula for calculating the overall growth rate (G) of a certain section from the growth rate (dx) in the horizontal direction (x direction) and the growth rate (dy) in the vertical direction (y direction) of the certain section. In other words, the square root is calculated for the sum of the squares of the growth rates in each direction. In the example of FIG. 6, 2 ² +(−1) ² = the square root of 5 (√(5)), which is about 2.24. Note that the calculation of the growth rate disclosed in FIG. 6 and the like is an example, and the calculation of the growth rate is not limited to the method shown and described.

The calculation unit 150 performs the calculations described in FIG. 6 for all sections. The schematic diagram in FIG. 7 is an example showing the growth degree in each section. FIG. 7(A) shows the growth degree (dx) on the x-axis, which is the horizontal direction, in all sections, and FIG. 7(B) shows the growth degree (dy) on the y-axis, which is the vertical direction, in all sections. FIG. 7(C) is a list of the growth degrees (G) in all sections.

As mentioned above, three steps are executed in sequence: (i) selecting an area to be analyzed from within a two-dimensional plane, (ii) creating individual sections by setting a grid in that area, and (iii) calculating the growth rate of each section. Then, for a particular section (e.g. section 41 mentioned above), it becomes possible to calculate the growth rate (dx) in the x-axis direction, which is the horizontal direction, and the growth rate (dy) in the y-axis direction, which is the vertical direction, of that section.

Then, the output unit 160 outputs the growth degree. The output of the growth degree is displayed as an image on the display unit 7 (display) of FIG. 1. FIG. 8 is an example of a display on the display unit 7, and in FIG. 8(A), the growth degree is displayed as an arrow on a two-dimensional plane.

In the example of FIG. 8(A), two types of arrows are provided, a black arrow 22 for a growth level above a given threshold, and a hollow arrow 23 for a growth level above another given threshold. In FIG. 8(B), the growth level is displayed numerically for each individual document, in the illustrated example in relation to the document number and the inventor.

In the case of FIG. 8(A), the locations and colors of the arrows 22 and 23 make it possible to objectively grasp areas with high growth rates on a two-dimensional plane. In addition, in the case of FIG. 8(B), the growth rate for each document can be specifically grasped as a numerical value.

Next, the document mapping display method of the first embodiment will be explained together with the document mapping display program.

The literature mapping display method of the first embodiment is executed by the CPU 11 of the literature mapping display device 1 based on the literature mapping display program of the first embodiment. The literature mapping display method causes the CPU 11 of the literature mapping display device 1 to execute a literature acquisition function, a text acquisition function, a literature arrangement function, an extraction function, a calculation function, an output function, and further a crawling function. Each function overlaps with the above description, so details are omitted.

The flowcharts in Figures 9, 10, and 11 show the overall flow of the document mapping display method in the CPU 11 of the document mapping display device 1 of the first embodiment. In Figure 9, a document acquisition step (S110), a text acquisition step (S120), a document placement step (S130), an extraction step (S140), a calculation step (S150), and an output step (S160) are executed, while in Figure 10, a crawling step (S111) is executed. In Figure 11, a vectorization step (S131) is executed.

The document acquisition function acquires a plurality of documents (S110; document acquisition step). The text acquisition function acquires a predetermined text from the plurality of documents (S120; text acquisition step). The document arrangement function arranges the plurality of documents on a two-dimensional plane according to the similarity of the predetermined text contained in each of the plurality of documents (S130; document arrangement step). The extraction function extracts the amount of change in the number of documents in a predetermined area present on the two-dimensional plane (S140; extraction step). Furthermore, the extraction function extracts the change in the set of predetermined texts in the two-dimensional plane, or the difference in the change in the set of predetermined texts, based on the amount of change in the number of documents in the predetermined area. The calculation function calculates the growth rate based on the amount of change in the number of documents in a predetermined area present on the two-dimensional plane (S150; calculation step). The output function outputs the growth rate (S160; output step). The crawling function acquires a plurality of documents through an Internet line (S111; crawling step) (see FIG. 10). The vectorization function generates a feature vector for each of the plurality of documents by natural language processing (S131; vectorization step) (see FIG. 11).

Next, the individual functional units of the document mapping display device 1 (computer) of the second embodiment will be described in order with reference to the block diagram of FIG. 12 and the like. The mechanical configuration of the document mapping display device 1 of the second embodiment is the same as that of the document mapping display device 1 of the first embodiment, and therefore the description will be omitted. As shown in the block diagram of FIG. 12, the document mapping display device 1 of the second embodiment includes functional units such as a document acquisition unit 110, a text acquisition unit 120, a document arrangement unit 130, an extraction unit 140, a calculation unit 150, an accumulation detection unit 170, a time series change output unit 180, a crawling unit 111, and a vectorization unit 131. Note that in the second embodiment, the same reference numerals are used for components common to the document mapping display device 1 of the first embodiment, and duplicated descriptions will be omitted.

In the block diagram of FIG. 12, the document acquisition unit 110 acquires multiple documents. The function of the document acquisition unit 110 is the same as in the first embodiment. The function of the crawling unit 111 included in the document acquisition unit 110 is the same as in the first embodiment. The text acquisition unit 120 acquires a predetermined text from multiple documents. The function of the text acquisition unit 120 is the same as in the first embodiment.

The document arrangement unit 130 of the second embodiment arranges multiple documents on a two-dimensional plane according to the similarity of specific sentences contained in each of the multiple documents. The arrangement on the two-dimensional plane in accordance with the similarity of specific sentences contained in the documents by the document arrangement unit 130 of the second embodiment is based on natural language processing. Words and the like present in the sentences of the documents have fluctuations and inconsistencies in expression that are specific to the language. Therefore, in order to facilitate a comparison of the similarity between the specific sentences, it is desirable to use natural language processing. The similarity between specific sentences contained in the specific documents referred to here refers to the closeness of the meaning of the documents.

The document arrangement unit 130 also includes a vectorization unit 131. This vectorization unit 131 generates a feature vector for each of the multiple documents by natural language processing. The function of the vectorization unit 131 is the same as in the first embodiment. The feature vectors generated by the document arrangement unit 130 also have their number of dimensions reduced to two dimensions (dimensional compression) as in the first embodiment. Therefore, it is possible to display them on a two-dimensional plane as shown in the schematic diagram of FIG. 13.

The schematic diagram in FIG. 13 is an example of a plurality of documents in the second embodiment arranged on a two-dimensional plane 25. The gray parts in the diagram correspond to collections of individual documents. The division by the shade of gray indicates the rough boundaries of technical fields. The actual display is a color display of multiple different colors, and is considered as a collection of points. For convenience, the illustration shows different shades of gray.

The extraction unit 140 extracts the amount of change in the number of documents in a specified area on the two-dimensional plane 25. The calculation unit 150 calculates the growth rate based on the amount of change in the number of documents in a specified area on the two-dimensional plane. The functions of the extraction unit 140 and the calculation unit 150 in the document mapping display device 1 of the second embodiment are the same as those of the first embodiment, and the same processing as that described above in Figures 5, 6, 7, and 8 is executed.

The accumulation detection unit 170 detects accumulation areas based on a plurality of documents in a specified area on a two-dimensional plane (clustering). Furthermore, the accumulation detection unit 170 detects accumulation areas based on the density of a plurality of documents in a specified area on a two-dimensional plane (clustering). When detecting, density-based clustering methods, k-means clustering, k-nearest neighbor clustering, etc. are used for a plurality of documents arranged on a two-dimensional plane. The number of accumulation areas is specified by the user's settings on the document mapping display device 1. This makes it possible to perform analysis according to the number of accumulation areas that meets the user's needs.

The time series change output unit 180 displays on a two-dimensional plane the degree of growth in a specified area and the time series change points in the two-dimensional plane of the accumulation area. More specifically, it displays on the two-dimensional plane 25 the degree of growth in a specified area present on the two-dimensional plane 25 and the time series change points in the two-dimensional plane 25 of the accumulation area of multiple documents. The display format is as shown in Figures 14 and 15 below.

The time series change output unit 180 calculates and then outputs the time series change points for each accumulation area (each cluster) detected by the accumulation detection unit 170. The time series change points for each accumulation area (each cluster) are calculated using the multiple documents belonging to each accumulation area (each cluster) detected by the accumulation detection unit 170 out of the multiple documents in a specific area present on the two-dimensional plane 25.

Furthermore, the time series change output unit 180 calculates the dense positions of the documents belonging to each cluster for each period, and connects the multiple periods to display the change points in the time series. A Gaussian distribution or the like is used to calculate the dense positions of the documents.

Specifically, as shown in the schematic diagram of FIG. 14, shapes such as circles are used to emphasize locations where many documents are concentrated (dense). These circles are displayed at various points on a two-dimensional plane 25.

Figure 14 shows an example of a display format for three years. Each accumulation area (area divided by shades of gray) contains three circles. For example, each circle corresponds to a location where multiple documents are densely packed this year, one year ago, and two years ago in the accumulation area in the two-dimensional plane 25. Since the figure shows an example for three years, there are three circles. Therefore, if it is a cumulative survey for five years, the number of circles is increased to five. Also, shapes other than circles (such as squares) may be used for each year (period). Furthermore, each circle is connected by a line in chronological order. With the display shown in Figure 14, both the locations where multiple documents are densely packed and the changes in the locations over time can be displayed together on the two-dimensional plane 25, making it easy to visually grasp. Of course, the period of display is not limited to the one shown in the figure and can be any appropriate period. For example, it may be every two years.

For more details, see the enlarged schematic diagram of FIG. 15, which is a partial enlargement of FIG. 14. FIG. 15 shows the changes over time for the past three years, 2018, 2019, and 2020. Circles 28a are displayed at locations where there is a high density of documents in 2018, circles 28b at locations where there is a high density of documents in 2019, and circles 28c at locations where there is a high density of documents in 2020. In order to clarify the display of the changes over time, circles 28a, 28b, and 28c are displayed in increasing gray in that order. The positions of the circles indicating the locations where such multiple documents are dense make it easier to understand the movement (locus of position) on the two-dimensional plane 25.

Furthermore, circles 28a and 28b, which indicate the location where multiple documents are densely packed, are connected by line 29p, and circles 28b and 28c are connected by line 29q. By arranging the lines, the length of the lines (the distance between the circles) is made clear, making it easier to grasp the amount of movement in two-dimensional plane 25 (whether there is a large movement or whether the position remains the same). In the illustration, the lines themselves are also made to be darker in gray in the order of line 29p, then line 29q, in order to clearly show the change over time.

In the example of FIG. 15, two types of arrows are provided, and when the growth level is above a predetermined threshold, a black arrow 26 is displayed, and when the growth level is above another predetermined threshold, a white arrow 27 is displayed. As in the first embodiment, the locations and colors of the arrows 26 and 27 make it possible to objectively grasp areas of high growth on the two-dimensional plane 25. Note that the types of arrows are not limited to the two types shown. In the illustration, the arrows 26 and 27 are shown as triangles. This is for the convenience of simplifying the display on the two-dimensional plane 25.

Next, the literature mapping display method of the second embodiment will be explained together with the literature mapping display program.

The literature mapping display method of the second embodiment is executed by the CPU 11 of the literature mapping display device 1 based on the literature mapping display program of the second embodiment. The literature mapping display method causes the CPU 11 of the literature mapping display device 1 to execute a literature acquisition function, a text acquisition function, a literature arrangement function, an extraction function, a calculation function, an accumulation detection function, and a time series change output function, and further executes a crawling function. Each function overlaps with the above description, so details will be omitted.

The flowcharts in Figures 16, 17, and 18 show the overall flow of the document mapping display method in the CPU 11 of the document mapping display device 1 of the second embodiment. In Figure 16, a document acquisition step (S110), a text acquisition step (S120), a document arrangement step (S130), an extraction step (S140), a calculation step (S150), an accumulation detection step (S170), and a time series change output step (S180) are executed, and in Figure 17, a crawling step (S111) is executed. In Figure 18, a vectorization step (S131) is executed.

The document acquisition function acquires a plurality of documents (S110; document acquisition step). The text acquisition function acquires a predetermined text from the plurality of documents (S120; text acquisition step). The document arrangement function arranges the plurality of documents on a two-dimensional plane according to the similarity of the predetermined text contained in each of the plurality of documents (S130; document arrangement step). The extraction function extracts the amount of change in the number of documents in a predetermined area present on the two-dimensional plane (S140; extraction step). Furthermore, the extraction function extracts the change in the set of predetermined texts in the two-dimensional plane, or the difference in the change in the set of predetermined texts, based on the amount of change in the number of documents in the predetermined area. The calculation function calculates the growth degree based on the amount of change in the number of documents in a predetermined area present on the two-dimensional plane (S150; calculation step). The accumulation detection function detects an accumulation area based on the plurality of documents in a predetermined area present on the two-dimensional plane (based on the density of the plurality of documents) (S170; accumulation detection step). The time series change output function displays the growth degree in the predetermined area and the time series change points in the two-dimensional plane of the accumulation area on the two-dimensional plane (S180; time series change output step). The crawling function also acquires multiple documents via an Internet line (S111; crawling step) (see FIG. 17). The vectorization function generates feature vectors for each of the multiple documents through natural language processing (S131; vectorization step) (see FIG. 18).

The computer program of the present invention described above may be recorded on a processor-readable recording medium, and the recording medium may be a "non-transitory tangible medium" such as a disk, card, semiconductor memory, or programmable logic circuit.

The computer program can be implemented using, for example, a scripting language such as ActionScript or JavaScript (registered trademark), an object-oriented programming language such as Objective-C or Java (registered trademark), or a markup language such as HTML5.

1 Literature mapping display device 2 Fixed media 3 Internet line 4 Reading unit 5 Keyboard 6 Mouse 7 Display unit (display)
11 CPU
12 RAM
13 ROM
14 Memory unit 15 I/O (input/output interface)
20, 25 Two-dimensional plane 22, 23, 26, 27 Arrows 28a, 28b, 28c Circles 29p, 29q Lines 30 Grid 31 Grid spacing 32, 41, 42, 43, 44, 45 Sections 110 Document acquisition section 111 Crawling section 120 Text acquisition section 130 Document placement section 131 Vectorization section 140 Extraction section 150 Calculation section 160 Output section 170 Accumulation detection section 180 Time series change output section

Claims (20)

a document acquisition unit for acquiring a plurality of documents;
a text acquisition unit for acquiring a predetermined text from the plurality of documents;
a document arrangement unit that arranges the plurality of documents on a two-dimensional plane according to similarities between predetermined sentences included in each of the plurality of documents;
an extraction unit that extracts a time series change in the number of documents in a predetermined area existing on the two-dimensional plane;
a calculation unit that calculates a growth rate based on a time series change in the number of documents in a predetermined area present on the two-dimensional plane;
and an output unit that outputs the growth degree. a document acquisition unit for acquiring a plurality of documents;
a text acquisition unit for acquiring a predetermined text from the plurality of documents;
a document arrangement unit that arranges the plurality of documents on a two-dimensional plane according to similarities between predetermined sentences included in each of the plurality of documents;
an extraction unit that extracts a time series change in the number of documents in a predetermined area existing on the two-dimensional plane;
a calculation unit that calculates a growth rate based on a time series change in the number of documents in a predetermined area present on the two-dimensional plane;
an accumulation detection unit that detects an accumulation area based on the plurality of documents in a predetermined area existing on the two-dimensional plane;
a time series change output unit that displays, on a two-dimensional plane, a growth rate in a predetermined area and time series change points in the two-dimensional plane of the accumulation area. The document mapping display device according to claim 1 or 2, wherein the document acquisition unit includes a crawling unit and acquires the plurality of documents via an Internet line. tag information indicating characteristics of the document is attached to each of the plurality of documents in accordance with the document;
The document mapping display device according to claim 3 , wherein the document acquisition section acquires the plurality of documents based on the tag information. The document mapping display device according to claim 1 or 2, wherein the document placement unit places the documents on a two-dimensional plane according to the similarity between the specified sentences based on natural language processing. The document mapping display device according to claim 5, wherein the document placement unit includes a vectorization unit that generates a feature vector for each of the plurality of documents by the natural language processing. The document mapping display device according to claim 1 or 2, wherein the extraction unit extracts the number of documents present within a given region, the given region being one of the sections that is generated by dividing the two-dimensional plane into a grid shape. The document mapping display device according to claim 1 or 2, wherein the extraction unit extracts a time-series change in the number of documents based on a comparison of the number of documents in a predetermined area present on the two-dimensional plane. The document mapping display device according to claim 1 or 2, wherein the extraction unit extracts changes in a set of predetermined sentences on a two-dimensional plane. The document mapping display device according to claim 7 , wherein the calculation unit calculates the growth rate of the one section from a time series change in the number of documents present in another section adjacent to the one section. The document mapping display device according to claim 1, wherein the output unit displays the degree of growth on the two-dimensional plane using a type of arrow. The document mapping display device according to claim 1, wherein the output unit displays the growth rate for each document as a numerical value. The document mapping display device according to claim 2, wherein the accumulation detection unit detects an accumulation area based on the density of the plurality of documents in a predetermined area present on the two-dimensional plane. The document mapping display device according to claim 2, in which any number of accumulation areas can be specified for the accumulation areas. The document mapping display device according to claim 2, wherein the time series change output unit displays time series change points in the accumulation area of the plurality of documents on the two-dimensional plane as circles and displays the circles as lines connecting each other. The document mapping display device according to claim 2, wherein the time series change output unit displays the degree of growth on the two-dimensional plane using a type of arrow. The computer
A document acquisition step of acquiring a plurality of documents;
a sentence acquisition step of acquiring a predetermined sentence from the plurality of documents;
a document arrangement step of arranging the plurality of documents on a two-dimensional plane according to similarities between predetermined sentences included in each of the plurality of documents;
an extraction step of extracting a time series change in the number of documents in a predetermined area on the two-dimensional plane;
a calculation step of calculating a growth rate based on a time series change in the number of documents in a predetermined area present on the two-dimensional plane;
and outputting the growth degree. On the computer,
A document acquisition function that acquires multiple documents;
A text acquisition function for acquiring a specified text from a plurality of documents;
a document arrangement function for arranging the plurality of documents on a two-dimensional plane in accordance with similarities between predetermined sentences contained in each of the plurality of documents;
an extraction function for extracting a time series change in the number of documents in a predetermined area on the two-dimensional plane;
a calculation function for calculating a growth rate based on a time series change in the number of documents in a predetermined area present on the two-dimensional plane;
and an output function for outputting the growth degree. The computer
A document acquisition step of acquiring a plurality of documents;
a sentence acquisition step of acquiring a predetermined sentence from the plurality of documents;
a document arrangement step of arranging the plurality of documents on a two-dimensional plane according to similarities between predetermined sentences included in each of the plurality of documents;
an extraction step of extracting a time series change in the number of documents in a predetermined area on the two-dimensional plane;
a calculation step of calculating a growth rate based on a time series change in the number of documents in a predetermined area present on the two-dimensional plane;
an accumulation detection step of detecting an accumulation region based on the plurality of documents in a predetermined region existing on the two-dimensional plane;
and a time series change output step of displaying on a two-dimensional plane the degree of growth in a predetermined region and time series change points in the two-dimensional plane of the accumulation region. On the computer,
A document acquisition function that acquires multiple documents;
A text acquisition function for acquiring a predetermined text from the plurality of documents;
a document arrangement function for arranging the plurality of documents on a two-dimensional plane in accordance with similarities between predetermined sentences contained in each of the plurality of documents;
an extraction function for extracting a time series change in the number of documents in a predetermined area on the two-dimensional plane;
a calculation function for calculating a growth rate based on a time series change in the number of documents in a predetermined area present on the two-dimensional plane;
an accumulation detection function for detecting an accumulation region based on the plurality of documents in a predetermined region existing on the two-dimensional plane;
and a time series change output function for displaying on a two-dimensional plane the degree of growth in a predetermined area and time series change points in the two-dimensional plane of the accumulation area.

Images