JP2016040723A - Article arrangement determination device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an article arrangement determination device capable of properly arranging content in accordance with article importance.SOLUTION: Virtual article area generating means 2 generates a virtual article area having area based on importance 6 of each article recorded in a recording part 4. Virtual article area arranging means 10 arranges the generated virtual article area on a virtual plane. Virtual spring setting means 12 sets a virtual spring having a spring constant corresponding to degree of association between articles recorded in the recording part 4 between corresponding virtual article areas. Virtual article area arrangement determining means 16 changes and determines arrangement of the virtual article areas arranged on the virtual plane so as to make total power by power total calculating means small. Article area arrangement determining means 18 determines the arrangement of the article areas on the basis of the determined arrangement of the virtual article areas.SELECTED DRAWING: Figure 1a

Description

  The present invention relates to an apparatus that automatically and appropriately arranges a plurality of articles according to their importance.

  Browsing the content recorded in the server device on the terminal device via the Internet. At this time, terminal devices used for browsing include PCs, smartphones, tablets, and the like, and the screens have various sizes.

  For this reason, if you browse news articles that are optimally arranged for devices with a large screen such as a PC on a device that has only a small screen such as a smartphone, the entire article may not be displayed or it may be extremely There was a problem that the characters had to be made smaller.

  Further, even if the terminal device to be browsed is determined, it is troublesome for a human to manually determine the article layout.

  Therefore, an apparatus for automatically arranging contents has been proposed. For example, Patent Document 1 discloses an apparatus that automatically rearranges content. With this device, it is possible to search for content corresponding to the theme of the page and rearrange it.

JP 2002-318143 A

  However, the apparatus of Patent Document 1 performs automatic rearrangement when it is desired to replace an article once arranged, and does not determine the layout from the beginning. Also, it was not possible to automatically determine an appropriate article arrangement according to the importance of the article.

  An object of the present invention is to solve the above-described problems and to provide an article arrangement determination device that can appropriately arrange contents according to the importance of articles.

(1) (2) An article placement determination device according to the present invention is an article placement determination device that places a plurality of articles on a plane, and records a plurality of articles and the importance of each article, and between each article. A recording section that records the degree of relevance of the article, a virtual article area generating unit that generates a virtual article area corresponding to each article and having an area corresponding to the importance of each article, and a spring corresponding to the degree of association between the articles Virtual spring setting means for setting a virtual spring having a constant between the corresponding virtual article areas; virtual article area arrangement determining means for determining the arrangement of the virtual article areas so that the total attractive force of the virtual springs is reduced; An article area arrangement determining means for determining the arrangement of the article area based on the determined arrangement of the virtual article area.

  Therefore, based on the total force of the virtual springs, a preferable article arrangement can be determined from the viewpoint of importance and relevance.

(3) The article arrangement determining apparatus according to the present invention further includes force total calculation means for calculating a total value of attractive forces of virtual springs arranged between the virtual article areas arranged on the virtual plane, the virtual article area The placement determination means includes a virtual article area placing means for placing the virtual article area on the virtual plane, and a virtual article area that changes the placement of the virtual article area on the virtual plane so that the total value of the attractive force of the virtual spring is reduced. Virtual article region arrangement change determining means for determining the arrangement of the article.

  Accordingly, it is possible to select an appropriate arrangement by changing the arrangement and looking at the change in the total force of the spring.

(4) The article arrangement determining apparatus according to the present invention is characterized in that the virtual article region generating means generates a circle having an area corresponding to the importance of each article as the virtual article region.

  Therefore, the article arrangement simulation can be easily performed.

(5) The article placement determining device according to the present invention is characterized in that the virtual article region placement means places the virtual article region having the highest importance fixed at the upper right or upper left on the plane on which the article is placed. Yes.

  Therefore, articles with a high degree of importance can be arranged in places with a high degree of attention.

(6) The article arrangement determining apparatus according to the present invention is characterized in that the virtual article area arranging means arranges the virtual article areas in order from a pair of virtual article areas having a high degree of association.

  Therefore, articles with a high degree of relevance can be arranged without being separated.

(7) The article arrangement determining apparatus according to the present invention is characterized in that the virtual spring setting means sets the virtual spring so that no attractive force is generated between the virtual article regions that are in contact with each other. .

  Therefore, an attractive force is applied when trying to separate both virtual article regions, and an appropriate arrangement can be selected by reducing the total size of the attractive force.

(8) The article placement determining apparatus according to the present invention is characterized in that the virtual spring setting means increases the spring constant as the degree of association between articles increases.

  Therefore, it is possible to perform a simulation so that an article having a higher degree of relevance has a greater attractive force when it is released.

(9) The article arrangement determining apparatus according to the present invention is characterized in that the virtual article area arrangement change determining means changes the arrangement by paying attention to a virtual article area related to a virtual spring having the greatest attraction.

  Therefore, there is a high possibility that the total force is reduced by the arrangement change.

(10) (11) An article placement determination device according to the present invention is an article placement determination device that places a plurality of articles on a plane, and records a plurality of articles and the importance of each article, and between each article. A recording section that records the degree of relevance of the article, a virtual article area generating unit that generates a virtual article area corresponding to each article and having an area corresponding to the importance of each article, and at least the most generated virtual article area A virtual article area having a large importance is fixedly arranged so as not to overlap each other on the virtual plane, and other virtual article areas are allowed to overlap each other on the virtual plane. Virtual article region placement means as a movable virtual article region placed far away, and virtual spring setting means for setting a virtual spring having a spring constant according to the degree of association between each article between the corresponding virtual article regions; , The movable virtual article region arranged on the virtual plane is moved toward the fixed virtual article region in accordance with the set attractive force of the virtual spring, and the movable virtual article region is moved to the fixed virtual article region. The article area arrangement determining means for determining the arrangement of the virtual article area by making the movable virtual article area a fixed virtual article area.

  Therefore, an appropriate arrangement can be easily realized without changing the arrangement.

(12) The article arrangement determining device according to the present invention analyzes an article to be analyzed, extracts at least a noun constituting the article as a word, and each of the articles included in the article extracted by the word extracting means Importance calculating means for calculating the importance of each word and calculating the importance of each article based on the importance of each word.

  Therefore, the importance of an article can be calculated based on the importance of a word.

(13) The article arrangement determining apparatus according to the present invention is characterized by further comprising a relevance calculation means for calculating a relevance between articles based on words included in each article.

  Therefore, the degree of association between both articles can be calculated based on the word.

(14) In the article arrangement determining device according to the present invention, the degree-of-relevance calculating means determines important words from the top with high importance in each article as important words,
Relevance between words that search for important words contained in one article for calculating the degree of association and important words contained in the other article for which the degree of association is calculated on a search site on the web and obtain the number of hits The relevance obtained by the inter-word relevance calculating means is changed by changing the combination of the inter-word relevance calculating means, the important word included in the one article, and the important word included in the other article. And an inter-article relevance calculating means for calculating the relevance between both articles.

  Therefore, the degree of association between articles can be calculated using the search site.

(15) The article arrangement determining method according to the present invention is a method of determining a plurality of article arrangements on a plane by a computer, the computer corresponding to each article, and having an area corresponding to the importance of each article. A virtual article region having the virtual article region is arranged on a virtual plane, and a virtual spring having a spring constant according to the degree of association between each article is set between the corresponding virtual article regions, The placement of the virtual article area on the virtual plane is changed, the placement of the virtual article area is determined such that the total attractive force of the virtual spring is reduced, and the placement of the article area is determined based on the determined placement of the virtual article area. It is characterized by deciding.

  Therefore, based on the total force of the virtual springs, a preferable article arrangement can be determined from the viewpoint of importance and relevance.

  In the embodiment, “virtual article region generation means” corresponds to step S71.

  In the embodiment, the “virtual article region arrangement determining unit” corresponds to steps S22 to S32, S40, and S41.

  In the embodiment, the “virtual article area arranging means” corresponds to the CPU 34 that executes a process of randomly arranging the virtual article areas.

  In the embodiment, the “virtual article area arrangement change determining unit” corresponds to the CPU 34 that executes the process of changing the arrangement of the virtual article areas.

  In the embodiment, “virtual spring setting means” corresponds to step S22.

  In the embodiment, the “total force calculation unit” corresponds to the CPU 34 that executes the process of totaling the attractive forces of the springs.

  In the embodiment, “article region arrangement determining means” corresponds to step S33.

  In the embodiment, “word extracting means” corresponds to step S2.

  In the embodiment, “importance calculation means” corresponds to steps S3 to S8.

  In the embodiment, “relevance calculation means” corresponds to steps S10 to S18.

The “program” is a concept that includes not only a program that can be directly executed by the CPU, but also a source format program, a compressed program, an encrypted program, and the like.

It is a functional block diagram of the article arrangement | positioning determination apparatus by one Embodiment of this invention. It is a functional block diagram of an importance / relevance calculation device according to an embodiment of the present invention. This is a hardware configuration of an article area arrangement device and an importance / relevance calculation device. It is a flowchart of the importance / relevance calculation program. It is a flowchart of the importance / relevance calculation program. It is a flowchart of a morphological analysis program. It is a figure which shows the result of a morphological analysis. It is a figure which shows the result of a morphological analysis. It is a figure which shows the process which calculates importance. It is a figure which shows the importance of each calculated article. It is a flowchart of an IDF value calculation process. It is a flowchart of a Hot coefficient calculation process. It is a figure which shows the reply data from a dictionary site. It is a coefficient table | surface for calculating a Hot coefficient. It is a figure which shows the important word of articles K1 and K2. It is a figure which shows the calculated relevance degree. It is a flowchart of an article arrangement | positioning determination program. It is a flowchart of an article arrangement | positioning determination program. It is a flowchart of an article arrangement | positioning determination program. It is a table of relevance arranged in descending order. It is a figure for demonstrating the arrangement | positioning procedure of a virtual article area | region. It is a figure for demonstrating the arrangement | positioning procedure of a virtual article area | region. It is a figure for demonstrating the arrangement | positioning procedure of a virtual article area | region. It is a figure for demonstrating the arrangement | positioning procedure of a virtual article area | region. It is a figure for demonstrating the arrangement | positioning procedure of a virtual article area | region. It is a figure for showing a procedure which transforms a virtual article field into an article field. It is a figure which shows the determined article arrangement | positioning. It is a figure which shows a virtual spring. It is a figure which shows arrangement | positioning determination of the virtual article area | region by other embodiment. It is a flowchart of a summary sentence generation program. It is a flowchart of a summary sentence generation program. It is a figure which shows the example of an article. It is a figure which shows the article decomposed | disassembled into the sentence. It is a figure which shows the importance calculated about each sentence. It is a figure which shows the produced | generated summary sentence.

1. Functional Block Diagram of Article Placement Determination Device FIG. 1a shows a functional block diagram of an article placement determination device according to an embodiment of the present invention. The virtual article region generating unit 2 generates a virtual article region having an area based on the importance 6 of each article recorded in the recording unit 4. The virtual article area arrangement unit 10 arranges the generated virtual article area on a virtual plane.

  The virtual spring setting means 12 sets a virtual spring having a spring constant corresponding to the degree of association between articles recorded in the recording unit 4 between corresponding virtual article regions. The virtual article area arrangement change determining unit 16 changes and determines the arrangement of the virtual article areas arranged on the virtual plane so that the total force by the force total calculating unit is reduced. The article area arrangement determining means 18 determines the arrangement of the article area based on the determined arrangement of the virtual article area.

FIG. 1b shows a functional block diagram of an importance / relevance calculation device used with the article arrangement determination device of FIG. 1a. The word extraction unit 22 extracts words included in each article 20 recorded in the recording unit 4. The importance calculation means 24 calculates the importance of the extracted word, and calculates the importance of each article based on the importance of each word. The calculated importance of each article is recorded in association with each article. The degree-of-association calculating means 26 calculates the degree of association between articles based on the importance of words included in each article. For example, a search is performed on a search site by combining words having high importance in both articles, and the relevance is determined based on the number of hits by the combination of the words. The importance level of articles and the degree of association between articles calculated as described above are recorded in the recording unit 4 in FIG. 1a.

2. Hardware Configuration FIG. 2 shows a hardware configuration of the article arrangement determination device and the importance / relevance calculation device shown in FIGS. 1a and 1b. A memory 36, a display 38, a hard disk 40, a CD-ROM drive 42, a keyboard / mouse 44, and a communication circuit 46 are connected to the CPU 34. The communication circuit 46 is for connecting to the Internet or the like.

On the hard disk 40, an operating system 50 such as WINDOWS (trademark), an importance / relevance calculation program 52, an article placement determination program 53, a morpheme dictionary 54, and a plurality of articles 20 are recorded. The operating system 50, the importance / relevance calculation program 52, and the article arrangement determination program 53 are installed on the CD-ROM 48 via the CD-ROM drive 42. The morpheme dictionary 54 records words and their parts of speech in association with each other.

3. Importance / Relevance Calculation Program Flow charts of the importance / relevance calculation program 52 are shown in FIGS. The CPU 34 sets the first article 20 among the articles 20 recorded on the hard disk 40 as a target article (step S1).

  The CPU 34 performs morphological analysis on all sentences included in the article (step S2). FIG. 5 shows details of the morphological analysis process. First, the CPU 34 refers to the morpheme dictionary 54 recorded on the hard disk 40 and analyzes the morpheme of the article (step S21). This morphological analysis can be performed by using a general method that has been conventionally performed. The result of morphological analysis is shown in FIG.

  Next, the CPU 34 uses the communication circuit 46 to access an encyclopedia or dictionary (for example, Wikipedia (trademark) or e-Words (trademark)) on the web via the Internet. Then, continuous words obtained by the morphological analysis shown in FIG. 6 are combined and searched for in an encyclopedia, dictionary, or search site on the web (step S22). For example, “plain” and “ha” are combined to generate the word “plain is” and whether this word is found as an entry word in an encyclopedia or dictionary on the web, or more than a predetermined number ( (For example, 10,000 or more) Search whether it hits. For example, “Plain + Ha”, “Extraordinary +”, “IT + Solution”, etc. are often used as clauses on the web, so it is possible to obtain more than a predetermined number of hits in a search on a search site. In addition, words such as “electronic + book” may exist as headwords in encyclopedias and dictionaries.

  In this embodiment, the first word of the sentence is used as the target word, the subsequent words are sequentially combined, and the longest word exists in the encyclopedia / dictionary on the web, or a predetermined number or more on the search site. Extract hits. The CPU 34 records the combined words (for example, “plain”) extracted in this way in the table of FIG. 6, and each word (for example, “plain” “ha”) constituting the combined word in FIG. 6. Delete from the table. As a result, the table becomes as shown in FIG. Further, the CPU 34 additionally registers the found combined word in the morpheme dictionary 54.

  Next, the next word after the extracted word (or the next word after the target word if it is not extracted) is used as the next target word, and subsequent words are sequentially combined, and the dictionary on the web as the longest word Extract what exists in. This process is repeated until the end of the sentence.

  Therefore, according to this embodiment, even a word that is not registered in the morpheme dictionary 54 can be extracted as a word.

  Subsequently, the CPU 34 performs processing for combining the words obtained above (step S23). For example, when nouns are continuous, they are combined into one noun. If the four nouns “Fukushima”, “Daily”, “Nuclear Power”, and “Power Plant” are consecutive, this is designated as “Fukushima Daiichi Nuclear Power Plant”. Similarly, if the surname and first name of a person name are consecutive, they are combined. In addition, words are combined using a method generally used for combining morphemes. The CPU 34 records the obtained combined word (for example, “Fukushima Daiichi Nuclear Power Station”) in the table of FIG. 7 and also makes individual words (for example, “Fukushima”, “First”, “ "Nuclear power" "Power plant") is deleted from the table of FIG.

  As described above, morphological analysis can be performed.

(3) TF Value Calculation Next, the CPU 34 calculates a TF (Term Frequency) value for the word extracted above (in this embodiment, only words including nouns are targeted) (step S3). That is, the TF value is calculated for a word whose part of speech recorded in the table of FIG. 7 is a noun and a word including the noun as a part of speech.

  Here, the TF value is a value obtained by dividing the number of times the word appears in the sentence by the total number of words in the sentence.

TF = number of appearances of words / total number of appearances of each word FIG. 8a shows the TF value calculated for each word. The CPU 34 records the calculated TF value on the hard disk 40.

  In this embodiment, by obtaining the appearance frequency of each word in the sentence, how much each word is emphasized in the sentence is calculated.

(4) IDF value calculation Next, the CPU 34 calculates an IDF (Inverse Document Frequency) value for each word (step S4). Here, the IDF value represents the rarity of each word (that is not common as a word circulating in the world). The higher the rarity of the word, the higher the IDF value.

  In this embodiment, the IDF value is calculated by the following equation.

IDF = log (number of hits (number of pages) in the search engine of the word / total number of pages on the search engine)
FIG. 9 shows details of the IDF value calculation process. The CPU 34 sets the first word recorded in the table of FIG. 7 (in this embodiment, only a word including a noun is targeted) as a target word (step S41). The CPU 34 searches for a target word (for example, “Peace is”) on a search site (for example, Yahoo! (trademark) or google (trademark)) on the Internet (step S42).

  In response to this search request, the search result is returned from the search site, so the CPU 34 receives it. The CPU 34 acquires the number of hits of the target word included in the search results (the number of pages including the word) and the total number of pages targeted by the search site, and records them in the memory 36 (step S43).

  Subsequently, the CPU 34 calculates log (number of hits (number of pages) in the search engine of the word / total number of pages on the search engine) and calculates an IDF value (step S44). The CPU 34 records the calculated IDF value in the hard disk 40 in association with each word (see FIG. 8a).

  The CPU 34 determines whether or not IDF values have been calculated for all words (step S45). If an unprocessed word remains in the table of FIG. 7, the CPU 34 sets the next word as a target word (step S46). Then, step S42 and subsequent steps are repeated. In this way, IDF values are calculated for all target words.

(5) TF-IDF value calculation Subsequently, the CPU 34 calculates a TF-IDF value for each word based on the calculated TF value and IDF value (step S5). In this embodiment, the TF-IDF value is calculated by calculating the product of the TF value and the IDF value.

(6) Hot coefficient calculation Subsequently, the CPU 34 calculates a Hot coefficient (step S6). Here, the Hot coefficient is an index representing the current attention level of the target word. In this embodiment, a target word is searched in a dictionary on the web, and how often the target word is updated and how close it is to the present time is acquired and calculated based on this. Yes.

  Specifically, for the word, a predetermined number (for example, 100) of update dates are acquired from the latest update date, an average date of these update dates is calculated, and a difference from the current date is obtained. The smaller the difference, the larger the Hot coefficient.

  FIG. 10 shows details of the Hot coefficient calculation process. The CPU 34 sets the first word recorded in the table of FIG. 7 (in this embodiment, only a word including a noun is targeted) as a target word (step S61). The CPU 34 searches for a target word by accessing a dictionary (web dictionary) on the Internet (step S62).

  If the target word exists as a headword, data such as an explanation for the target word is returned. If it does not exist as a headword, a message that it does not exist (or an error) is returned. In response to this, the CPU 34 determines whether or not the target word exists as a headword in the dictionary on the web (step S63).

  If it exists as a headword, the CPU 34 obtains a predetermined number (in this case, 100) of update dates recorded in the returned commentary data from the latest (step S64).

  FIG. 11 shows an example of comment data returned. In addition to items such as headwords and contents, items of update date are provided. On this update date, the history in which the contents etc. have been updated is shown in order. The CPU 34 acquires 100 update dates from the latest. If the number of updates is 100 or less, the hot coefficient of the word is set to 1.0 and the process proceeds to step S68.

  Next, the CPU 34 calculates the average date of the acquired 100 update dates. Next, the CPU 34 calculates the difference (absolute value) between the current date (based on the internal clock) and the average date (step S66). Further, a Hot coefficient is calculated based on the calculated difference (step S67).

  In this embodiment, the Hot coefficient is determined based on a coefficient table (see FIG. 12) recorded in advance on the hard disk 40. For example, if the calculated difference is 10 days, the Hot coefficient is determined as 2.6. The CPU 34 records the calculated Hot coefficient in association with the target word (see FIG. 8a).

  The CPU 34 determines whether or not the Hot coefficient has been calculated for all words (step S68). If an unprocessed word remains in the table of FIG. 7, the CPU 34 sets the next word as a target word (step S69). Then, step S62 and subsequent steps are repeated. In this way, the Hot coefficient is calculated for all target words.

  If the target word does not exist as an entry word in the dictionary on the web, the Hot coefficient is set to 1.0 (step S70).

(7) Calculation of Importance / Relevance The CPU 34 calculates the importance for each word by multiplying the TF-IDF value by the Hot coefficient (step S7 in FIG. 3). The CPU 34 records the importance calculated in this way in association with the target word (see FIG. 8a). Subsequently, the CPU 34 totalizes the importance levels of the words included in the target article, and calculates the importance level of the article (step S8). The CPU 34 records the calculated importance on the hard disk 40 in association with the corresponding article (see FIG. 8b).

  Subsequently, the CPU 34 determines whether or not all articles have been processed (step S9). If an unprocessed article remains, the next article is set as a target article (step S19). If all articles have been processed, the CPU 34 proceeds to a process for calculating the degree of association between articles.

  The CPU 34 first sets i = 1 and j = i + 1 (step S10). Next, ten words are extracted from the words included in the article K1 (numbers indicate the order of the articles recorded in the hard disk 40) from the most important words (step S11). Also, ten words are extracted from the words having the highest importance among the words included in the article K2 (step S12). Here, as shown in FIG. 13, it is assumed that the important word of the article K1 and the important word of the article K2 are selected.

  Next, the CPU 34 combines one of the important words of the article K1 and one of the important words of the article K2, and searches for the combined terms using a search engine on the Internet via the communication circuit 46. The CPU 34 acquires the number of hits sent as a search result (step S13). Such a search is performed for all combinations of the important words of the article K1 and the important words of the article K2. Therefore, here, 10 × 10 = 100 searches are performed.

  The CPU 34 totals the number of hits as the search results, and records 2.5 billion hits (rounded to the nearest 100 million) on the hard disk 40 as the relevance level R1,2 between the article K1 and the article K2 (step S14). ).

  Next, j is incremented (step S15), and the degree of association R1,3 between the article K1 and the article K3 is calculated by the same process as described above (steps S10 to S14). When this process is repeated until the degree of association R1,5 between the article K1 and the article K5 is calculated (that is, when j = 5 is exceeded), the CPU 34 increments i (steps S16 and S17). Further, j is set to i + 1 (step S17). As a result, i becomes 2 and j becomes 3.

  The CPU 34 repeats step S11 and subsequent steps. Thereby, the relevance R2,3, R2.4, R2,5 is calculated. Further, i is incremented, and the relevance degrees R3,4, R3,5 are calculated. Subsequently, the relevance R4, 5 is calculated. As described above, the importance of each article and the relevance between each article are calculated.

(8) Article Arrangement Next, the CPU 34 determines the arrangement of each article based on the article arrangement determination program 53. Flow charts of the article placement determination program 53 are shown in FIGS. In this embodiment, it is assumed that virtual springs proportional to the degree of relevance between articles are provided between the virtual article areas, and the total attractive force of the virtual springs when the virtual article areas are arranged is reduced. I try to arrange it.

  First, the CPU 34 generates a virtual article region having a size proportional to the importance of each article as a circle (step S71). Next, the CPU 34 stores pairs of articles Ki and Kj in descending order of the degree of relevance R (see step S72, FIG. 18). Further, the CPU 34 applies a virtual spring having a spring constant corresponding to the degree of relevance between these article pairs.

  In this embodiment, as shown in FIG. 26A, in a state where two virtual article regions are in contact, a virtual spring is applied between the virtual articles so that the virtual spring is in a natural length state. When the virtual article region is separated, an attractive force according to the magnitude of the spring constant is applied. In other words, if an article pair with high relevance is released, a large attractive force is generated in the virtual spring.

  Therefore, in this embodiment, the arrangement is determined so that the total attractive force of the virtual springs when the virtual article area is arranged becomes small. As a result, articles with high relevance are arranged so as to be in nearby positions.

  Next, the CPU 34 arranges the virtual article region corresponding to the article with the highest importance at the upper right of the virtual plane (see step S73, FIG. 19A). Here, the description will be made assuming that the importance of the article K1 is the highest. As shown in FIG. 19A, the virtual article region corresponding to the article K1 is arranged at the upper right of the virtual plane. The virtual plane is a virtual plane set corresponding to the area for displaying the article. The size (vertical / horizontal) of the virtual plane is set to change in proportion to the size (vertical / horizontal) of the article display area (display size of the terminal device). In this embodiment, the article with the highest importance is fixed at the upper right place with the highest attention.

  Next, the pair of articles having the highest degree of association is read out from the table of FIG. 18 (step S74). Here, a pair of article K1 and article K5 is selected. The CPU 34 determines whether at least one of the selected article pairs has already been arranged on the virtual plane (step S75). Here, since the article K1 is already arranged, the process proceeds to step S76. In step S76, the acquired article pair is arranged on a virtual plane. In the case of placement, the newly placed article is placed so as to be in contact with the paired article and as close to the upper right as possible (in this embodiment, it is on the right side from the top). Priority.)

  Here, as shown in FIG. 19B, an article K1 and an article K5 to which a virtual spring is applied are arranged. Since article K1 has already been arranged, it will not be arranged again. It should be noted that the virtual article areas are not overlapped when arranged.

  When the article pair is arranged as described above, the CPU 34 deletes the article pair from the table of FIG. 18 (step S77). Next, the CPU 34 determines whether or not the separately placed article is on the virtual plane (step S78). Here, since there is no separately placed article, the process proceeds to step S80.

  In step S80, the CPU 34 determines whether an article pair remains in the table of FIG. Here, since the article pair still remains, the process returns to step S73.

  In step S73, the article pair having the highest degree of association is selected from the article pairs remaining in the table of FIG. Here, the article pair K3, K5 is selected. Of the article pairs K3 and K5, the article K5 is already arranged on the virtual plane, so this article pair is arranged on the virtual plane (steps S75 and S76). Here, as shown in FIG. 20A, an article K3 with a virtual spring is placed between the article K5 and the article K5.

  When the above processing is completed, the CPU 34 reads and arranges the article pairs K1 and K3 from the table of FIG. Here, as shown in FIG. 20A, since the article K1 and the article K3 are already arranged, a process of applying a virtual spring between the article K1 and the article K3 is performed.

  By repeating the above processing, all articles K1 to K5 are arranged on a virtual plane. Thereby, as shown in FIG. 20B, articles K1 to K5 with virtual springs applied to each other are arranged.

  Note that some examples of processing when processing is performed using a table different from the table of FIG. 18 described above are shown below. For example, as shown in FIG. 21A, it is assumed that a pair of articles K1 and K5, a pair of articles K1 and K2, and a pair of articles K2 and K3 are arranged. In this arrangement state, when a pair of articles K2 and K5 is read from the table, K3 is moved in the direction of the arrow, and K2 is rotated in the direction of the arrow so as to contact K3. Thus, it arrange | positions like FIG. 21B.

  As shown in FIG. 22A, when the article K3 protrudes from the virtual space when the article K3 is to be arranged, a position that does not protrude is found and arranged. Alternatively, as shown in FIG. 22B, a position where the protruding area is as small as possible is found and arranged.

  If none of the article pairs read from the table is arranged in the virtual space, the article pair is placed separately (steps S75, S90, S91). For example, as shown in FIG. 23A, it is assumed that a pair of articles K3 and K4 is read from the table in a state where articles K1 and K5 are arranged. In this case, a pair of articles K3 and K4 is placed separately as shown in FIG. 23A.

  If any of the separately placed article pairs is placed in the virtual space, the separately placed article pair is placed in the virtual space (steps S78 and S79). For example, in the state of FIG. 23A, article pairs K1 and K4 are read from the table and article K4 is arranged (see FIG. 23B). Since the article K4 is now placed separately with the article K3, these pairs are arranged in the virtual space. The arrangement method is the same as in the case of arranging a normal pair.

  Hereinafter, the description returns to the table of FIG. If no article pair remains in the table of FIG. 18, the CPU 34 proceeds from step S80 to step S81. In step S81, it is determined whether or not a separately placed article remains. If there are any articles left separately, they are all arranged so as not to overlap with other articles on the virtual plane and as close to the upper right as possible (step S82).

  As described above, when the arrangement of the virtual article areas of all articles is completed, the circular article area is transformed into an article area consisting of a rectangle or a combination of rectangles, and the article area is determined (step S83). The transformation into a rectangle is replaced with a square having the same area without changing the center (a rectangle with a predetermined ratio may be defined in advance). For example, in the case shown in FIG. 24A, it is deformed as shown in FIG. 24B.

Subsequently, the CPU 34 deforms each article region so as to avoid an overlapping portion of the article regions and to fill a gap between the article regions. In this way, an article region as shown in FIG. 25 can be obtained.

4). Other embodiments
(1) In the above embodiment, a search is performed using a combination of important words of two articles, and the relevance is calculated based on the total number of hits. However, the association degree may be calculated as follows. The number of important words in one article is calculated in the other article, and this is performed for all the important words and totaled. Thus, the degree of association from one article to the other article can be obtained. Next, how many important words of the other article are included in one article is calculated, and this is performed for all the important words and totaled. Thereby, the degree of relevance to one article can be obtained from the other article. Finally, the degree of association from one article to the other article and the degree of association from the other article to one article are summed to obtain the degree of association.

(2) In the above embodiment, the article pairs are arranged in the virtual space in order from the article pair having the larger spring constant (that is, the relevance degree). By doing in this way, it aims at arrangement | positioning that the attractive force of a spring becomes small.

  However, in the state where the virtual article area is arranged, the attractive force acting on each virtual spring is calculated, and the virtual article area is changed so that the total of the attractive forces acting on the virtual spring becomes small, and the arrangement is determined. Also good.

  For example, after the virtual article area of the article K1 having the highest importance is arranged on the upper right, the virtual article areas of other articles are arranged randomly. This state is shown in FIG.

  Here, the CPU 34 sums up the attractive forces acting on each spring based on the spring constant between the articles. That is, the total force F is calculated by the following equation.

F = F1,2 + F1,3 + F1,4 + F1,5 + F1,6 + F2,3 + F2,4 + F2,5 + F2,6 + F3,4 + F3,5 + F3,6 + F4,5 + F4,6 + F5,6
Here, Fi, j is the attractive force acting on the articles Ki and Kj, and how much the spring constant pi, j of the virtual spring applied between the articles Ki and Kj and the outer peripheral surfaces of the articles Ki and Kj. It is obtained by the product of the distances xi, j indicating whether they are separated (see FIG. 26B).

  Next, the CPU 34 performs the above-described F1,2, F1,3, F1,4, F1,5, F1,6, F2,3, F2,4, F2,5, F2,6, F3,4, F3, 5, F3,6, F4,5, F4,6, F5,6 The largest one is selected. For example, assume that F2,5 is the largest. That is, it can be seen that the greatest force is working between the article K2 and the article K5. Therefore, the total force F may be reduced by rearranging the articles K2 and K5 so as to be close to each other. In other words, there is a possibility that more relevant articles can be placed nearby.

  The CPU 34 searches for an article that attracts the article K2 and the article K5 with a force greater than zero. For example, article K6 is in contact with article K2 and has an attractive force of 0. Similarly, article K3 is in contact with article K5 and has an attractive force of 0. Therefore, only the article K4 that is not in contact with any of the articles K2 and K5 is selected as satisfying this condition. The article K1 that is fixedly arranged is not a subject here.

  The CPU 34 recalculates the total force F by replacing the articles K2 and K5 where the greatest force is working with the selected article K4. That is, the location of the selected article K4 and article K2 is switched. Then, the total force F is recalculated. Similarly, the places of the article K4 and the article K5 are exchanged, and the total force F is recalculated.

  Then, among the original arrangement, the arrangement in which the articles K4 and K2 are exchanged, and the arrangement in which the articles K4 and K5 are exchanged, an arrangement having the smallest force F is selected.

  If the arrangement with the smallest force F is the original arrangement, the CPU 34 determines that the original arrangement is the preferred arrangement. If the arrangement in which the articles K4 and K2 are exchanged or the arrangement in which the articles K4 and K5 are exchanged is the arrangement where the force F is the smallest, the combination of the articles with the greatest attraction is found in the arrangement, and the above processing is repeated. In this way, an arrangement where the force F is reduced to some extent can be determined.

  In the above embodiment, for both the article K2 and the article K5 where the greatest force is working, an article that attracts with a force larger than 0 is searched and replaced with the virtual article region of the article. However, in FIG. 27, articles K6 and K3 may also be selected as objects to be exchanged. This is because article K6 is in contact with K2 but not in contact with K5, article K3 is in contact with K5 but not with K2, and K6 and K3 are not in contact with each other. In other words, it is connected by a virtual spring in a state where the force is not 0 in the paths K2, K3, K6, and K5.

  In this case, the CPU 34 exchanges the virtual article area of the article K3 and the virtual article area of the article K2, and exchanges the virtual article area of the article K6 and the virtual article area of the article K5 to see the change in the total power.

  In the above embodiment, virtual article areas are randomly arranged. However, the above-described replacement may be performed after the virtual article region is arranged by the processing shown in FIGS.

  Furthermore, in the said embodiment, the arrangement | positioning where force F becomes small by replacement is searched. However, after the virtual article area of the article having the highest importance is fixedly arranged, other virtual article areas are temporarily arranged at positions far away from each other, and may be determined to be naturally fitted by the force of the virtual spring. . In this case, other virtual article areas (virtual article areas other than the fixed article area) are allowed to overlap during movement. However, the overlap with other virtual article areas is prohibited at the moment of contact with the fixedly arranged virtual article area.

  In the above description, the method of separately placing article pairs that are not arranged in the virtual space has been described in steps S24, S40, and S41. However, instead of placing the article pair separately, place the article pair so that it touches the already placed article group (at this stage, provide a virtual spring between the article pair and the already placed article group. Finally, the total attractive force acting on the virtual spring may be reduced by the above replacement.

  In the above description, the attractive force of the virtual spring by the spring constant k is F = k × x, but it is sufficient that F is a monotonically increasing function of k and x. For example, if F = k × log (x), the influence of distance can be evaluated smaller.

  By performing such a simulation, the CPU 34 comes into contact with a virtual article region having a larger spring constant and fixedly arranged earlier, and appropriate arrangement can be realized.

(3) In the above embodiment, the article area is fixedly arranged at the upper right. This is because the attention level is the highest in the vertically written article. In the case of horizontal writing, it may be arranged at the upper left.

(4) In the above embodiment, a circle is used as the virtual article area. However, other figures such as a rectangle, a triangle, and a hexagon may be used.

(5) In the above embodiment, a fixedly arranged virtual article region is provided. However, all the virtual article areas may be movable.

5. Summary sentence generation program In the above embodiment, the arrangement of article areas according to the importance / relevance of articles has been described. The article is displayed in each article area arranged in this way. However, if the number of characters in the article is larger than the number of characters that can be displayed in the article area (determined by the size of the character font and the size of the article area), the entire contents of the article cannot be displayed. Therefore, it is necessary to create a summary within the number of characters that can be displayed in the article area and display it in the article area.

  The summary sentence generation program flowcharts are shown in FIGS.

(1) Sentence Extraction In step S101, the CPU 34 extracts a sentence from article text data recorded on the hard disk 40 (step S101). In this embodiment, a sentence is extracted based on a punctuation mark (“.” Or “.”). That is, a sentence until a punctuation is found is extracted as one sentence.

  For example, in the case of sentence data (article) as shown in FIG. 30, as shown in FIG. 31, six sentences with sentence numbers “1” to “6” are extracted. The CPU 34 records the sentence extracted in this way on the hard disk 40 together with the sentence number indicating the order. At this time, the CPU 34 also records the number of characters of each sentence.

(2) Morphological analysis Next, the CPU 34 performs morphological analysis on all sentences included in the sentence data (step S102). Details of the morphological analysis processing are the same as those shown in FIG. The result of morphological analysis is shown in FIG.

(3) TF Value Calculation Next, the CPU 34 calculates a TF (Term Frequency) value for the word extracted above (in this embodiment, only words including nouns are targeted) (step S103). That is, the TF value is calculated for a word whose part of speech recorded in the table of FIG. 7 is a noun and a word including the noun as a part of speech.

  Here, the TF value is a value obtained by dividing the number of times the word appears in the sentence by the total number of words in the sentence.

TF = number of appearances of words / total number of appearances of each word FIG. 8a shows the TF value calculated for each word. The CPU 34 records the calculated TF value on the hard disk 40.

  In this embodiment, by obtaining the appearance frequency of each word in the sentence, how much each word is emphasized in the sentence is calculated.

(4) IDF Value Calculation Next, the CPU 34 calculates an IDF (Inverse Document Frequency) value for each word (step S104). The details of the IDF calculation process are the same as those shown in FIG. The CPU 34 records the calculated IDF value in the hard disk 40 in association with each word (see FIG. 8a).

(5) TF-IDF value calculation Subsequently, the CPU 34 calculates a TF-IDF value based on the calculated TF value and IDF value for each word (step S105). In this embodiment, the TF-IDF value is calculated by calculating the product of the TF value and the IDF value.

(6) Hot coefficient calculation Subsequently, the CPU 34 calculates a Hot coefficient (step S106). The details of the Hot coefficient calculation process are the same as those shown in FIG. The CPU 34 records the calculated Hot coefficient in association with the target word (see FIG. 8a).

(7) Calculation of Importance The CPU 34 calculates the importance for each word by multiplying the TF-IDF value by the Hot coefficient (step S107). The CPU 34 records the importance calculated in this way in association with the target word (see FIG. 8a).

  Next, the CPU 34 determines an important word based on the importance of each word (step S108). In this embodiment, an average value of importance levels of words is calculated, and words having importance levels exceeding the average value are extracted as important words. For example, in the case shown in FIG. 8a, the average word importance is 11.40135438. Therefore, the CPU 34 selects the words “electronic book”, “library”, “book”,... “Use value” having importance exceeding the average value as important words.

  Next, the CPU 34 calculates an importance level for a sentence including an important word among the sentences. The importance is calculated by summing the importance of each word included in the sentence (step S109). For example, since the first sentence of the sentence shown in FIG. 31 “Plain is plain ...” does not include an important word, the importance of the first sentence is not calculated. In the second sentence “Our company has been here ...”, the important words “Our company”, “Library” and “Book” are included, so the importance is calculated. The CPU 34 acquires the importance of each word with reference to FIG. 8a and adds them together to obtain the importance 57.83187.

  The CPU 34 records the calculated importance of each sentence on the hard disk 40 in association with each sentence (see FIG. 31).

(8) Generation of Summary Sentence Next, the CPU 34 extracts the sentence with the highest importance based on the calculated importance of each sentence (step S110). Here, the importance of the fourth sentence shown in FIG. Since it is the highest, the fourth sentence is selected. The CPU 34 determines whether or not the total number of characters of the selected sentence exceeds a predetermined number of designated characters (step S111). Here, the number of characters that can be displayed in the article area is the designated number of characters. Here, the description proceeds assuming that the designated number of characters is 250 characters.

  The number of characters in the fourth sentence is 62 characters (see FIG. 31). Therefore, since the designated number of characters does not exceed 250, the process proceeds to step S12. In step S12, the next most important sentence is extracted. Here, the fifth sentence is selected. The fifth character number is 97 characters. Therefore, the total number of characters of the selected sentence is 160 characters, which is the sum of 62 characters and 97 characters. Therefore, since the designated number of characters does not exceed 250, the process proceeds from step S111 to step S112.

  In step S112, the next most important sentence is extracted. Such processing is repeated until the total number of characters of the selected sentence exceeds the designated number of characters 250. Here, when the fourth sentence (62 characters), the fifth sentence (97 characters), the third sentence (74 characters), and the second sentence (56 characters) are extracted, the total number of characters becomes 289 characters and the designated number of characters. Over.

  When the specified number of characters is exceeded, the CPU 34 removes the last extracted sentence from the extracted sentence (step S113). Here, since the 2nd sentence is the sentence extracted last, it removes from what extracted the 2nd sentence. As a result, as shown in FIG. 32, the fourth sentence, the fifth sentence, and the third sentence are extracted.

  Next, the CPU 34 rearranges the extracted sentences according to the order in the original sentence (that is, the sentence number is in ascending order) (step S114). Thereby, the extracted sentences in the order as shown in FIG. 33 can be obtained.

  The CPU 34 outputs the extracted sentences rearranged in this way as a summary (step S115). Here, the summary “The digitization of the book… I want to connect” shown in FIG. 33 is output to the corresponding display area.

The above processing is performed for all display areas. Thereby, the summary sentence of the corresponding article is displayed in each display area. The CPU 34 displays the link information to the full article along with the summary sentence.

6). Other embodiments
(1) In the above embodiment, the importance level of a word is calculated based on the TF value (enhancement level), IDF value (rareness level), and Hot coefficient (attention level). However, the importance may be calculated using any one of these, or the importance may be calculated by combining any two of these.

(2) In the above embodiment, the word emphasis degree is calculated based on the number of appearances of the word in the sentence. However, the degree of word emphasis may be determined based on the appearance position of the word in the sentence. For example, the word appearing at the back of the sentence may be enhanced. further. The enhancement degree may be calculated by combining the number of appearances and the appearance position.

(3) In the above embodiment, the rarity (IDF value) is calculated based on the ratio between the number of hits of the word on the search site and the total number of pages to be searched. However, how many times the word is used in dictionaries and encyclopedias (regardless of whether they are on the web or local), including how many times the word is used in the content Based on this, the degree of rarity of the word may be calculated.

(4) In the above embodiment, the current attention level (Hot coefficient) is calculated based on the difference between the average value of the updated dictionaries on the web and the current date. However, the current attention level may be calculated based on the difference from the latest update date. Alternatively, the number of updates performed within a predetermined period may be calculated, and the current attention level may be calculated based on this (the higher the update count, the higher the attention level).

(5) In the above embodiment, important words are extracted using the average value of the importance of each word as a threshold value. However, important words may be extracted in consideration of standard deviation and variance.

(6) In the above embodiment, the importance is calculated only for the sentence including the important word. However, importance may be calculated for all sentences.

(7) In the above embodiment, sentences are selected within a range not exceeding the specified number of characters. However, the sentence may be selected so that the absolute value of the difference between the specified number of characters and the total number of characters is minimized.

(8) In the above embodiment, a sentence is selected based on the designated number of characters. However, the sentence may be selected based on the total number of sentences, the total number of lines, the total data amount of the sentences, and the like.

(9) In the above embodiment, it is assumed that sentence selection criteria such as the number of designated characters are predetermined. However, the number of designated characters may be made variable according to the number of characters in the target sentence (for example, half the number of characters in the sentence is designated as the designated number of characters). Thereby, a summary sentence having a length corresponding to the length of the sentence can be obtained.

(10) In the above embodiment, the summary sentence creation process is performed on the terminal device side. However, the summary sentence creation process may be performed on the server device side.

(11) In the above embodiment, “sentence” is extracted from the sentence based on the punctuation mark. However, some sentences, such as headings, have no punctuation. Therefore, the part determined to be a headline (determined based on the fact that the font is larger than the other parts, the attribute of being a headline, etc.) may be determined as a whole sentence. .

(12) In the above embodiment, when calculating the importance of a word, the genre and attributes of the whole sentence (whether it is a social article, a three-faced article, a paper or a miscellaneous sentence, etc.) Thus, a coefficient may be provided for the word. For example, the “National Diet” in society is a word that usually appears, so the coefficient is set low. On the other hand, “National Diet” in the three-page article is a word that does not appear so often, so the coefficient is increased. Multiply the calculated importance by this coefficient to obtain the final importance.

(13) In the above embodiment, the importance of a sentence is calculated without considering the relation between the sentence and the sentence. However, the importance may be calculated in consideration of the relationship between sentences. For example, when the pronouns such as “that” and “he” appear in the sentence, in the extracted sentence selected according to the importance and rearranged, the sentence before the sentence is not extracted. Reduce the coefficient of sentences containing the pronoun. On the other hand, if the sentence contains a conjunction that expands the topic, such as “so” or “that is”, the coefficient of the sentence is increased. Multiply the calculated importance by this coefficient to obtain the final importance.

Claims (15)

An article placement determination device that places a plurality of articles on a plane,
A recording unit that records multiple articles and the importance of each article, and records the degree of association between each article,
Virtual article region generation means for generating a virtual article region corresponding to each article and having an area according to the importance of each article;
Virtual spring setting means for setting a virtual spring having a spring constant according to the degree of association between each article between the corresponding virtual article regions;
Virtual article area arrangement determining means for determining the arrangement of the virtual article area such that the total attractive force of the virtual springs is reduced;
An article area arrangement determining means for determining the arrangement of the article area based on the determined arrangement of the virtual article area;
Article placement determination device with An article arrangement determining program for realizing an article arrangement determining apparatus for arranging a plurality of articles on a plane by a computer,
Virtual article region generation means for generating a virtual article region corresponding to each article and having an area according to the importance of each article;
Virtual spring setting means for setting a virtual spring having a spring constant according to the degree of association between each article between the corresponding virtual article regions;
Virtual article area arrangement determining means for determining the arrangement of the virtual article area such that the total attractive force of the virtual springs is reduced;
An article arrangement determination program for functioning as article area arrangement determination means for determining the arrangement of article areas based on the determined arrangement of virtual article areas. In the apparatus of claim 1 or the program of claim 2,
The apparatus further includes force total calculation means for calculating a total value of attractive forces of virtual springs arranged between the virtual article regions arranged on the virtual plane,
The virtual article area arrangement determining means changes the arrangement of the virtual article area on the virtual plane and the virtual article area arrangement means for arranging the virtual article area on the virtual plane so that the total value of the attractive force of the virtual spring is reduced. An apparatus or program comprising virtual article area arrangement change determining means for determining arrangement of a virtual article area. In the apparatus or program in any one of Claims 1-3,
The virtual article region generation means generates a circle having an area corresponding to the importance of each article as the virtual article region. In the apparatus or program in any one of Claims 1-4,
The virtual article area arranging unit arranges the virtual article area having the highest importance fixedly arranged at the upper right or upper left on the plane. In the apparatus or program in any one of Claims 1-5,
The virtual article area arranging unit arranges virtual article areas in order from a pair of virtual article areas having a high degree of relevance. In the apparatus or program in any one of Claims 1-6,
The virtual spring setting means sets a virtual spring so that no attractive force is generated between virtual article regions that are in adjacent contact with each other. In the apparatus or program in any one of Claims 1-7,
The virtual spring setting means increases the spring constant as the degree of association between articles increases. In the apparatus or program in any one of Claims 1-8,
The virtual article area arrangement change determining means performs an arrangement change by paying attention to a virtual article area related to a virtual spring having the greatest attraction. An article placement determination device that places a plurality of articles on a plane,
A recording unit that records multiple articles and the importance of each article, and records the degree of association between each article,
Virtual article region generation means for generating a virtual article region corresponding to each article and having an area according to the importance of each article;
Among the generated virtual article areas, those having the greatest importance are set as fixed virtual article areas fixedly arranged so as not to overlap each other on the virtual plane, and other virtual article areas are arranged on the virtual plane. In the virtual article region arrangement means to be a movable virtual article region that is arranged far away to allow overlapping with each other,
Virtual spring setting means for setting a virtual spring having a spring constant according to the degree of association between each article between the corresponding virtual article regions;
The movable virtual article region arranged on the virtual plane is moved toward the fixed virtual article region in accordance with the set attractive force of the virtual spring, and the movable virtual article region is moved to the fixed virtual article region. The article area arrangement determining means for determining the arrangement of the virtual article area by making the movable virtual article area as a fixed virtual article area,
Article placement determination device with An article arrangement determining program for realizing an article arrangement determining apparatus for arranging a plurality of articles on a plane by a computer,
Virtual article region generation means for generating a virtual article region corresponding to each article and having an area according to the importance of each article;
Among the generated virtual article areas, those having the greatest importance are set as fixed virtual article areas fixedly arranged so as not to overlap each other on the virtual plane, and other virtual article areas are arranged on the virtual plane. In the virtual article region arrangement means to be a movable virtual article region that is arranged far away to allow overlapping with each other,
Virtual spring setting means for setting a virtual spring having a spring constant according to the degree of association between each article between the corresponding virtual article regions;
The movable virtual article region arranged on the virtual plane is moved toward the fixed virtual article region in accordance with the set attractive force of the virtual spring, and the movable virtual article region is moved to the fixed virtual article region. An article arrangement determining program for causing the movable virtual article area to function as article area arrangement determining means for determining the arrangement of the virtual article area by making the movable virtual article area a fixed virtual article area. In the apparatus or program in any one of Claims 1-11,
The apparatus further comprises:
A word extracting means for analyzing an article to be analyzed and extracting at least nouns constituting the article as a word;
Importance calculating means for calculating the importance of each word included in the article extracted by the word extracting means, and calculating the importance of each article based on the importance of each word;
A device or program comprising: In the apparatus or program in any one of Claims 1-12,
The apparatus further comprises:
Relevance calculating means for calculating relevance between articles based on words included in each article;
A device or program comprising: The apparatus or program of claim 13,
The relevance calculation means includes:
An important word determination means for determining a predetermined number of words as important words from the highest importance in each article;
Relevance between words that search for important words contained in one article for calculating the degree of association and important words contained in the other article for which the degree of association is calculated on a search site on the web and obtain the number of hits An inter-word relevance calculating means for calculating
Change the combination of the important words contained in the one article and the important words contained in the other article, and sum the relevance obtained by the inter-word relevance calculation means to calculate the relevance between both articles Between-article relevance calculation means,
A device or program comprising: A method of determining a plurality of article arrangements on a plane by a computer, the computer comprising:
Create a virtual article area that corresponds to each article and has an area according to the importance of each article,
Place the generated virtual article area on the virtual plane,
A virtual spring having a spring constant corresponding to the degree of association between each article is set between the corresponding virtual article regions,
Changing the placement of the virtual article area on the virtual plane, determining the placement of the virtual article area so that the total attractive force of the virtual springs is reduced,
An article arrangement determining method, wherein the article area arrangement is determined based on the determined virtual article area arrangement.

Images