Tabulation table integration system and program

Abstract

To provide a tabulation table integration system for enabling any worker to easily integrate tabulation tables without time and effort.SOLUTION: In a tabulation table integration system 10, inter-element similarity calculation means 24 calculates the degree of similarity between head elements and front elements for tabulation tables at different times to determine the head elements and front elements whose semantic contents correspond to each other, and then inter-element similarity calculation means 25 calculates the degree of similarity between head element sets and front element sets by considering the head elements and front elements whose semantic contents correspond to each other as the same element, determines whether or not tabulation tables at different times can be integrated, and when it is determined that integration is possible, integration means 26 prepares an integrated tabulation table.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a summary table integration system and program configured by a computer that executes processing to integrate summary tables from different points in time, and can be used, for example, when summary table data from different points in time are automatically created from raw data from different points in time by a summary table creation system, and the summary table data are arranged and integrated so that they can be compared. [Background technology]

[0002] Generally, data collected from various surveys such as questionnaire surveys and property inspections is used to create a summary table showing the number of samples (how many people or how many items) that apply, or a summary table showing the percentage of samples (what percentage of people or items) that apply.

[0003] For example, after a survey, raw data is first created, as shown in Figure 16. Raw data means raw data that shows survey results in an unprocessed state, and in the example of Figure 16, response data for each survey item for one sample (one respondent) is represented in one line and arranged horizontally. Specifically, for example, the top row (excluding the title row listing the survey items) lists the response data for sample ID=S0001 (Tenko Taro), and for survey item Q1="First, tell us about yourself. Where is your high school located?", the response data "12" (Tokyo) is entered; for survey item Q2="Tell us about your own career choice. What was your deviation score?", the response data "1" (very important) is entered; for survey item Q3="Tell us about your own career choice. How well-known it was?", the response data "3" (neither) is entered; and for survey item Q4="Department that you were accepted into at XX University?", multiple selections are allowed, so the response data is in flag format (1,0 format) indicating whether or not each department was accepted.

[0004] Below that, an example of a summary table with summary table ID (summary table identification information) = [N0001] is shown. This summary table has a table head portion consisting of a table head element and a table head title at the top of the table (there may be no table head title, but in this case, it is still called the table head portion), and a table side portion consisting of a table side element and a table side title at the left end of the table (there may be no table side title, but in this case, it is still called the table side portion). Each data that fills the matrix (matrix) formed by each table head element arranged horizontally (Aomori Prefecture, Tokyo, etc.) and each table side element arranged vertically (Childhood Studies, English Literature, etc.) is all summary data, and in this application, a set of these summary data or their arrangement position may be called a summary data portion, distinguished from the table head and table side portions. Furthermore, when "summary table data" (the same applies to "integrated summary table data" and "combined summary table data") is used instead of "summary data," it refers to the entire summary table data, combining the top table section, side table section, and summary data section. Note that this summary table is an example of a summary table that displays data in frequency (number of people), but the names of the top table section, side table section, and summary data section are the same even if it displays data in percentage (% of number of people).

[0005] Furthermore, there are two types of summary tables created from raw data: simple summary tables and cross summary tables. The summary table in FIG. 16 is an example of a cross summary table.

[0006] The former simple summary table is used to summarize data (answer data for each survey item) collected by questionnaire surveys, etc., for each survey item. Specifically, using the raw data in FIG. 16, for example, if Q1 = "First, tell us about yourself. Where is your high school located?" is selected as one survey item to be summarized, the answer data to this is how many people are in 1 (Aomori Prefecture), how many people are in 2 (Iwate Prefecture), how many people are in 3 (Miyagi Prefecture), etc., and simple summary (Grand Total) is used to create summary data, and the simple summary table (GT table) is used to display this in a table format. Therefore, in the case of this simple summary table, 1 (Aomori Prefecture), 2 (Iwate Prefecture), 3 (Miyagi Prefecture), etc. are simply lined up as either the top or side table elements, and the top table elements are not lined up horizontally and the side table elements are not lined up vertically, so it is not a summary table to which the present invention is applicable.

[0007] The latter cross-tabulation table uses data collected by questionnaire surveys, etc. (response data for each survey item) to combine and tabulate multiple survey items (responses to the survey items). Therefore, the top element of the table is the response to a certain survey item, and the side elements are responses to other survey items. Specifically, as shown in the tabulation table in Figure 16, the top elements "Chiba Prefecture" and "Tokyo" are responses to a certain survey item, and the side elements "Childhood Studies Department" and "Food Studies Department" are responses to other survey items, and tabulated data is created by combining these. There are 103 people in total whose high school is located in "Chiba Prefecture", of which 14 people passed the entrance exam for "Childhood Studies Department" and 3 people passed the entrance exam for "Food Studies Department". In addition, there are 326 people in total whose high school is located in "Tokyo", of which 28 people passed the entrance exam for "Childhood Studies Department" and 7 people passed the entrance exam for "Food Studies Department".

[0008] Incidentally, such questionnaire surveys and the like are conducted to measure changes over time, and therefore are often conducted repeatedly, for example, every year, month, or quarter, and the survey objectives, methods, and items are often almost the same each time. Therefore, it would be convenient to be able to create an integrated summary table that integrates data from summary tables that are repeatedly conducted at different points in time (periods) and arranges the data from those different points in time in a way that makes it easy for people to visually compare them.

[0009] On the other hand, traditionally, raw data from different points in time are integrated before creating a summary table. When raw data from different points in time are integrated, the samples (respondents in the example of Figure 16) from each point in time are arranged vertically, and the number of rows (number of samples) increases. In this case, there is no overlap in the samples from each point in time, or even if there is an overlap, they are considered to be different samples (in the example of Figure 16, the sample from 2019, Tokkyo Taro, and the sample from 2020, Ronin Tokkyo Taro, are considered to be different samples) and are integrated. Therefore, when creating an integrated simple summary table (GT table), the aggregation process becomes easier. However, the aggregation process that is said to be easier here is the process of totaling raw data from multiple points in time, for example, the raw data from 2019 and 2020 are arranged vertically and linked to obtain the total number of people for the two years combined into one. However, when creating an integrated summary table to compare the total number of people in 2019 and 2020, it is necessary to perform aggregation processing for each point in time (for each fiscal year in the example of Figure 16), so arranging raw data from multiple points in time vertically and integrating them does not ensure effective aggregation processing. In other words, if the term "integration" means both the totalization of data and the comparative display of data, then when performing the latter kind of integration, arranging raw data vertically is almost useless.

[0010] Furthermore, when the summary table for each time point is a cross-tabulation table rather than a simple summary table (GT table), multiple survey items are combined and tabulated, so in this case as well, if the raw data from multiple time points is to be aggregated, it makes sense to create a cross-tabulation table after arranging and concatenating the raw data from multiple time points vertically (for example, after arranging and concatenating the raw data from 2019 and 2020 vertically). However, when creating an integrated summary table to compare cross-tabulation table data from different time points, for example, the data from the cross-tabulation table for 2019 and the data from the cross-tabulation table for 2020, tabulation processing is required for each time point (for each fiscal year in the example of Figure 16), so even if raw data from multiple time points is arranged vertically and integrated, effective tabulation processing cannot be performed.

[0011] As a system for integrating tables, for example, a form recognition device that recognizes table structures and automatically determines the attributes of the minimum rectangle is known (see Patent Document 1). As a system that uses the top and side parts of a table, for example, a data processing device that enables vectorization of natural language descriptions in table data is known (see Patent Document 2). [Prior art documents] [Patent documents]

[0012] [Patent Document 1] JP 2010-61693 A (paragraphs

[0016] ,

[0019] ,

[0042] ,

[0043] ,

[0133] ,

[0141] ) [Patent Document 2] Patent No. 7192991 Publication (Paragraph

[0030] ) Summary of the Invention [Problem to be solved by the invention]

[0013] As mentioned above, the conventional practice is to vertically line up and concatenate raw data from multiple points in time and then create a summary table. However, when performing an integration process to compare aggregated data from different points in time, integrating raw data from multiple points in time is of little use.

[0014] Therefore, instead of integrating raw data, it is possible to directly create integrated summary table data using summary table data from different points in time. However, this work must be done manually, which is time-consuming and laborious, and there is a risk of data transcription errors.

[0015] More specifically, as mentioned above, questionnaire surveys and the like are conducted repeatedly to measure changes over time, so the purpose, method, and items of each survey do not change significantly. However, for the following reasons, it is often not possible to carry out a simple integration task.

[0016] In other words, if the survey time (survey period) is different, the number of survey items used as the top and side titles in the summary table may increase or decrease (change in the column configuration of raw data), or the wording of the survey items may change (change in the content of the character strings that make up the questions). It may also involve an increase or decrease in the top and side elements (an increase or decrease in the number of answer options for a question), a change in the wording of the top and side elements (change in the content of the character strings that make up the answers), or a change in the order of the top and side elements.

[0017] Furthermore, among the table head elements and table side elements, there are some that do not simply use the numerical values ​​of the raw data but rather use some kind of calculation to obtain aggregate data through synthesis processing, such as aggregating aggregate data for multiple elements such as "Aomori Prefecture," "Iwate Prefecture," and "Akita Prefecture" to obtain aggregate data for the element "Tohoku," or calculating the average value of aggregate data for several elements to obtain aggregate data for another element, and the contents of this synthesis processing can also change over time. For example, at one point in time, synthesis processing was being carried out to obtain aggregate data for the element "Tohoku," but at another point in time, this might change to the element "Hokkaido and Tohoku."

[0018] For this reason, when creating an integrated summary table manually, it is necessary to judge the identity of the table top title and table side titles, or the identity of the table top elements and table side elements. This not only takes time and effort, but the results of the identity judgment may differ depending on the knowledge and experience of the person performing the integration work, which may result in differences in the contents of the created integrated summary table.

[0019] An object of the present invention is to provide a spreadsheet integrating system and program that enables any worker to easily integrate spreadsheets without spending a lot of time and effort. [Means for solving the problem]

[0020] The present invention is characterized by using each character string in the top and side parts of a summary table, particularly each character string in the top and side elements of the table, to determine whether summary tables from different points in time (i.e., multiple points in time) can be integrated, or to detect items to be integrated, and to perform an integration process.Specifically, the following configuration can be adopted.

[0021] <Configuration (first configuration) of calculating the similarity between table head elements and table side elements, determining table head elements and table side elements whose semantic contents correspond, and then calculating the similarity between table head element groups and table side element groups to determine whether or not to integrate the summary tables>

[0022] The present invention provides a summary table integration system configured by a computer that executes a process for integrating summary tables at different times, an inter-element similarity calculation means for calculating a table top element similarity indicating a similarity between character strings of a plurality of table top elements constituting the table tops of the summary tables at different times to be integrated and written in the summary table file, and determining table top elements having corresponding semantic contents between the summary tables at different times by comparing the calculated table top element similarity with a preset threshold value, and for calculating a table side element similarity indicating a similarity between character strings of a plurality of table side elements constituting the table side parts of the summary tables at different times to be integrated, and for comparing the calculated table side element similarity with a preset threshold value to determine table side elements having corresponding semantic contents between the summary tables at different times; an inter-set similarity calculation means for determining whether or not to integrate the summary tables at different times based on whether the table head elements having corresponding semantic contents determined by the inter-element similarity calculation means are the same element, calculating an inter-table head element set similarity indicating the similarity between table head element sets consisting of a plurality of table head elements constituting the table head, comparing the calculated inter-table head element set similarity with a preset threshold, determining whether or not to integrate the summary tables at different times based on whether the table head element set similarity and the inter-table side element set similarity are both high; an integration means for executing an integration process for creating an integrated summary table by vertically or horizontally arranging the summary data for the table head element and the table side element adjacent to each other when the semantic contents of the summary tables at different times correspond to each other for both the table head element and the table side element among the respective summary data of the summary tables at different times that are determined to be integrable by the inter-set similarity calculation means; The present invention is characterized by comprising:

[0023] Here, the "summary tables of the integration targets at different times written in a summary table file" in the "inter-element similarity calculation means" means that the summary tables at each time may be written in separate summary table files, or may be written together in a single summary table file. The former case is, for example, when a summary table file for a certain time Ta and a summary table file for a later time Tb are prepared. The latter case is when summary tables for different times are written in different sheets in the same file in a spreadsheet software (for example, Microsoft Excel (registered trademark), etc.). The same applies to other inventions.

[0024] The integrated tally table file into which the "integrated tally table" created by the "integration means" is written may be a file prepared separately from the tally table files into which the individual tally tables before integration are written, or may be the same file. The same applies to other inventions.

[0025] In addition, the "different points in time" in the "process of integrating summary tables at different points in time" may be, for example, two points in time, such as 2020 and 2021, or may be three or more points in time, such as 2019, 2020, 2021, .... The same applies to other inventions.

[0026] When integrating tally sheets for three or more points in time, it is preferable to repeat the integration process for two points in time, for example, by first integrating 2019 and 2020, and then integrating the integrated tally sheet with the tally sheet for 2021, in order to facilitate the calculation process. Therefore, the "point in time" when referring to "different points in time" also includes a representative point in time for a plurality of points in time when tally sheets for a plurality of points in time are integrated. For example, when 2019 and 2020 are integrated, this representative point in time may be either of those points in time (the point in time selected as the creation reference point of 2019 or 2020), or may be a virtual point in time that does not exist in reality (for example, "2019 & 2020", "2019-2020 based on 2020", etc.). Furthermore, for example, when integrating summary tables for four points in time, 2019, 2020, 2021, and 2022, an integrated summary table for 2019 and 2020 may be created, as well as an integrated summary table for 2021 and 2022, and these integrated summary tables may then be further integrated. Therefore, integration of "different points in time" also includes integration of a representative point in time (e.g., "2019 & 2020") and a representative point in time (e.g., "2021 & 2022").

[0027] Instead of repeating the integration process for two time points, it is also possible to integrate the summary tables for three or more time points at once. In that case, the calculation process becomes complicated because various similarities between each time point are calculated in a brute-force manner. In other words, if the number of time points to be integrated is N, N Various similarities between two points in time (similarity between table top elements, similarity between table side elements, similarity between table top element sets, and similarity between table side element sets) are calculated in C2 ways, and compared with a threshold value to determine whether they are high or low. For example, when integrating summary tables for three points in time, 2019, 2020, and 2021, 3C2 similarities, i.e., various similarities between 2019 and 2020, various similarities between 2020 and 2021, and various similarities between 2019 and 2021, are calculated, and all of the various similarities between two points in time in these 3C2 ways are compared with a threshold value to determine whether the various similarities between any two points in time are high or low.

[0028] Furthermore, the "summary table" in this invention includes a summary table that displays frequency (displaying number of people, number of items, number of cases, etc.) and a summary table that displays percentage (displaying percentage, per mille, percentage, etc.). Note that there are simple summary tables (GT tables) and cross summary tables, but this invention is applicable to cross summary tables (see the explanation of FIG. 16 above).

[0029] In the summary table integration system of the present invention, the inter-element similarity calculation means calculates the similarity between table head elements and table side elements to determine the table head elements and table side elements which correspond in meaning, and then the inter-set similarity calculation means calculates the similarity between table head element sets and table side element sets to determine whether summary tables from different points in time (i.e., multiple points in time) can be integrated, and if it is determined that integration is possible, the integration means creates an integrated summary table.

[0030] Therefore, the person performing the integration work does not need to judge the identity of the table top elements and table side elements, or judge whether the summary tables can be integrated, so that the integrated summary table can be easily created without time or effort. Also, if an appropriate threshold is set in advance (the threshold may be set by the person performing the integration work, or by someone else), the integration process can be executed without the intervention of human judgment, so that any worker can create a uniform integrated summary table regardless of the worker's knowledge or experience, thereby achieving the above-mentioned object.

[0031] <A configuration in which summary tables at different time points to be integrated are detected from among a plurality of summary tables existing for each of the different time points, and then, for the summary tables to be integrated that have been detected, similarities between table head elements and table side elements are calculated, table head elements and table side elements whose semantic contents correspond are determined, and then similarities between sets of table head elements and sets of table side elements are calculated, and whether or not the summary tables can be integrated is judged (second configuration)>

[0032] Furthermore, in the above-mentioned summary table integration system (first configuration), when there are multiple summary tables for each different point in time, the summary tables for the different points in time to be integrated may be specified by a human or may be detected by the system; however, from the viewpoint of eliminating the need for manual specification and thus saving labor, the following configuration (second configuration) may be used.

[0033] That is, in the summary table integration system described above, a merging target detection means for detecting a tally table at a different time point to be merged from among a plurality of tally tables existing for each of the different time points written in the tally table file; The integrated object detection means includes: (1) For each of the summary tables at different times, a table head / side part concatenated string is created by concatenating a table head element concatenated string obtained by concatenating each character string of a plurality of table head elements constituting the table head and / or a table head string including a character string of the table head title, and a table side element concatenated string obtained by concatenating each character string of a plurality of table side elements constituting the table side and / or a table side part string including a character string of the table side title, and using the created table head / side part concatenated strings of the summary tables at different times, a table head / side part similarity indicating the similarity of the table head and table side part of the summary tables at different times is calculated by an algorithm different from the algorithm for calculating similarity used in the inter-set similarity calculation means, and the calculated table head / side part similarity is compared with a preset threshold value, and depending on whether the table head / side part similarity is high or not, the summary tables at different times to be integrated are detected, or (2) A table top similarity indicating the similarity of the table tops of the summary tables at different times is calculated by an algorithm different from the similarity calculation algorithm used in the set similarity calculation means, using a table top element concatenated string obtained by concatenating the strings of each of a plurality of table top elements constituting the table tops of the summary tables at different times and / or a table top string including a string of a table top title, and the calculated table top similarity is compared with a preset threshold value, and at the same time, a table side portion similarity indicating the similarity of the table side portions of the summary tables at different times is calculated by an algorithm different from the similarity calculation algorithm used in the set similarity calculation means, using a table side element concatenated string obtained by concatenating the strings of each of a plurality of table side elements constituting the table side portions of the summary tables at different times and / or a table side portion string including a string of a table side title, and the calculated table side portion similarity is compared with a preset threshold value, and the summary tables at different times to be integrated are detected depending on whether or not both the table top similarity and the table side portion similarity are high. The element similarity calculation means The processing may be performed by using summary tables at different times detected by the integration target detection means as integration targets.

[0034] Here, the "table head string including a table head element concatenated string and / or a table head title string" in the "integration target detection means" may be only a table head element concatenated string, only a table head title string, or may include both a table head element concatenated string and a table head title string. Similarly, the "table side part string including a table side element concatenated string and / or a table side title string" in the "integration target detection means" may be only a table side element concatenated string, only a table side title string, or may include both a table side element concatenated string and a table side title string. The reason why only a table head title or a table side title string is sufficient is that the inter-set similarity calculation means will ultimately determine whether or not summary tables at different points in time (i.e., multiple points in time) can be integrated, and therefore it is sufficient that there is a possibility of integration at the detection stage by the integration target detection means.

[0035] In this manner, when the configuration is provided with an integration target detection means (second configuration), the integration target detection means detects summary tables at different times to be integrated from the multiple summary tables that exist for each of the different times, and then the inter-element similarity calculation means calculates the similarity between the top elements and the side elements for the detected summary tables to be integrated, and determines the top elements and side elements which correspond in meaning. Thereafter, the inter-set similarity calculation means calculates the similarity between the top element sets and the side element sets to determine whether the summary tables at different times (i.e., multiple time points) can be integrated, and if it is determined that integration is possible, the integration means creates an integrated summary table.

[0036] Therefore, since the integration target detection means detects the target summary tables, it is possible to eliminate the need for a person to specify the target summary tables, thereby saving labor and time.

[0037] <A configuration in which summary tables at different times to be integrated are detected from among a plurality of summary tables existing for each of the different times, and then similarities between table head elements and table side elements are calculated for the summary tables to be integrated detected, and table head elements and table side elements having corresponding semantic contents are determined (third configuration)>

[0038] Furthermore, the present invention provides a summary table integration system configured by a computer that executes a process for integrating summary tables at different times, comprising: an integration target detection means for detecting a summary table at a different time point to be integrated from among a plurality of summary tables existing for each of the different time points written in the summary table file; an inter-element similarity calculation means for calculating inter-table head element similarities indicating the similarity between character strings of a plurality of table head elements constituting the table heads of the summary tables at different times to be integrated detected by the integration target detection means, and determining table head elements whose semantic contents correspond to each other between the summary tables at different times by comparing the calculated inter-table head element similarities with a preset threshold value, and for calculating inter-table side element similarities indicating the similarity between character strings of a plurality of table side elements constituting the table side parts of the summary tables at different times to be integrated, and for comparing the calculated inter-table side element similarities with a preset threshold value to determine table side elements whose semantic contents correspond to each other between the summary tables at different times; an integration means for executing an integration process for creating an integrated summary table by vertically or horizontally arranging the summary data for the table head element and the table side element adjacent to each other when the semantic contents of the summary tables at different times of the integration target detected by the integration target detection means correspond to each other for both the table head element and the table side element, The integration target detection means includes: (1) For each of the summary tables at different times, a table head element concatenated string is created by concatenating the character strings of multiple table head elements constituting the table head or a table head string by adding a character string of a table head title to this table head element concatenated string, and a table side element concatenated string by concatenating the character strings of multiple table side elements constituting the table side or a table side string by adding a character string of a table side title to this table side element concatenated string, and using the created table head and side part concatenated strings of the summary tables at different times, a table head and side part similarity indicating the similarity between the table heads and the table side parts of the summary tables at different times is calculated, and the calculated table head and side part similarity is compared with a preset threshold value, and depending on whether the table head and side part similarity is high or not, the summary tables at different times to be integrated are detected, or (2) A table head similarity indicating a similarity between the table heads of the summary tables at different times is calculated using a table head element concatenated string obtained by concatenating the character strings of a plurality of table head elements constituting the table heads of the summary tables at different times, or a table head string obtained by adding a character string of a table head title to the table head element concatenated string, and the calculated table head similarity is compared with a preset threshold value. At the same time, a table side portion similarity indicating a similarity between the table side portions of the summary tables at different times is calculated using a table side element concatenated string obtained by concatenating the character strings of a plurality of table side elements constituting the table side portions of the summary tables at different times, or a table side portion string obtained by adding a character string of a table title to the table side element concatenated string, and the calculated table side portion similarity is compared with a preset threshold value. The summary tables at different times to be integrated are detected depending on whether or not both the table head similarity and the table side portion similarity are high.

[0039] Here, the "table head element concatenated string or the table head string obtained by adding the table head title string to the table head element concatenated string" in the "integration target detection means" may be only the table head element concatenated string, or may include both the table head element concatenated string and the table head title string, but the purpose is to exclude the table head title string alone. Similarly, the "table side element concatenated string or the table side part string obtained by adding the table side title string to the table side element concatenated string" in the "integration target detection means" may be only the table side element concatenated string, or may include both the table side element concatenated string and the table title string, but the purpose is to exclude the table side title string alone. The reason for excluding the table head title string alone or the table side title string alone is that in this third configuration, the inter-set similarity calculation means does not determine whether or not the summary tables can be integrated, so at the integration target detection stage by the integration target detection means, at least the table head element concatenated string or the table side element concatenated string is used to determine whether or not integration can be performed.

[0040] In such a summary table integration system (third configuration) of the present invention, an integration target detection means detects summary tables at different points in time to be integrated from among the multiple summary tables that exist for each of the different points in time, and also determines whether or not they can be integrated. Next, an inter-element similarity calculation means calculates the similarity between the top elements and the side elements of the detected summary tables to be integrated, determines the top elements and side elements whose semantics correspond, and the integration means creates an integrated summary table.

[0041] Therefore, the person performing the integration work does not need to judge the identity of the table top elements and table side elements, or judge whether the summary tables can be integrated, so that the integrated summary table can be easily created without time or effort. Also, if an appropriate threshold is set in advance (the threshold may be set by the person performing the integration work, or by someone else), the integration process can be executed without the intervention of human judgment, so that any worker can create a uniform integrated summary table regardless of the worker's knowledge or experience, thereby achieving the above-mentioned object.

[0042] <Configuration with division means>

[0043] In addition, in the above-described configurations (first, second, and third configurations), In the summary table file, a group of summary tables consisting of a plurality of summary tables for different time points is written, and as table of contents information for each of the summary table groups, a table head title constituting the table head, a table side title constituting the table side, and summary table identification information are written in association with each other for each summary table constituting the summary table group, a division means for dividing, when a combined summary table is included in the summary table group as a divisible summary table, the combined summary table into unit summary tables as summary tables that cannot be divided, using the table of contents information; This division means: In the table of contents information, by determining whether a plurality of table side titles are associated with the same table top title or a plurality of table top titles are associated with the same table side title in a state where they are associated with the same summary table identification information, it is determined whether a combined summary table is included in the summary table group, and if it is determined that a combined summary table is included, the combined summary table portion in the table of contents information is divided into unit summary tables associated with one table top title and one table side title in the table of contents information, The table of contents title similarity indicating the similarity of unit summary tables at the title level in the table of contents information is calculated using a concatenated title string obtained by concatenating the character strings of the top title and the side title of a unit summary table obtained by dividing a combined summary table included in a summary table group at one of the different points in time in the table of contents information, and a concatenated title string obtained by concatenating the character strings of the top title and the side title of a unit summary table obtained by dividing a combined summary table included in a summary table group at the other point in time in the table of contents information, or a summary table that is in the form of a unit summary table from the beginning, and the calculated table of contents title similarity is compared with a preset threshold value to determine whether or not there is a unit summary table with a high table of contents title similarity in the summary table group at the other point in time.If there is a unit summary table with a high table of contents title similarity, the data of the combined summary table included in the summary table group at one point in time is divided into data of the unit summary table in accordance with the division of the combined summary table into unit summary tables in the table of contents information.

[0044] When the configuration is provided with a splitting means in this manner, even when a combined tally table is included in the group of tally tables at each point in time and these combined tally tables cannot be integrated, or when a combined tally table is included in the group of tally tables at one point in time but not at the other point in time, it may be possible to integrate the tally tables by splitting the combined tally table into unit tally tables.

[0045] <Configuration in which the inter-element similarity calculation means swaps the top and side portions of the table vertically and horizontally>

[0046] Furthermore, in the above-mentioned configurations (first and second configurations), When it is determined by the inter-set similarity calculation means that it is not possible to integrate the summary tables at different times, The element similarity calculation means The similarity between the top and side elements may also be calculated for a state in which the top and side portions of a summary table at one of the different time points are swapped vertically and horizontally, and the calculated similarity between the top and side elements may be compared with respective preset threshold values ​​to determine the top and side elements whose semantics correspond between the summary tables at different time points.

[0047] In this manner, when the inter-element similarity calculation means is configured to swap the top and side parts of the table vertically and horizontally, even if the inter-set similarity calculation means determines that it is not possible to integrate summary tables from different points in time, there are cases in which integration is possible by swapping the top and side parts of the table vertically and horizontally.

[0048] <Configuration with inclusion relationship storage means>

[0049] In addition, in the above-described configurations (first, second, and third configurations), a storage unit for storing inclusion relationships that stores character strings that can be table top elements or table side elements and that are inclusive of each other and correspond to each other; The integration means are When the table head element or table side element of the summary table to be integrated contains both upper and lower character strings that are in an inclusive relationship with each other and are stored in the inclusion relationship storage means, the upper and lower character strings may be arranged adjacent to each other in the horizontal direction as the table head element of the integrated summary table, or adjacent to each other in the vertical direction as the table side element.

[0050] In a configuration equipped with such an inclusion relationship storage means, for example, when character strings in an inclusion relationship exist in the head or side elements of table tables at different times, such as when the Faculty of Science and Faculty of Engineering are the top or side elements at a certain point in time and the top or side element at a later point in time is the Faculty of Science and Engineering, or vice versa, it becomes possible to arrange the summary data for those elements in an easy-to-read manner in the integrated summary table.

[0051] <Program invention>

[0052] The program of the present invention is for causing a computer to function as the spreadsheet integration system (first, second, and third configurations) described above.

[0053] The above-mentioned program or a part of it can be recorded on a recording medium such as a magneto-optical disk (MO), a compact disk (CD), a digital versatile disk (DVD), a flexible disk (FD), a magnetic tape, a read-only memory (ROM), an electrically erasable and programmable read-only memory (EEPROM), a flash memory, a random access memory (RAM), a hard disk drive (HDD), a solid-state drive (SSD), a flash disk, etc., and can be stored or distributed, and can be transmitted using a transmission medium such as a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a wired network such as the Internet, an intranet, an extranet, a wireless communication network, or a combination of these, and can also be carried on a carrier wave. Furthermore, the above-mentioned program may be a part of another program, or may be recorded on a recording medium together with a separate program. Effect of the Invention

[0054] As described above, according to the present invention, each character string at the top and side of the table of a summary table, particularly each character string at the top and side elements of the table, is used to determine whether or not summary tables from different points in time (i.e., multiple points in time) can be integrated, or to detect items to be integrated, and to execute the integration process, thereby having the effect that any worker can easily integrate summary tables without incurring time and effort. [Brief description of the drawings]

[0055] [Figure 1] 1 is a diagram showing the overall configuration of a spreadsheet integration system according to an embodiment of the present invention; [Diagram 2] 6A and 6B are diagrams illustrating summary tables at two time points to be integrated in the embodiment. [Diagram 3] FIG. 4 is a diagram illustrating an integrated summary table according to the embodiment. [Figure 4] FIG. 4 is an explanatory diagram of the integration target detection process according to the embodiment. [Diagram 5] 5A to 5C are diagrams for explaining pre-designation, real-time designation, and automatic detection of integration targets according to the embodiment. [Figure 6] FIG. 13 is an exemplary diagram of a divisible combined spreadsheet according to the embodiment. [Figure 7] FIG. 4 is an explanatory diagram of a division process according to the embodiment. [Figure 8] 11A and 11B are explanatory diagrams of the integration process of table top elements or table side elements having an inclusion relationship in the embodiment. [Figure 9] FIG. 4 is a diagram showing the configuration of a summary table at two time points to be integrated in the embodiment. [Figure 10] 11A to 11C are diagrams illustrating a vertically long integrated type integration process of the summary table according to the embodiment. [Figure 11] 11A and 11B are diagrams illustrating a landscape-oriented integration process for a spreadsheet according to the embodiment. [Figure 12] 5A to 5C are explanatory diagrams of a calculation process of element similarity and set similarity in the embodiment. [Figure 13] FIG. 11 is a flowchart showing the flow (part 1) of the summary table integration process according to the embodiment. [Figure 14] FIG. 11 is a flowchart showing the flow (part 2) of the summary table integration process according to the embodiment. [Figure 15]FIG. 4 is a flowchart showing the flow of a division process according to the embodiment. [Figure 16] FIG. 1 is an example diagram of a conventional process for creating a summary table from raw data. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0056] An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows the overall configuration of a summary table integration system 10 of this embodiment. FIG. 2 is an example of summary tables at two points in time to be integrated, and FIG. 3 is an example of an integrated summary table. FIG. 4 is an explanatory diagram of an integration target detection process, and FIG. 5 is an explanatory diagram of advance designation, real-time designation, and automatic detection of an integration target. FIG. 6 is an example of a splittable merged summary table, and FIG. 7 is an explanatory diagram of a division process. FIG. 8 is an explanatory diagram of an integration process of table top elements or table side elements in an inclusion relationship, FIG. 9 is a configuration diagram of summary tables at two points in time to be integrated, and FIGS. 10 and 11 are explanatory diagrams of an integration process of a vertically integrated type summary table and a horizontally integrated type summary table. FIG. 12 is an explanatory diagram of a calculation process of an element similarity and a set similarity. FIGS. 13 and 14 show a flow chart of an integration process of a summary table, and FIG. 15 shows a flow chart of a division process.

[0057] (Summary of Spreadsheet Integration System 10) The tally table integration system 10 is a system that executes a process of creating an integrated tally table from multiple tally tables that have different creation times (survey time, data collection time) for the original raw data (see FIG. 16). The created integrated tally table displays comparative summary data (see FIG. 16) from different times. Note that, as already described in detail, when integrating tally tables from three or more times, the integration process for two times is repeated. Therefore, here, as shown in FIG. 2, a case will be described where a tally table from time Ta (e.g., 2020) is integrated with a tally table from time Tb (e.g., 2021).

[0058] As shown in FIG. 2, the survey objectives and survey methods of the summary table at time Ta and the summary table at time Tb are almost the same at each time (each year), but the survey items, etc. may differ slightly each time. This point has already been described in detail as [Problem to be solved by the invention], so a specific example will be described here. In the example of FIG. 2, the table title at the top of the table is different between "Q1 First, to you..." and "Q1 First, to you...". In addition, the table elements at the top of the table are different between "Chiba Prefecture, Tokyo" in the order and "Tokyo, Chiba Prefecture" in the reverse order. Furthermore, at time Tb, "Life Science Department" is added to the table side elements. In the case of such a difference, the integration targets at these two times are judged to be integrateable by the summary table integration system 10, and are integrated, and an integrated summary table as shown in FIG. 3 is automatically created.

[0059] In the example of Figure 2, elements have been added, but some elements may have been decreased, or some elements may have been added and some elements may have been decreased, and as long as there are some common elements, they can be integrated. In addition, not only the order of elements may be changed, but even if the character string of an element is changed, for example, "Izu Islands" and its Chinese translation "Izu Islands", they may be considered as common elements and a decision may be made as to whether they can be integrated.

[0060] In the example of FIG. 3, there is no aggregated data for the "Department of Life Science" for the time Ta (2020), so it is left blank. This blank (a blank in the integrated tally table) does not necessarily have to be blank, and may be a mark (for example, "-", "...", "*", etc.) indicating that there is no aggregated data to compare, and these are collectively referred to as "blank" in this application. In addition, in the example of FIG. 3, the integrated tally table is a vertically long integrated table in which the aggregated data to compare are arranged adjacent to each other vertically, but it may also be a horizontally long integrated table in which the aggregated data are arranged adjacent to each other horizontally, and details of this point will be described later with reference to FIG. 10 and FIG. 11.

[0061] In the example of FIG. 3, the reference time for creating the integrated summary table is set to time Tb (2021). This reference time for creating the integrated summary table can be freely selected and set by the user, as described later in the explanation of the setting means 28, but the default is a later time, time Tb. The idea is to adapt the old format to the new format. Therefore, at the top of the table of the created integrated summary table, the top title of time Tb (2021), "Q1 First, to you...", is adopted, and the order of the top elements of time Tb (2021), "Tokyo, Chiba Prefecture", is adopted.

[0062] (Overall configuration of summary table integration system 10) 1, the tally sheet integrating system 10 is configured with one or more computers, and is equipped with a main body 10A, input means 60 such as a mouse or keyboard connected to the main body 10A, and display means 70 such as a liquid crystal display. The main body 10A is equipped with a processing means 20 that executes a process of integrating tally sheets at different points in time (multiple points in time), and a storage means 40 that stores various data required for the execution of various processes by the processing means 20. In addition, although not shown, the system may be equipped with a printing device.

[0063] The processing means 20 includes an integration target advance designation means 21, an extraction means 22, an integration target detection means 23, an inter-element similarity calculation means 24, an inter-set similarity calculation means 25, an integration means 26, a division means 27, a setting means 28, and an inclusion relationship registration means 29. The inter-element similarity calculation means 24 includes a table head element similarity calculation means 24A and a table side element similarity calculation means 24B. The inter-set similarity calculation means 25 includes a table head element set similarity calculation means 25A, a table side element set similarity calculation means 25B, and an integration feasibility determination means 25C.

[0064] Here, each of the means 21 to 29 included in the processing means 20 is realized by a central processing unit (CPU) provided inside the main body 10A of the spreadsheet integrating system 10, one or more programs that define the operation procedures of this CPU, and a working memory such as a main memory and a cache memory. The details of each of these means 21 to 29 will be described later.

[0065] The storage means 40 is configured to include a summary table file storage means 41, an integration target pre-designation information storage means 42, an extraction result storage means 43, an inter-element similarity calculation result storage means 44, an integrated summary table storage means 45, an integrated summary table file storage means 46, a setting information storage means 47, and an inclusion relationship storage means 48.

[0066] Here, the storage means 40 may be implemented with, for example, a hard disk drive (HDD), a solid state drive (SSD), or the like, but may also be implemented with volatile memory (a storage area temporarily reserved in a working memory such as a main memory or a cache memory). This will be described later.

[0067] (Configuration of integration target advance designation means 21: Fig. 5) The integration target advance designation means 21 receives designation input by a user operating the input means 60 for combinations of summary tables at different points in time (two points in time in this example) to be integrated, and stores the received designation information in the integration target advance designation information storage means 42 (see FIG. 5). The designation information is combination information of summary table identification information (summary table IDs) for the summary tables at the two points in time to be integrated. Note that even if a user designates them as integration targets here, they are not necessarily integrated in reality, and only summary tables that are determined to be integrable by the integration feasibility determination means 25C of the inter-set similarity calculation means 25 are integrated.

[0068] In this embodiment, as shown in FIG. 5, there is a summary table file for a time point Ta (e.g., 2019) and a summary table file for a time point Tb (e.g., 2020) that is later than time point Ta, and multiple summary tables (summary table IDs = [N0001], [N0002], [N0003], ...) are written in each summary table file, so the specified information is a combination of the file identification information (file name) of the summary table file for one time point Ta and the summary table ID within that file, and the file identification information (file name) of the summary table file for the other time point Tb and the summary table ID within that file. Specifically, in the example of Figure 5, combination information of the summary table ID = [N0001] of the summary table file at time Ta and the summary table ID = [N0001] of the summary table file at time Tb, combination information of the summary table ID = [N0002] of the summary table file at time Ta and the summary table ID = [N0003] of the summary table file at time Tb, etc. are stored in the integration target pre-designated information storage means 42 as information designated by the user.

[0069] (Configuration of extraction means 22) The extraction means 22 reads a tally table file stored in the tally table file storage means 41, and executes a process of extracting data from the read tally table file (all tally tables including tally tables that are not ultimately subjected to processing including various similarity judgments as integration targets) written therein, distinguishing between the table head, table side, and tally data portions, using the setting information stored in the setting information storage means 47, and storing the extraction results in the extraction result storage means 43. Since this is an extraction process for executing subsequent processing without much time lag, the extraction result storage means 43 may be a main memory, but may also store the data in a non-volatile memory (HDD, SSD, etc.).

[0070] The tally table file is a spreadsheet software file (such as Microsoft Excel (registered trademark)). In this tally table file, data from multiple tally tables at the same survey time (multiple tally tables with the same raw data collection time) are written on the same sheet in the same file, for example, as shown in Fig. 5.

[0071] Note that the data of multiple tally sheets at the same time of investigation may be written not in the same sheet in the same file but distributed across multiple sheets in the same file. Also, the data of multiple tally sheets at the same time of investigation may be written not in the same file but distributed across multiple files, in which case the multiple tally sheet files are read in sequence to execute the extraction process.

[0072] Typically, the tally sheets are compiled into one or more tally sheet files for each survey point as described above. For example, since a similar survey is repeatedly conducted at each survey point, such as in 2020, 2021, etc., raw data (see FIG. 16) is created for each survey point, and a tally sheet is created for each survey point using the raw data. Therefore, for example, there are one or more files for 2020, one or more files for 2021, etc. However, when a post-facto compilation work is performed, tally sheets for multiple points may be written in one tally sheet file. In that case, for example, in the same file, there may be a sheet on which each tally sheet for time Ta is written and another sheet on which each tally sheet for time Tb is written, or, assuming that there is information identifying which tally sheet is for which point, the tally sheet for time Ta and the tally sheet for time Tb may be written in a mixed state in the same sheet.

[0073] Furthermore, the creation of a summary table using raw data (including the creation of a summary table file) is usually performed automatically by an existing summary table creation system, but it may also be created manually by a person. The summary tables to which the present invention is applicable may be those created automatically by a system or those created manually by a person.

[0074] Existing spreadsheet creation systems can be divided into several types depending on the contents of their programs (creation algorithms), and so there are also several types of spreadsheet files that are created. The type of spreadsheet file is determined by the system provider (a company that manufactures and sells the system, or an individual who provides it free of charge on the web, etc.), but even if the provider is the same, different versions can be considered to be different types.

[0075] Therefore, the data extraction process by the extraction means 22 for the table head, table side, and summary data section that make up each summary table uses a different algorithm for each type of summary table file, so the user selects the type of summary table file to be used and stores it in the setting information storage means 47 as setting information.

[0076] Specifically, for example, in the output file of the spreadsheet creation system provided by company α (spreadsheet file type α), the spreadsheet starts from the first column in the sheet, and in the output file of the spreadsheet creation system provided by company β (spreadsheet file type β), a blank line is inserted in the row following the top of the table, or a new spreadsheet starts when there is a character string in the first column after one blank line. The extraction algorithms for each type of spreadsheet file are described in each program, but numerical values ​​that can be output to the outside of the program and may be changed (numerical values ​​in column X, row Y, etc.) may be stored as setting information in the setting information storage means 47 in association with type identification information of the spreadsheet file (for example, company name such as company α or company β, developer name, version information), and may be read when the program is executed.

[0077] The extraction means 22 also displays the extraction results on the display means 70. At this time, in order to indicate the extracted locations (positional information within the matrix), the table head, table side, and tabulated data of each extracted tally table are displayed separately by painting them in different colors, framing them in different colors, or the like. This allows the user to check whether the extraction has been performed correctly by looking at the screen display. If the extracted locations (positional information within the matrix) are misaligned, the user can operate the input means 60 to correct the locations on the screen of the display means 70, i.e., while looking at the display screen. The corrected locations (positional information within each matrix of the table head, table side, and tabulated data) are accepted by the system and overwritten and saved in the extraction result storage means 43 as information under processing, and are stored in the setting information storage means 47 as setting information specific to the tally table file, that is, as setting information specific to the tally table file, not as setting information for each type of tally table file, in association with the file identification information (e.g., file name) of the tally table file. The reason why the information is stored not only in the extraction result storage means 43 but also in the setting information storage means 47 as setting information is that it can be used when the summary table file is processed again, and there is a possibility that the corrected information of the extraction results of the summary table file will be reflected in the setting information for that type of summary table file (in other words, there is a possibility that it will be upgraded and used not only for that summary table file).

[0078] Specifically, for example, when the position of the table top displayed in red is out of alignment with the actual position, the user performs a correction operation to make the table top position coincide with the actual position by dragging and moving the red display part on the screen, or by moving each edge of the top, bottom, right, and left to widen or narrow it. The system accepts the input of this correction operation, grasps the position information after the correction operation, and stores it in the extraction result storage means 43 (overwrites and saves it), and stores it in the setting information storage means 47 in association with the file identification information (e.g., file name) and the tally table ID (tally table identification information) of the tally table file as setting information. The position information after the correction operation to be stored here is the row number and column number that specify the respective positions of the table top, table side, and tally data part of the tally table, and since the areas of each of these parts have different widths and shapes, for example, the row number and column number of each cell in the upper left and lower right of the area. The position of the table side part displayed in yellow, for example, and the position of the tally data part displayed in green, for example, can also be corrected and stored in the same manner. The area of ​​the top of the table is extracted and corrected separately into the area of ​​the top element and the area of ​​the top title (which may not exist). The area of ​​the side of the table is extracted and corrected similarly, separately into the area of ​​the top element and the area of ​​the top title (which may not exist).

[0079] Furthermore, if a summary table does not actually exist but the top, side, and summary data parts of the extracted summary table are displayed as separate parts (by coloring or framing), these separate parts can be erased. On the other hand, if a summary table exists but no separate parts have been displayed as the extraction result, separate parts can be added to any position for the top, side, and summary data parts of the summary table, and the positions of the separate parts can be modified by the above-mentioned modification operations such as dragging.

[0080] (Configuration of integration target detection means 23: FIG. 4) The integration target detection means 23 executes a process of detecting summary tables for different points in time (times Ta and Tb) to be integrated from among the multiple summary tables that exist for each of the different points in time (times Ta and Tb) written in the summary table file (in the example of Figure 5, each point in time is a separate file, such as a file for time Ta and a file for time Tb).

[0081] Therefore, it is possible to automatically detect integration targets without the need for a user to specify them in advance or in real time using the integration target advance specification means 21. The summary tables at different time points to be integrated and automatically detected by the integration target detection means 23 may be further used as a target for determining whether or not integration is possible by the inter-set similarity calculation means 25 (the second configuration of the present invention described above) depending on the user's selection (this selection information is stored in the setting information storage means 47 as setting information), or may be actually integrated as is without further determination of whether or not integration is possible by the inter-set similarity calculation means 25 (the third configuration of the present invention described above).

[0082] Furthermore, whether or not automatic detection processing by the integration target detection means 23 is required is selected by the user, and is stored as setting information by the setting means 28 in the setting information storage means 47. When a selection is made that automatic detection processing is not required, the user himself / herself performs advance designation by the integration target advance designation means 21, or performs real-time designation.

[0083] The automatic detection process by this integration target detection means 23 includes (1) a process of linking the top and side of the table and making a similarity judgment, and (2) a process of separately making a similarity judgment of the top and side of the table and integrating the results of both judgments. Either (1) or (2) may be made user-selectable as setting information, or only one of the processes may be prepared.

[0084] In this embodiment, as shown in FIG. 5, a plurality of tally tables for different time points (time points Ta and Tb) are written in each of the tally table files for the corresponding time points, so the integration target detection means 23 calculates the similarity between each tally table in the tally table files for these two time points in a brute-force manner, and automatically detects the integration targets. In other words, when time Tb is set as the reference creation time, first the similarity is calculated between the summary table with summary table ID = [N0001] in the summary table file at time Tb and each of the summary tables with summary table IDs = [N0001], [N0002], [N0003], ... in the summary table file at the other time Ta, and then the similarity is calculated between the summary table with summary table ID = [N0002] in the summary table file at time Tb and each of the summary tables with summary table IDs = [N0001], [N0002], [N0003], ... in the summary table file at the other time Ta, ... and so on to automatically detect combinations of summary tables with high similarity (more precisely, similarity of the top and side parts of the tables).

[0085] 4, the table head of any of the schedule tables A in the schedule file at time Ta is AX, the table head title is AXT, the table head element is AXi (i=1 to N1, N1 is the number of elements of the table head element), the table side part is AY, the table side title is AYT, and the table side element is AYj (j=1 to N2, N2 is the number of elements of the table side element). Also, the table head of any of the schedule tables B in the schedule file at time Tb is BX, the table head title is BXT, the table head element is BXf (f=1 to N3, N3 is the number of elements of the table head element), the table side part is BY, the table side title is BYT, and the table side element is BYg (g=1 to N4, N4 is the number of elements of the table side element).

[0086] In the process of (1), the integration target detection means 23 first prepares a table head element concatenated string ΣAXi (i=1-N1), ΣBXf (f=1-N3) obtained by concatenating the strings of the multiple table head elements AXi (i=1-N1), BXf (f=1-N3) constituting the table head AX, BX for each of the summary tables A and B at different times (times Ta, Tb) and / or a table head string including the string of the table head title AXT, BXT. At this time, when it is selected to omit the judgment of whether or not integration is possible by the inter-set similarity calculation means 25, it is preferable to include the table head element concatenated string in order to increase the accuracy of the similarity judgment by this integration target detection means 23, that is, it is preferable to avoid using only the table head title, and the system may be configured so that it is not possible to select to use only the table head title. On the other hand, when it is selected to judge whether or not integration is possible by the inter-set similarity calculation means 25, the accuracy of the similarity judgment by this integration target detection means 23 does not need to be so high, so only the table head title may be used. If there is no table top title, the table top element concatenated character string will inevitably be used.

[0087] Therefore, in the process of (1), for the table head AX, prepare ΣAXi (i = 1 to N1) as the table head character string, prepare {ΣAXi (i = 1 to N1) + AXT}, or prepare only AXT. However, if the determination of whether or not integration is possible by the inter-set similarity calculation means 25 is omitted, it is preferable to avoid using only AXT. Also, for the table head BX, prepare ΣBXf (f = 1 to N3), prepare {ΣBXf (f = 1 to N3) + BXT}, or prepare only BXT. However, if the determination of whether or not integration is possible by the inter-set similarity calculation means 25 is omitted, it is preferable to avoid using only BXT.

[0088] In the process of (1), the integration target detection means 23 prepares, for each of the summary tables A and B at different times (times Ta and Tb), a table side part string including a table side element concatenated string ΣAYj (j=1-N2), ΣBYg (g=1-N4) obtained by concatenating the strings of the multiple table side elements AYj (j=1-N2), BYg (g=1-N4) constituting the table side parts AY and BY, and / or a table side title AYT, BYT. In this case, if it is selected to omit the judgment of whether or not integration is possible by the inter-set similarity calculation means 25, it is preferable to include the table side element concatenated string as in the case of the table head described above, that is, it is preferable to avoid using only the table side title, and the system may be configured so that it is not possible to select to use only the table side title. Note that, if there is no table side title, the table side element concatenated string will inevitably be used.

[0089] Therefore, in the process of (1), for the front side part AY, ΣAYj (j = 1 to N2) or {ΣAYj (j = 1 to N2) + AYT} or only AYT is prepared as the front side part character string. However, if the determination of whether or not integration is possible by the inter-set similarity calculation means 25 is omitted, it is preferable to avoid using only AYT. Also, for the front side part BY, ΣBYg (g = 1 to N4) or {ΣBYg (g = 1 to N4) + BYT} or only BYT is prepared. However, if the determination of whether or not integration is possible by the inter-set similarity calculation means 25 is omitted, it is preferable to avoid using only BYT.

[0090] Then, in process (1), for each of summary tables A and B at different points in time (points Ta and Tb), a table head-side concatenated string is created by concatenating the table head string and table side string prepared in this manner.

[0091] Next, in the process of (1), the integration target detection means 23 uses the created table head / side part concatenated strings of the summary tables A and B at different times (times Ta and Tb) to calculate a table head / side part similarity indicating the similarity between the table head / side part concatenated string for the table head AX and the table side AY of the summary table A at time Ta and the table head / side part concatenated string for the table head BX and the table side BY of the summary table B at time Tb by an algorithm different from the algorithm for calculating similarity used by the set similarity calculation means 25, compares the calculated table head / side part similarity with a preset threshold (stored in the setting information storage means 47 as setting information), and detects the summary tables A and B at different times (times Ta and Tb) to be integrated depending on whether the table head / side part similarity is high or not. That is, when the similarity is high (when the similarity is equal to or exceeds the threshold), the pair of summary tables A and B is detected as an integration target.

[0092] In the process of (2), the integration target detection means 23 first calculates a table head similarity indicating the similarity between the table head AX of the summary table A at time Ta and the table head BX of the summary table B at time Tb by using a table head element concatenated strings ΣAXi (i = 1 to N1), ΣBXf (f = 1 to N3) obtained by concatenating the strings of the multiple table head elements AXi (i = 1 to N1), BXf (f = 1 to N3) constituting the table heads AX, BX of the summary tables A, B at different times (times Ta, Tb) and / or a table head string including the strings of the table head titles AXT, BXT, by an algorithm different from the similarity calculation algorithm used by the inter-set similarity calculation means 25. Then, the calculated table head similarity is compared with a preset threshold value (stored in the setting information storage means 47 as setting information). In this case, if it is selected to omit the determination of whether or not integration is possible by the inter-set similarity calculation means 25, it is preferable to include the head element concatenated character string as in the above-mentioned process (1), that is, it is preferable to avoid using only the head title, and the system may be configured so that it is not possible to select using only the head title. Note that if there is no head title, the head element concatenated character string will inevitably be used.

[0093] In addition, in the process of (2), the integration target detection means 23 calculates a table side portion similarity indicating the similarity between the table side portion AY of the summary table A at time Ta and the table side portion BY of the summary table B at time Tb by an algorithm different from the similarity calculation algorithm used by the inter-set similarity calculation means 25, using table side element concatenated strings ΣAYj (j = 1 to N2), ΣBYg (g = 1 to N4) obtained by concatenating the strings of each of the multiple table side elements AYj (j = 1 to N2), BYg (g = 1 to N4) that constitute the table side portions AY, BY of the summary tables A, B at different times (times Ta, Tb) and / or table side portion strings including the strings of the table titles AYT, BYT, and compares the calculated table side portion similarity with a predetermined threshold value (stored in the setting information storage means 47 as setting information). In this case, if it is selected to omit the determination of whether or not integration is possible by the inter-set similarity calculation means 25, it is preferable to include the front side element concatenated string as in the case of the above-mentioned process (1), that is, it is preferable to avoid using only the front side title, and the system may be configured so that it is not possible to select using only the front side title. Note that if there is no front side title, the front side element concatenated string will inevitably be used.

[0094] Next, in the process of (2), the integration target detection means 23 detects the summary tables A and B at different times (times Ta and Tb) to be integrated depending on whether or not both the table head similarity and the table side similarity are high in the above two comparisons. That is, when both similarities are high (when both similarities are equal to or higher than their respective thresholds or exceed the thresholds), the pair of summary tables A and B are detected as integration targets. The threshold for the table head similarity and the threshold for the table side similarity may be set to different values. When the threshold for the table head similarity is set higher, differences in the table side (differences in the number of table side elements, etc.) are relatively tolerated, whereas when the threshold for the table side similarity is set higher, differences in the table head (differences in the number of table head elements, etc.) are relatively tolerated.

[0095] In the above process (1), as an algorithm for similarity calculation of the integration target detection means 23, for example, a method can be adopted in which the head and side concatenated character strings of the summary tables A and B are used to obtain text vectors of the summary tables A and B by Doc2Vec, and the cosine similarity or the like of these text vectors is used as the head and side similarity. Also, in the process (2), a method can be adopted in which the head character strings of the summary tables A and B are used to obtain text vectors of the summary tables A and B by Doc2Vec, and the cosine similarity or the like of these text vectors is used as the head similarity, and the side character strings of the summary tables A and B are used to obtain text vectors of the summary tables A and B by Doc2Vec, and the cosine similarity or the like of these text vectors is used as the side similarity. In this embodiment, these methods are set as default algorithms, for example.

[0096] The algorithm for calculating the similarity of the integration target detection means 23 is not limited to the above-mentioned Doc2Vec method, and may be other methods as long as the purpose can be achieved. For example, in the process of (1), a method can be adopted in which word importance vectors such as TF-IDF vectors for each table top / table side concatenated character string of the summary tables A and B are obtained, and the cosine similarity of these word importance vectors is used as the table top / table side similarity. In the process of (2), a method can be adopted in which word importance vectors such as TF-IDF vectors for each table top character string and each table side character string of the summary tables A and B are obtained, and the cosine similarity of these word importance vectors is used as the table top similarity and table side similarity.

[0097] In this case, when determining the word importance vector, morphological analysis is performed on the head and side table concatenated strings, or the head and side table strings, of not only the pair of summary tables A and B to be compared, but also all summary tables in all summary table files (not only for two time points, but all summary tables at all time points), to decompose them into words, and any unnecessary symbols that should be removed are removed. Using all the remaining words (which may be only nouns), a word importance vector consisting of the TF-IDF value or other word importance index value of each word for each summary table (the head and side table concatenated strings, or the head and side table strings of each summary table), is calculated, and the obtained word importance vector is stored in a word importance vector storage means (not shown) in association with the file identification information (file name, etc.) of the summary table file and the summary table ID (summary table identification information).

[0098] Here, the word importance index value is not limited to the TF-IDF value, and may be, for example, Okapi-BM25. TF (Term Frequency) is the frequency of occurrence of a word in a document (here, the top and side concatenated strings of a summary table, or the top and side strings of the table), and is the value obtained by dividing the number of occurrences of each word in a string by the sum of the number of occurrences of all words in the string. IDF (Inverse Document Frequency) is the inverse document frequency, and is the logarithm of the value obtained by dividing the total number of documents (here, the total number of top and side concatenated strings of a summary table, or the total number of top and side strings of the table) by the number of strings containing each word. The word importance vector is a p-dimensional vector, where p is the number of words.

[0099] (Configuration of the element similarity calculation means 24: FIG. 12) As shown in FIG. 12, the table head element similarity calculation means 24A of the inter-element similarity calculation means 24 calculates table head element similarities indicating the similarity between character strings of each of the multiple table head elements AXi (i = 1 to N1), BXf (f = 1 to N3) that constitute the table heads AX, BX of each of the summary tables A, B at different times (times Ta, Tb) to be integrated that are written in the summary table file, and by comparing the calculated table head element similarity with a preset threshold value (stored in the setting information storage means 47 as setting information), determines table head elements (AXn, BXm) whose meanings correspond between the summary tables A, B at different times (times Ta, Tb), and stores the determination results in the inter-element similarity calculation result storage means 44. This determination result is temporarily stored for use in subsequent processing by the set similarity calculation means 25, so the element similarity calculation result storage means 44 may be a main memory, but it may also be stored in a non-volatile memory such as an HDD or SSD.

[0100] As shown in FIG. 12, the surface side inter-element similarity calculation means 24B of the inter-element similarity calculation means 24 calculates surface side inter-element similarities indicating the similarity between strings of each of the multiple surface side elements AYj (j = 1 to N2) and BYg (g = 1 to N4) that constitute the surface side sections AY and BY of the summary tables A and B at different times (times Ta and Tb) to be integrated that are written into the summary table file, and by comparing the calculated surface side inter-element similarities with a predetermined threshold value (stored in the setting information storage means 47 as setting information), it determines surface side elements (AYh, BYk) whose meanings correspond between the summary tables A and B at different times (times Ta and Tb), and stores the determination results in the inter-element similarity calculation result storage means 44.

[0101] At this time, the table head element similarity calculation means 24A calculates the table head element similarity by exhaustively searching multiple table head elements AXi (i = 1 to N1) of the table head AX and multiple table head elements BXf (f = 1 to N3) of the table head BX, and when the table head element similarity is high (above or above a threshold), it determines the pair of table head elements AXi and BXf at that time as table head elements (AXn, BXm) whose semantic contents correspond between the summary tables A and B at different times (times Ta and Tb). However, even if it is called exhaustive, it is not necessary to calculate the table head element similarity for common elements that have already been determined as table head elements whose semantic contents correspond. Note that AXn and BXm are not symbols that indicate a specific table head element, but symbols that represent all table head elements as common elements whose semantic contents correspond (the symbol of any one table head element among the table head elements as common elements).

[0102] Specifically, for example, when the time Tb is the creation reference time, first, the similarity between the table head element BX1 of the table head BX and the table head elements AX1, AX2, AX3, ... of the table head AX is calculated, and then the similarity between the table head element BX2 of the table head BX and the table head elements AX1, AX2, AX3, ... of the table head AX is calculated. At this time, if it has already been determined that BX1 and AX1 are common elements whose meanings correspond, it is sufficient to calculate the similarity between the table head element BX2 of the table head BX and the table head elements AX2, AX3, AX4, ... of the table head AX, excluding AX1. Next, if it has already been determined that BX2 and AX2 are common elements, it is sufficient to calculate the similarity between the table head element BX3 of the table head BX and the table head elements AX3, AX4, AX5, ... of the table head AX, excluding AX2. The same process is repeated for the table head element BX4 of the table head BX and onwards.

[0103] Similarly, the surface element similarity calculation means 24B calculates the surface element similarity between multiple front elements AYj (j=1 to N2) of the surface side part AY and multiple front elements BYg (g=1 to N4) of the surface side part BY by exhaustive calculation, and when the surface element similarity is high (above or above a threshold value), the pair of surface elements AYj, BYg at that time are determined as surface elements (AYh, BYk) whose semantic contents correspond between the summary tables A and B at different times (times Ta, Tb). However, even if it is called exhaustive calculation, it is not necessary to calculate the surface element similarity for common elements that have already been determined as surface elements whose semantic contents correspond, as in the case of the surface element similarity calculation means 24A described above. Note that AYh, BYk are not symbols that indicate a specific one surface element, but symbols that represent all of the surface elements as common elements whose semantic contents correspond (the symbol of any one surface element among the surface elements as common elements).

[0104] The threshold value for the similarity between front head elements and the threshold value for the similarity between front side elements are set as setting information by the setting means 28, and it is possible to set both threshold values ​​to different values. However, in the case of similarity between these elements, unlike when determining overall similarity as in the integration target detection means 23 or the inter-set similarity calculation means 25, the similarity between short character strings is determined, so they may normally be set to the same value.

[0105] As an algorithm for calculating similarity by the inter-element similarity calculation means 24, for example, the Levenshtein distance can be used. Note that, if the distance value is large, the similarity is low, and if the distance value is small, the similarity is high. Therefore, if the distance value is equal to or less than the threshold value, the similarity is determined to be high. The Levenshtein distance is a distance represented by the minimum edit distance between one character string and another character string. In this embodiment, as an example, the Levenshtein distance is adopted as the default for the similarity between table top elements and the similarity between table side elements.

[0106] In addition, the algorithm for calculating similarity by the inter-element similarity calculation means 24 is not limited to the above-mentioned Levenshtein distance method, and other methods may be used as long as the purpose can be achieved. For example, the Jaro-Winkler distance, the Hamming distance, etc. may be used.

[0107] Furthermore, a morphological analysis may be performed on the character strings of each head element and each side element, each character string may be broken down into words, and a word vector may be obtained using Word2vec or the like for each word that constitutes the character strings in one (any one) head element or side element, and a representative word vector (averaged word vector or the like) may be obtained for each head element or side element by calculating an average vector or the like of the word vectors for each word that constitutes each character string in each head element or side element, and the cosine similarity or the like of the obtained representative word vector may be used as the similarity between head elements and side elements. Note that if the character string of an element is one word, the average or the like cannot be calculated, so the word vector for that one word itself becomes the representative word vector.

[0108] (Configuration of the set similarity calculation means 25: FIG. 12) As shown in FIG. 12, the table head element set similarity calculation means 25A of the inter-set similarity calculation means 25 regards table head elements (AXn, BXm) whose semantic contents correspond between summary tables A and B at different times (times Ta, Tb) determined by the table head element inter-similarity calculation means 24A of the element inter-element similarity calculation means 24 as the same element, calculates a table head element set similarity indicating the similarity between a table head element set consisting of a plurality of table head elements AXi (i = 1 to N1) constituting the table head AX and a table head element set consisting of a plurality of table head elements BXf (f = 1 to N3) constituting the table head BX, compares the calculated table head element set similarity with a preset threshold value (stored in the setting information storage means 47 as setting information), and executes a process of determining that the table head elements are similar as a set if the similarity is high (greater than or equal to the threshold value).

[0109] At this time, as shown in Fig. 12, the table element set consisting of a plurality (N1) table element AXi (i = 1 to N1) constituting the table head AX is divided into N5 common elements AXn determined as table head elements having corresponding semantic contents between the table heads AX and BX (there is a corresponding table head element in the other table head), and (N1-N5) non-common elements AXu determined as table head elements having no corresponding semantic contents between the table heads AX and BX (there is no corresponding table head element in the other table head), as determined by the table head element inter-element similarity calculation means 24A. In the example of Fig. 12, the table head elements determined as common elements AXn are AX1, AX2, AX3, ..., and the table head elements determined as non-common elements AXu are AX13, AX14, ....

[0110] As shown in Fig. 12, the table head element set consisting of a plurality (N3) of table head elements BXf (f = 1 to N3) constituting the table head BX is divided into N5 common elements BXm determined as table head elements having a corresponding meaning between the table heads AX and BX, and (N3-N5) non-common elements BXv determined as table head elements having no corresponding meaning between the table heads AX and BX, by the determination of the table head element set similarity calculation means 24A. In the example of Fig. 12, the table head elements determined as the common elements BXm are BX1, BX2, BX3, ..., and the table head elements determined as the non-common elements BXv are BX10, BX11, .... Therefore, the table head element set similarity calculation means 25A calculates the table head element set similarity by utilizing the situation between the table head element sets shown in Fig. 12.

[0111] In the example of Figure 12, "BXm(AXn)" is written for the N5 common elements, which means that AXn and BXm may be completely identical or may differ slightly, and if they differ slightly, the character string BXm will be adopted as the notation for the integrated tally table, assuming that the later point in time Tb is the reference point for creating the integrated tally table, as a rule. However, if there is space, they may be written in parallel.

[0112] As shown in FIG. 12, the surface side element set similarity calculation means 25B of the inter-set similarity calculation means 25 regards surface side elements (AYh, BYk) whose semantic contents correspond between summary tables A and B at different times (times Ta, Tb) determined by the surface side element inter-similarity calculation means 24B of the inter-element similarity calculation means 24 as the same element, calculates a surface side element set similarity indicating the similarity between a surface side element set consisting of a plurality of surface side elements AYj (j = 1 to N2) constituting the surface side part AY and a surface side element set consisting of a plurality of surface side elements BYg (g = 1 to N4) constituting the surface side part BY, compares the calculated surface side element set similarity with a predetermined threshold value (stored in the setting information storage means 47 as setting information), and executes a process of determining that the surface side elements are similar as a set if the similarity is high (greater than or equal to the threshold value).

[0113] In this case, as shown in Fig. 12, the surface element set consisting of a plurality (N2) of surface elements AYj (j = 1 to N2) constituting the surface part AY is divided into N6 common elements AYh determined as surface elements having corresponding semantic contents between the surface parts AY and BY (corresponding surface elements are present in the surface part of the other side) and (N2-N6) non-common elements AYr determined as surface elements having no corresponding semantic contents between the surface parts AY and BY (no corresponding surface element is present in the surface part of the other side) by the determination of the surface element set by the surface element similarity calculation means 24B. In the example of Fig. 12, the surface elements determined as common elements AYh are AY1, AY2, AY3, ..., and the surface elements determined as non-common elements AYr are AY9, AY10, ....

[0114] As shown in Fig. 12, the surface element set consisting of a plurality (N4) of surface elements BYg (g = 1 to N4) constituting the surface part BY is divided into N6 common elements BYk determined as surface elements having a corresponding meaning between the surface parts AY and BY, and (N4-N6) non-common elements BYs determined as surface elements having no corresponding meaning between the surface parts AY and BY by the determination of the surface element inter- similarity calculation means 24B. In the example of Fig. 12, the surface elements determined as the common elements BYk are BY1, BY2, BY3, ..., and the surface elements determined as the non-common elements BYs are BY11, BY12, .... Therefore, the surface element set similarity calculation means 25B calculates the surface element set similarity by utilizing the situation between the surface element sets shown in Fig. 12.

[0115] In the example of Figure 12, "BYk(AYh)" is written for the N6 common elements, which means that AYh and BYk may be completely identical or may differ slightly, and if they differ slightly, the character string BYk will be adopted as the notation for the integrated tally table, assuming that the later point in time Tb is the reference point for creating the integrated tally table, as a rule. However, if there is space, they may be written in parallel.

[0116] The integration feasibility determination means 25C of the inter-set similarity calculation means 25 executes a process of determining whether or not the summary tables A and B at different points in time (points Ta and Tb) can be integrated, depending on whether or not both the similarity between the table top element sets and the similarity between the table side element sets are high. In other words, if both similarities are high (if both similarities are equal to or greater than their respective thresholds), it is determined that the summary tables A and B can be integrated.

[0117] Note that the threshold for the similarity between front element sets and the threshold for the similarity between front side element sets may be set to different values. When the threshold for the similarity between front element sets is set higher, differences in the front side (such as differences in the number of front side elements) are relatively more tolerated. On the other hand, when the threshold for the similarity between front side element sets is set higher, differences in the front heads (such as differences in the number of front head elements) are relatively more tolerated.

[0118] Moreover, whether or not to perform the judgment process by the inter-set similarity calculation means 25 is stored as setting information in the setting information storage means 47. When the selection setting is made to omit the judgment process by the inter-set similarity calculation means 25, it is necessary to perform the detection process of the integration target by the integration target detection means 23 as a substitute, so the selection setting is made to require the detection of the integration target by the integration target detection means 23. Note that it is also possible to select the selection setting to perform the processes of both the integration target detection means 23 and the inter-set similarity calculation means 25 (second configuration of the present invention), in which case, after the detection process of the integration target by the integration target detection means 23 is performed, the detection process of the integration target is performed by the inter-element similarity calculation means 24 for the detected integration target, and then the inter-set similarity calculation means 25 performs the judgment process of whether or not to integrate.

[0119] As an algorithm for similarity calculation by the inter-set similarity calculation means 25, for example, the Jaccard coefficient can be adopted. The Jaccard coefficient represents the ratio of common elements to the elements contained in two sets, and the larger the value of the coefficient, the higher the similarity between the two sets. In other words, the more common elements there are, the higher the similarity between the sets. In this embodiment, as an example, the Jaccard coefficient is set as a default.

[0120] 12, the Jaccard coefficient for the table head element set similarity calculated by the table head element set similarity calculation means 25A is a value obtained by dividing the number N5 of common elements BXm (AXn) of the table heads AX and BX by the total number of elements of the table heads AX and BX (N1+N3-N5: common elements AXn and BXm are not counted twice, but are counted as the same element.) For the table side element set similarity calculated by the table side element set similarity calculation means 25B, the Jaccard coefficient for the table side element set similarity calculated by the table side element set similarity calculation means 25B is a value obtained by dividing the number N6 of common elements BYk (AYh) of the table side parts AY and BY by the total number of elements of the table side parts AY and BY (N2+N4-N6: common elements AYh and BYk are not counted twice, but are counted as the same element.).

[0121] However, the algorithm for calculating the similarity of the set similarity calculation means 25 is not limited to the method using the Jaccard coefficient, and other methods may be used as long as the purpose can be achieved. For example, the Dice coefficient, the Simpson coefficient, etc. may be used.

[0122] (Configuration of integration means 26: Figs. 9, 10, 11, and 12) The integration means 26 is configured to extract data (each data constituting the aggregate data portion (see FIG. 16 )) of the summary tables A and B at different times (times Ta and Tb) that are determined to be integrable by the inter-set similarity calculation means 25, or the summary tables A and B at different times (times Ta and Tb) that are detected as integration targets by the integration target detection means 23 when the inter-set similarity calculation means 25 is set to omit the determination of integration feasibility, for both the table head elements AX and BX and the table side elements AY and BY. When the semantic contents of summary tables A and B at times Ta and Tb correspond to each other, an integration process is performed to create an integrated summary table by arranging the summary data for the table head elements AXn and BXm and the table side elements AYh and BYk (A(h,n), B(k,m) in FIG. 9) adjacent to each other vertically (see FIG. 10) or horizontally (see FIG. 11), and then a process is executed to store the created integrated summary table in integrated summary table storage means 45 in association with automatically or manually assigned summary table identification information (summary table ID). Information on whether to select the vertical or horizontal orientation is stored as setting information in setting information storage means 47.

[0123] More specifically, as shown in FIG. 12, as a result of the determination by the inter-element similarity calculation means 24, the table top elements AXi (i = 1 to N1), BXf (f = 1 to N3) and table side elements AYj (j = 1 to N2), BYg (g = 1 to N4) of the summary tables A and B to be integrated are each divided into common elements and non-common elements, and therefore the integration means 26 performs integration processing by utilizing the situation shown in FIG. 12.

[0124] 9, in the schedule A at the top of the figure, the aggregated data of the cell where the row of the table side element AYh as a common element intersects with the column of the table top element AXn as a common element is A(h,n). Meanwhile, in the schedule B at the bottom of the figure, the aggregated data of the cell where the row of the table side element BYk as a common element intersects with the column of the table top element BXm as a common element is B(k,m). In this case, the aggregated data A(h,n) of schedule A and the aggregated data B(k,m) of schedule B are adjacently arranged aggregated data.

[0125] 9, the aggregate data of a cell where the row of the table side element AYh as a common element intersects with the column of the table top element AXu as a non-common element is A(h,u). The aggregate data of a cell where the row of the table side element AYr as a non-common element intersects with the column of the table top element AXn as a common element is A(r,n). The aggregate data of a cell where the row of the table side element AYr as a non-common element intersects with the column of the table top element AXu as a non-common element is A(r,u).

[0126] Similarly, in the bottom summary table B in Fig. 9, the aggregated data of the cell where the row of the table side element BYk as a common element intersects with the column of the table top element BXv as a non-common element is B(k,v). The aggregated data of the cell where the row of the table side element BYs as a non-common element intersects with the column of the table top element BXm as a common element is B(s,m). The aggregated data of the cell where the row of the table side element BYs as a non-common element intersects with the column of the table top element BXv as a non-common element is B(s,v).

[0127] In the table top AX of the upper summary table A in Fig. 9, the order of the table top element AXn as a common element and the table top element AXu as a non-common element is not limited to this, and since AXn and AXu are representative elements (any one of the corresponding table top elements) of the table top elements as the common element and the non-common element, respectively, both may be mixed. In addition, in the table top BX of the lower summary table B in Fig. 9, the order of the table top element BXm as a common element and the table top element BXv as a non-common element is not limited to this, and since they are representative elements, both may be mixed. Therefore, the order of BXm (AXn), AXu, and BXv in the table top MX of the integrated summary table M in Figs. 10 and 11 is also not limited to this, and the three may be mixed.

[0128] Similarly, in the table side part AY of the upper schedule A in Fig. 9, the order of the table side element AYh as a common element and the table side element AYr as a non-common element is not limited to this, and since AYh and AYr are representative elements of the table side elements as the common element and the non-common element, respectively (any one of the corresponding table side elements), both may be mixed. Also, in the table side part BY of the lower schedule B in Fig. 9, the order of the table side element BYk as a common element and the table side element BYs as a non-common element is not limited to this, and since they are representative elements, both may be mixed. Therefore, the order of BYk (AYh), AYr, and BYs in the table side part MY of the integrated schedule M in Figs. 10 and 11 is also not limited to this, and the three may be mixed.

[0129] First, for the table head title MXT of the table head MX of the integrated summary table M, in either the vertical integration process (see FIG. 10) or the horizontal integration process (see FIG. 11), the integration means 26 adopts the character string of the table head title AXT of the table head AX of the summary table A or the table head title BXT of the table head BX of the summary table B, whichever is set as the reference time of creation (the default is the later time Tb), but they may be written in parallel. Also, for the table side title MYT of the table side part MY, the table side title AYT of the table side part AY of the summary table A or the table side title BYT of the table side part BY of the summary table B, whichever is set as the reference time of creation (the default is the later time Tb), but they may be written in parallel. Note that the table head title MXT and the table side title MYT are not necessary.

[0130] Next, the integration means 26 adopts the character string that is the creation reference time (the later time Tb by default) for the table head elements AXn and BXm as common elements whose semantic contents correspond between the summary tables A and B in either the vertical integration process (see FIG. 10) or the horizontal integration process (see FIG. 11), but may write them in parallel. On the other hand, for the table head element AXu (there is no corresponding table head element in the table head BX of the other party) and table head element BXv (there is no corresponding table head element in the table head AX of the other party) as non-common elements that do not have a corresponding table head element in the table head of the other party, the integration means 26 uses the character strings AXu and BXv as they are, since there is no character string on the other party's side. As already mentioned, the order of the three elements BXm (AXn), AXu, and BXv is not limited to this order, but the order of each table head element in the table head MX of the integrated summary table M is adjusted to the order of the creation reference time (the later time Tb by default). This is because the table head element similarity calculation means 24A calculates the table head element similarity in a brute force manner regardless of the order of the table head elements to search for corresponding table head elements (see FIG. 12), so a criterion for determining the order after integration is required.

[0131] In addition, the integration means 26 adopts the character string that is the creation reference time (the later time Tb by default) for the notation of the table side elements AYh and BYk as common elements whose semantic contents correspond between the schedules A and B in either the vertical integration process (see FIG. 10) or the horizontal integration process (see FIG. 11), but may write them in parallel. On the other hand, for the notation of the table side element AYr (there is no corresponding table side element in the table side part BY of the other party) and the table side element BYs (there is no corresponding table side element in the table side part AY of the other party) as non-common elements that have no corresponding table side element in the table side part of the other party, the character strings AYr and BYs of the table side element AY are used as they are, since there is no character string on the other party. As already mentioned, the order of the three elements BYk (AYh), AYr, and BYs is not limited to this order, but the order of the table side elements in the table side part MY of the integrated schedule M is adjusted to the order of the creation reference time (the later time Tb by default). This is because the surface element similarity calculation means 24B calculates the surface element similarity in a brute-force manner regardless of the order of the surface elements, and searches for corresponding surface elements (see Figure 12), so a criterion for determining the order after integration is required.

[0132] Furthermore, the integration means 26 arranges each piece of aggregated data constituting the aggregated data part of the integrated summary table M as follows. An explanation will be given using the aggregated data defined in Fig. 9. With regard to aggregated data A(h,n) at time Ta and aggregated data B(k,m) at time Tb, in which both the table top element and table side element are common elements, they are arranged adjacent to each other vertically in the vertical integration type integration process shown in Fig. 10, and arranged adjacent to each other horizontally in the horizontal integration type integration process shown in Fig. 11.

[0133] More specifically, in the vertical integration type integration process shown in Fig. 10, corresponding aggregated data (aggregated data for two points in time to be compared) at different points in time (points Ta and Tb) are arranged vertically, so that the integrated summary table M created has a row for point in time Ta and a row for point in time Tb arranged vertically. Note that either the row for point in time Ta or the row for the subsequent point in time Tb can be the upper row. Therefore, the arrangement of each aggregated data in the integrated summary table M in Fig. 10 will be viewed vertically.

[0134] First, looking vertically from the top of the column of the table top element BXm (AXn) as a common element, the rows of the table side element BYk (AYh) as a common element are provided with a row of time Ta and a row of time Tb, and there are two cells that intersect with these two rows. And since both the table top element BXm (AXn) and the table side element BYk (AYh) are common elements, the aggregated data A(h,n) and B(k,m) are placed in the two cells aligned vertically.

[0135] Below that, a row for time Ta is provided as a row for the surface element AYr, which is a non-common element, but because the surface element AYr is a non-common element, a row for time Tb is not provided, and so there is one cell that intersects with the row for time Ta, and the aggregated data A(r,n) is placed there.

[0136] Further below that, a row for time Tb is provided as a row for the surface element BYs, which is a non-common element, but because the surface element BYs is a non-common element, a row for time Ta is not provided, and so there is one cell that intersects with the row for time Tb, and the aggregated data B(s,m) is placed there.

[0137] Next, looking vertically from top to bottom at the column of the table top element AXu as a non-common element, there are two cells that intersect with the two rows of times Ta and Tb for the table side element BYk (AYh) as a common element, and the aggregated data A(h, u) for time Ta is placed therein. However, since the table top element AXu is a non-common element, there is no aggregated data for time Tb, so these are left blank.

[0138] Below that, there is one cell that intersects with the row of time Ta for the table side element AYr as an uncommon element, and the aggregated data A(r,u) is placed there. Although both the table top element AXu and the table side element AYr are uncommon elements, they are both elements at the same time Ta, so aggregated data for time Ta exists.

[0139] Further below that, there is one cell that intersects with the row of time Tb for the table side element BYs as an uncommon element, but no aggregate data is placed there and it is left blank because both the table top element AXu and the table side element BYs are uncommon elements and are not elements of the same time.

[0140] Next, looking vertically from top to bottom at the column of table top element BXv as a non-common element, there are two cells that intersect with the two rows of time points Ta and Tb for table side element BYk (AYh) as a common element, and the aggregated data B(k,v) for time point Tb is placed therein. However, since table top element BXv is a non-common element, there is no aggregated data for time point Ta, so these are left blank.

[0141] Below that, there is one cell that intersects with the row of time Ta for the table side element AYr as an uncommon element, but no aggregate data is placed there and it is left blank because both the table top element BXv and the table side element AYr are uncommon elements and are not elements of the same time.

[0142] Further below that, there is one cell that intersects with the row of time Tb for the table side element BYs as an uncommon element, and the aggregated data B(s,v) is placed there. Although both the table top element BXv and the table side element BYs are uncommon elements, they are both elements at the same time Tb, so aggregated data for time Tb exists.

[0143] On the other hand, in the horizontal integration type integration process shown in Fig. 11, corresponding aggregated data (aggregated data for two points in time to be compared) at different points in time (points Ta and Tb) are arranged horizontally, so that the integrated summary table M created has a column for point in time Ta and a column for point in time Tb arranged horizontally. Note that either the column for point in time Ta or the column for the subsequent point in time Tb can be placed in the left row. Therefore, the arrangement of each aggregated data in the integrated summary table M in Fig. 11 will be viewed horizontally.

[0144] First, looking horizontally from the left at the columns of the table side element BYk (AYh) as a common element, the columns of the table top element BXm (AXn) as a common element have columns of time Ta and time Tb, and there are two cells that intersect with these two columns. And since both the table side element BYk (AYh) and the table top element BXm (AXn) are common elements, the aggregated data A(h,n) and B(k,m) are placed in the two cells aligned horizontally.

[0145] To the right of that, a column for time Ta is provided as a column for table head element AXu, which is a non-common element, but because table head element AXu is a non-common element, a column for time Tb is not provided, and so there is one cell that intersects with the column for time Ta, and the aggregated data A(h, u) is placed there.

[0146] Further to the right, a column for time Tb is provided as a column for table head element BXv, which is a non-common element, but because table head element BXv is a non-common element, a column for time Ta is not provided, and so there is one cell that intersects with the column for time Tb, where the aggregated data B(k,v) is placed.

[0147] Next, looking horizontally from left to right at the row of the table element AYr as a non-common element, there are two cells that intersect with the two columns of times Ta and Tb for the table top element BXm (AXn) as a common element, and the aggregated data A(r,n) for time Ta is placed there. However, since the table element AYr is a non-common element, there is no aggregated data for time Tb, so these are left blank.

[0148] To the right of that is a cell that intersects with the column for time Ta for the table top element AXu as an uncommon element, and the aggregated data A(r,u) is placed there. Although the table top element AXu and the table side element AYr are both uncommon elements, they are both elements at the same time Ta, so aggregated data for time Ta exists.

[0149] Further to the right, there is one cell that intersects with the column of time Tb for table top element BXv as an uncommon element, but no aggregate data is placed there and it is left blank because both table side element AYr and table top element BXv are uncommon elements and are not elements of the same time.

[0150] Next, looking horizontally from left to right at the row of the table element BYs as a non-common element, there are two cells that intersect with the two columns of times Ta and Tb for the table top element BXm (AXn) as a common element, and the aggregated data B(s,m) for time Tb is placed there. However, since the table element BYs is a non-common element, there is no aggregated data for time Ta, so these are left blank.

[0151] To the right of that, there is one cell that intersects with the column of time Ta for the table top element AXu as an uncommon element, but no aggregate data is placed there and it is left blank because both the table side element BYs and the table top element AXu are uncommon elements and are not elements of the same time.

[0152] Further to the right, there is one cell that intersects with the column of time Tb for table top element BXv as an uncommon element, and the aggregated data B(s,v) is placed there. Although both table top element BXv and table side element BYs are uncommon elements, they are both elements at the same time Tb, so aggregated data for time Tb exists.

[0153] The integration means 26 also executes a process of creating an integrated tally table file, which is a spreadsheet software file (e.g., Microsoft Excel (registered trademark), etc.) using the data of each integrated tally table stored in the integrated tally table storage means 45, and storing the file in the integrated tally table file storage means 46.

[0154] Furthermore, the integration means 26 may determine whether or not both upper and lower character strings that are in an inclusive relationship with each other and are stored in the inclusion relationship storage means 48 are present in the table head elements AX, BX or table side elements AY, BY of the summary tables A, B to be integrated, and if both are present, perform a process of arranging these upper and lower character strings adjacent to each other horizontally as table head elements of the integrated summary table M, or adjacent to each other vertically as table side elements.

[0155] For example, as shown in FIG. 8, in the case where "Faculty of Science and Engineering" is registered as a higher-level character string and "Faculty of Science" and "Faculty of Engineering" are registered as lower-level character strings in the inclusion relationship storage means 48, when "Faculty of Science" and "Faculty of Engineering" are among the table head elements at time Ta and "Faculty of Science and Engineering" is among the table head elements at time Tb, "Faculty of Science" and "Faculty of Engineering" and "Faculty of Science and Engineering" are arranged adjacent to each other in the horizontal direction as table head elements of the integrated schedule M. Also, when "Faculty of Science" and "Faculty of Engineering" are among the table side elements at time Ta and "Faculty of Science and Engineering" is among the table side elements at time Tb, "Faculty of Science" and "Faculty of Engineering" and "Faculty of Science and Engineering" are arranged adjacent to each other in the vertical direction as table side elements of the integrated schedule M. It is considered that "Faculty of Science" and "Faculty of Engineering" have been integrated to become "Faculty of Science and Engineering" over time, so the idea is to arrange them close to each other so that they are easy to see. It should be noted that the relationship between time Ta and time Tb in this example may be reversed, in which case it would be considered that over time the "Faculty of Science and Engineering" would split into the "Faculty of Science" and the "Faculty of Engineering."

[0156] (Configuration of division means 27: Figs. 5, 6, 7, 15) When a summary table group consisting of multiple summary tables for each different point in time (points Ta and Tb) is written in a summary table file and table of contents information for each of these summary table groups is written, the division means 27 uses the table of contents information to determine whether the summary table group contains a combined summary table (a summary table that can be divided), and if so, executes a process of dividing the combined summary table into unit summary tables (summary tables that cannot be divided any further).

[0157] The division process by the division means 27 is executed when a setting is made that division is necessary according to the division necessity information stored in the setting information storage means 47 as the setting information. When division is set to be necessary, the combined tally table included in the tally table group is divided by the division means 27 into a unit tally table state, and in that state, the integration target detection means 23 detects integration targets, or the inter-set similarity calculation means 25 judges whether integration is possible. On the other hand, when division is set to be unnecessary, the combined tally table included in the tally table group is not divided, and in that state, the integration target detection means 23 detects integration targets, or the inter-set similarity calculation means 25 judges whether integration is possible. Note that even if the combined tally table is not divided, if a combined tally table with a similar configuration is included in the other tally table group, those combined tally tables become integration targets detected by the integration target detection means 23, or are judged to be able to be integrated by the inter-set similarity calculation means 25.

[0158] In this embodiment, as shown in Fig. 5, a summary table group consisting of a plurality of summary tables at time Ta is written in a summary table file at time Ta, and a summary table group consisting of a plurality of summary tables at time Tb is also written in a summary table file at time Tb. Also, as shown in Fig. 7, an INDEX sheet is provided in each of the summary table files at time Ta and Tb, and table of contents information of the summary table groups at time Ta and Tb is written in these INDEX sheets.

[0159] 7, the table of contents information at times Ta and Tb includes, for each spreadsheet constituting the respective spreadsheet groups at times Ta and Tb, a table head title constituting the table head, a table side title constituting the table side, and spreadsheet identification information (a spreadsheet ID such as [N0001]) associated with each other. In this embodiment, the table of contents information on the INDEX sheet and each spreadsheet constituting the spreadsheet group are linked, and when a spreadsheet ID portion such as [N0001] is clicked, the spreadsheet associated with that spreadsheet ID (the corresponding spreadsheet among the spreadsheet groups written on a sheet other than the INDEX sheet) is displayed on the screen.

[0160] As shown in Figure 6, each of the summary tables that make up the summary table group contains a combined summary table that can be split. In the example of Figure 6, the combined summary table with summary table ID = [N0001] has a table title "Q1 First, tell us about yourself. Where is your high school located?", which corresponds to one survey item in the raw data (see Figure 16). On the other hand, the table titles are "Q4 Department you passed" and "Q2 Tell us about your own career choice. What is your deviation score?", which correspond to two survey items in the raw data (see Figure 16). Therefore, the combined summary table shown in Figure 6 has two table titles, so it can be split into two unit summary tables.

[0161] The bottom of Figure 6 shows the table of contents information corresponding to this combined summary table. In the table of contents information, row No. 1 has the table top title Survey Item Q1, the table side title Survey Item Q4, and summary table ID = [N0001]. In addition, row No. 2 has the table top title Survey Item Q1, the table side title Survey Item Q2, and summary table ID = [N0001]. The links are set so that the same combined summary table is displayed on the screen when the summary table ID = [N0001] part of either row is clicked. Therefore, rows No. 1 and No. 2 correspond to the contents of the unit summary tables after the combined summary table is divided.

[0162] 7, the table of contents information of the INDEX sheet of the spreadsheet file at time Ta contains three rows, No. 1, No. 2, and No. 3, with the same spreadsheet ID [N0001], and these three rows have different table side titles for the same table top title, so the spreadsheet with spreadsheet ID = [N0001] is a combined spreadsheet that can be divided into three unit spreadsheets. Also, the table of contents information of the INDEX sheet of the spreadsheet file at time Tb contains two rows, No. 1 and No. 2, with the same spreadsheet ID [N0001], and these two rows have different table side titles for the same table top title, so the spreadsheet with spreadsheet ID = [N0001] is a combined spreadsheet that can be divided into two unit spreadsheets.

[0163] 15, the flow of processing by the division means 27 is shown in a flowchart as follows: First, the division means 27 acquires division necessity information stored in the setting information storage means 47, and judges whether or not there is a division request from the user (step S401).

[0164] Here, when there is a request for division, the division means 27 determines whether the group of tally sheets at times Ta and Tb includes a combined tally sheet by determining whether multiple table side titles are associated with the same table top title or multiple table top titles are associated with the same table side title in the table of contents information of each tally sheet file at times Ta and Tb stored in the tally sheet file storage means 41 while being associated with the same tally sheet identification information (tally sheet ID) (step S402).

[0165] If it is determined that a combined tally table is included (step S403), the combined tally table portion of the table of contents information is divided into unit tally tables that correspond to one table top title and one table side title within the table of contents information (step S404). Dividing within the table of contents information means dividing the table of contents information, not dividing the combined tally table data itself.

[0166] Then, the division means 27 creates a concatenated title string by concatenating the character strings of the table top title and table side title of the unit tally table obtained by dividing the combined tally table included in the tally table group at one of the different time points (for example, time Ta) in the table of contents information (step S405). In the example of Fig. 7, a concatenated title string is created by concatenating the character string "Location of High School" in the table top title and the character string "Passed Department" in the table side title in row No. 1 in the table of contents information at time Ta. Concatenated title strings are also created for rows No. 2 and No. 3.

[0167] The dividing means 27 also creates a concatenated title string by concatenating the character strings of the top title and the side title of the unit tally table obtained by dividing the combined tally table included in the tally table group at the other time point (time point Tb) in the table of contents information (step S405). In the example of FIG. 7, a concatenated title string is created by concatenating the character string "Location of High School" in the top title of the row No. 1 in the table of contents information at time point Tb and the character string "Passed Department" in the side title. A concatenated title string is also created for the row No. 2. A concatenated title string is also created by concatenating the character strings of the top title and the side title of the tally table that is in the state of a unit tally table from the beginning (step S405). In other words, if a unit tally table is included in the tally table group at the other time point (time point Tb), a concatenated title string for that unit tally table is also created.

[0168] Next, the division means 27 uses the created concatenated title string at time Ta and the concatenated title string at time Tb to brute-force calculate an in-table-of-contents title similarity indicating the similarity of unit summary tables at the title level in the table of contents information, and performs a process of determining whether or not there is a unit summary table with a high in-table-of-contents title similarity among the summary tables at the other time (time Tb) by comparing the calculated in-table-of-contents title similarity with a preset threshold value (stored as setting information in the setting information storage means 47) (step S405).

[0169] Here, the algorithm for calculating the title similarity in the table of contents by the division means 27 can be the same as the algorithm for calculating the similarity by the integration target detection means 23, but a different algorithm may also be used. Therefore, for example, a method can be used in which sentence vectors are obtained by Doc2Vec using each concatenated title character string at time points Ta and Tb, and the cosine similarity or the like of these sentence vectors is used as the title similarity in the table of contents. Also, a method can be used in which word importance vectors such as TF-IDF vectors are obtained using each concatenated title character string at time points Ta and Tb, and the cosine similarity or the like of these word importance vectors is used as the title similarity in the table of contents.

[0170] Then, when there is a unit tally table with a high title similarity in the table of contents (when the similarity is equal to or exceeds a threshold value), the division means 27 executes a process of dividing the data of the combined tally table included in the group of tally tables at one point in time (time Ta) into data of the unit tally tables in accordance with the division of the combined tally table into unit tally tables in the table of contents information (step S406).Then, the above process is also performed with time Ta and time Tb interchanged, and the data of the combined tally table at both points in time is divided if it makes sense to divide it (i.e., when a unit tally table corresponding to the unit tally table obtained by division exists from the beginning in the group of tally tables on the other side, or when it can be prepared on the other side by division).

[0171] (Configuration of setting means 28) The setting means 28 accepts operational inputs such as selection input, key input, correction operation, etc. by a user (including a worker, a work manager, and a system manager) who performs integration work by operating the input means 60, and executes a process of storing the accepted input data in the setting information storage means 47.

[0172] The setting means 28 also receives input from the system provider (including the developer, sales staff, and maintenance staff of the system) who operates the input means 60 to set initial values ​​(user-selectable values ​​and default states of options) and to set subsequent corrected values, and also executes a process of storing the received input data in the setting information storage means 47.

[0173] (Configuration of inclusion relationship registration means 29) The inclusion relationship registration means 29 receives input from the user via the input means 60 regarding upper and lower character strings that are in an inclusion relationship with each other and that can become table head elements or table side elements, and stores the received upper and lower character strings in association with each other.

[0174] (Configuration of summary table file storage means 41) The tally table file storage means 41 stores a plurality of tally table files. This tally table file storage means 41 is configured with a non-volatile memory such as an HDD or SSD. In this embodiment, it is a storage area formed by a folder in the computer, and stores tally table files at different points in time (for example, points in time Ta and Tb). It may also be a storage area formed by an external recording medium such as a USB memory or a DVD-RAM.

[0175] (Configuration of integration target advance designation information storage means 42) 5, the integration target advance designation information storage means 42 stores information on advance designation of integration targets by a user received by the integration target advance designation means 21. The integration target advance designation information storage means 42 is configured with a non-volatile memory such as an HDD or SSD. The advance designation information of integration targets has already been described in detail in the description of the integration target advance designation means 21, and therefore will not be described in detail here.

[0176] (Configuration of extraction result storage means 43) The extraction result storage means 43 stores the position information (row number, column number) of the table head, table side, and summary data portion constituting the summary table extracted by the extraction means 22, as well as the position information of each portion modified by the user's modification operation. This extraction result storage means 43 may be a main memory, but may also be a non-volatile memory.

[0177] (Configuration of Inter-element Similarity Calculation Result Storage Means 44) The inter-element similarity calculation result storage means 44 stores the determination results by the table top element similarity calculation means 24A and the table side element similarity calculation means 24B of the inter-element similarity calculation means 24. This inter-element similarity calculation result storage means 44 may be a main memory, but may also be a non-volatile memory.

[0178] (Configuration of integrated summary table storage means 45) The integrated tally table storage means 45 stores data of each integrated tally table created in the integration process by the integration means 26 in association with tally table identification information (tally table ID) that is automatically or manually assigned to each integrated tally table. This integrated tally table storage means 45 may be a main memory, but may also be a non-volatile memory.

[0179] (Configuration of integrated summary table file storage means 46) The integrated tally table file storage means 46 stores an integrated tally table file created by consolidating the data of each integrated tally table stored in the integrated tally table storage means 45. This integrated tally table file storage means 46 may be a common storage area (for example, a common folder) with the tally table file storage means 41. The file identification information (file name) of the integrated tally table file may be assigned automatically or manually.

[0180] (Configuration of setting information storage means 47) The setting information storage means 47 stores setting information obtained by operation input such as selection input, key input, and correction operation by a user (including a worker, a worker manager, and a system worker) performing an integration operation, and setting information prepared in advance by a system provider (including a developer, a sales person, and a maintenance person of this system). This setting information storage means 47 is configured with a non-volatile memory such as an HDD or an SSD. The setting information includes setting information by the setting means 28 and information after correction by the extraction means 22. Setting information determined for each type of tally table file is stored in association with type identification information of the tally table file (for example, company name such as α company, β company, developer name, and version information), setting information determined for each tally table file is stored in association with file identification information (for example, file name), and setting information determined for each tally table is stored in association with tally table identification information (tally table ID) or in association with file identification information and tally table identification information.

[0181] Specifically, the setting information includes, for example, selection information for the type of summary table file to be used, read data that is sent externally without being directly written into the program that executes the data extraction algorithm prepared for each type of summary table file, and corrected information on the position information (row number, column number) of the table head, table side, and summary data section that make up the summary table. However, since this setting information has already been described in detail in the explanation of the extraction means 22, a detailed explanation will not be given here.

[0182] The setting information also includes information on whether automatic detection processing is required by the merging target detection means 23, selection information on the merging target detection method (character string to be used, algorithm to be used) when merging target detection is selected, thresholds for the head and side table similarities when merging target detection is selected, or thresholds for the head and side table similarities, and selection information on whether to determine whether merging is possible based on the inter-set similarity when merging target detection is selected.

[0183] Furthermore, the setting information includes selection information for the inter-element similarity calculation method (algorithm to be used) by the inter-element similarity calculation means 24, a threshold value for the table head element similarity used by the table head element similarity calculation means 24A, and a threshold value for the table side element similarity used by the table side element similarity calculation means 24B.

[0184] The setting information also includes information on whether or not a merger is necessary to be determined by the inter-set similarity calculation means 25, selection information for the inter-set similarity calculation method (algorithm to be used) by the inter-set similarity calculation means 25, a threshold value for the similarity between table head element sets used by the table head element set similarity calculation means 25A, and a threshold value for the similarity between table side element sets used by the table side element set similarity calculation means 25B.

[0185] The setting information also includes selection information for the reference point in time (default = point in time Tb) when the integration means 26 creates an integrated summary table (whether to use a new or old point in time, the date, etc.), and selection information for whether the integration process by the integration means 26 will be a vertical integration type (see Figure 10) or a horizontal integration type (see Figure 11).

[0186] Furthermore, the setting information includes information on whether or not the combined summary table should be split by the splitting means 27 (settings can be made by summary table file type, by summary table file, or by summary table), selection information for the method of calculating the title similarity in the table of contents (algorithm to be used) when it is selected that splitting is necessary, and a threshold value for the title similarity in the table of contents.

[0187] (Configuration of inclusion relationship storage means 48) The inclusion relationship storage means 48 stores character strings that can be table top elements or table side elements and that are in an inclusion relationship between higher-level and lower-level character strings in association with each other. The inclusion relationship storage means 48 is composed of a non-volatile memory such as an HDD or SSD.

[0188] (Overall flow of integration process) In this embodiment, the tally sheets at different times are integrated by the tally sheet integrating system 10 in the following manner.

[0189] 13, first, a user (including a worker, a worker manager, and a system worker) performing an integration task operates the input means 60 to input necessary setting information (various selection information, thresholds, etc.), and the input is accepted by the setting means 28 and stored in the setting information storage means 47 (step S1). Note that prior to this, the system provider (including the developer, sales person, and maintenance person of this system) sets initial values.

[0190] Next, the integration target advance designation means 21 displays on the display means 70 the contents of the tally table file stored in the tally table file storage means 41, accepts advance designation of integration targets by the user (see FIG. 5), and stores the accepted advance designation information (designation information of the tally tables of multiple sets (pairs) to be integrated) in the integration target advance designation information storage means 42 (step S2). Note that the user can omit this advance designation work when automatic detection of integration targets is performed by the integration target detection means 23, or when real-time designation is performed pair by pair.

[0191] Then, processing is executed by the extraction means 22 (step S3). The extraction means 22 reads the tally table file stored in the tally table file storage means 41, and executes extraction processing of the table head, table side, and summary data part of each tally table in the tally table file, using the setting information stored in the setting information storage means 47. This extraction result is displayed on the screen of the display means 70, and is stored in the extraction result storage means 43 in association with the file identification information (file name) of the tally table file and the tally table ID of each tally table. In addition, correction input (correction of misalignment of the table head, table side, etc.) by the user who viewed the screen display is accepted, and the corrected information is overwritten and saved in the extraction result storage means 43, and is also stored in the setting information storage means 47 to be useful for subsequent maintenance work.

[0192] Furthermore, the division means 27 executes processing (step S4; see steps S401 to S406 in FIG. 15, which have already been described in detail). The division means 27 reads the information on whether or not splitting of the tally table is necessary, which is stored in the setting information storage means 47, and if there is a request for splitting, it automatically detects a combined tally table. If a combined tally table is found, it executes splitting processing of the combined tally table while judging whether a paired unit tally table exists in the other party's tally table group from the beginning, or whether it can be prepared on the other party's side by splitting the other party's combined tally table. The data of the unit tally table obtained by splitting the combined tally table (data of the table head, table side, and summary data part) is stored in the extraction result storage means 43 in association with the file identification information (file name) of the tally table file and the automatically or manually assigned tally table ID of the unit tally table.

[0193] Next, if information on advance designation of integration targets is stored in the integration target advance designation information storage means 42, it is read (step S5).

[0194] Then, it is determined whether or not pre-designated information for integration targets has been read (step S6). If so, data for the top and side of the table for each of the extracted summary tables for the two points in time are obtained, pair by pair, from the extraction result storage means 43 in accordance with the pre-designated information for integration targets (step S7).

[0195] Furthermore, if the pre-designated information for the integration targets has not been read in at step S6 above, a decision is made as to whether or not to designate them manually (step S8). If manual designation is selected, a user input is accepted for real-time designation of a pair of summary tables to be integrated, and data for the top and side of the tables of each of the summary tables at the two designated points in time is obtained from the extraction result storage means 43 (step S9).

[0196] On the other hand, if no manual designation is made in step S8, the integration target detection means 23 acquires each data of the top and side parts of each of the extracted summary tables, calculates the similarity of the summary tables (top-side part similarity, or top-side part similarity and side-side part similarity), and automatically determines a pair of summary tables to be integrated (step S10, see FIG. 4).

[0197] 14, thereafter, processing is executed by the inter-element similarity calculation means 24 (step S11). For a pair of summary tables to be processed, the inter-element similarity calculation means 24 calculates the similarity between table top elements and determines the corresponding table top element, calculates the similarity between table side elements and determines the corresponding table side element, and stores these determination results in the inter-element similarity calculation result storage means 44 (see FIG. 12).

[0198] Next, it is determined whether the pair of summary tables to be integrated is specified by the user (step S12), and if it is specified by the user (i.e., if the user specifies it in advance or in real time), the integration target detection means 23 has not automatically detected the integration target, so processing is executed by the inter-set similarity calculation means 25 (step S13). The inter-set similarity calculation means 25 calculates the similarities between the table head element sets and the table side element sets using the calculation results of the similarities between the table head elements and the table side elements stored in the inter-element similarity calculation result storage means 44, and determines whether integration is possible (see FIG. 12).

[0199] Then, it is determined whether or not integration is possible by the inter-set similarity calculation means 25 (step S14), and if integration is possible, integration processing is performed by the integration means 26 (see Figs. 9, 10, and 11), data for an integrated tally table is created, and stored in the integrated tally table storage means 45 (step S15). On the other hand, if integration is not possible, processing by the integration means 26 is not performed.

[0200] Furthermore, if the user has not specified the target for integration in step S12, this means that automatic detection processing of targets for integration has been performed by the integration target detection means 23, so a determination is made as to whether or not integration can be performed based on the similarity between sets using the setting information stored in the setting information storage means 47 (step S16), and if integration can be performed, the process proceeds to the processing by the inter-set similarity calculation means 25 in the above-mentioned step S13. On the other hand, if integration can be not performed, the process proceeds to the processing by the integration means 26 in the above-mentioned step S15 without performing processing by the inter-set similarity calculation means 25.

[0201] Thereafter, it is determined whether any integration targets remain (step S17), and if so, the process returns to step S6 in FIG. 13 described above; if not, the integration means 26 creates an integrated tally table file using the data of each integrated tally table stored in the integrated tally table storage means 45, and stores this in the integrated tally table file storage means 46 (step S18), thereby completing the series of processes.

[0202] (Effects of this embodiment) According to this embodiment, the following effects are obtained: In other words, the tally sheet integration system 10 uses the inter-element similarity calculation means 24 to calculate the similarity between table head elements and table side elements, determines table head elements and table side elements whose semantics correspond, then uses the inter-set similarity calculation means 25 to calculate the similarity between table head element sets and between table side element sets, judges whether tally sheets at different points in time (i.e., multiple points in time) can be integrated, and if it is judged that integration is possible, the integration means 26 can create an integrated tally sheet.

[0203] Therefore, the person performing the integration work does not need to judge the identity of the table top elements and table side elements, or judge whether the summary tables can be integrated, so that the integrated summary table can be easily created without time or effort. Also, if an appropriate threshold is set in advance, the integration process can be performed without human judgment, so that any worker can create a uniform integrated summary table regardless of their knowledge or experience.

[0204] Furthermore, since the tally sheet integrating system 10 is equipped with an integration target detection means 23, it can automatically detect tally sheets at different time points to be integrated from among multiple tally sheets (groups of tally sheets) that exist for each of the different time points. This eliminates the need for the user to specify the integration targets in advance or in real time, and allows the integration work to be performed easily without time or effort.

[0205] Furthermore, since automatic detection processing of integration targets by the integration target detection means 23 can be substituted for the processing of determining whether integration is possible by the inter-set similarity calculation means 25, the user can set to omit processing by the inter-set similarity calculation means 25. Note that a setting can also be made to improve the accuracy of determining whether integration is possible by performing processing by both the integration target detection means 23 and the inter-set similarity calculation means 25.

[0206] Furthermore, because the tally table integration system 10 is equipped with a splitting means 27, even when a combined tally table is included in the group of tally tables at each point in time and these combined tally tables cannot be integrated, or when a combined tally table is included in the group of tally tables at one point in time but not at the other point in time, the combined tally table can be split into unit tally tables to integrate the tally tables.

[0207] Furthermore, since the summary table integration system 10 is equipped with an inclusion relationship registration means 29 and an inclusion relationship storage means 48, when character strings in an inclusion relationship exist in the top or side elements of summary tables at different times, such as when the Faculty of Science and the Faculty of Engineering are top or side elements at a certain time and the top or side element at a later time is the Faculty of Science and Engineering, or vice versa, the integration means 26 can arrange the summary data for those elements in an easy-to-view manner in the integrated summary table.

[0208] <Transformation Form>

[0209] The present invention is not limited to the above-described embodiment, and modifications within the scope of the present invention are included in the present invention.

[0210] For example, the spreadsheet integration system 10 of the embodiment has been described as a standalone configuration as shown in Fig. 1, but it may be a server-client system. In this case, the functions of the main body 10A may be provided in the server, and the functions of the input means 60 and the display means 70 may be provided in the client terminal, or the functions of the main body 10A may be distributed between the server and the client terminal.

[0211] In the above embodiment, if the inter-set similarity calculation means 25 determines that the schedules A and B at different times (times Ta and Tb) cannot be integrated, the integrated schedule M cannot be created. In this case, the inter-element similarity calculation means 24 may be configured to calculate the inter-table top element similarity and the inter-table side element similarity for a state in which the top and side parts of one of the schedules A and B at different times (times Ta and Tb) are swapped vertically and horizontally, and to determine the top and side elements whose meanings correspond between the schedules at different times by comparing the calculated inter-table top element similarity and the inter-table side element similarity with respective preset thresholds. In this case, when the inter-set similarity calculation means 25 determines whether integration is possible again after the top and side parts are swapped vertically and horizontally and the processing of the inter-element similarity calculation means 24 is performed, there are cases in which it is determined that integration is possible. If it is determined that the data can be integrated, the integration means 26 naturally performs the integration process with the top and side portions of the front page swapped vertically and horizontally. [Industrial Applicability]

[0212] As described above, the spreadsheet integration system and program of the present invention are suitable for use in cases where, for example, when there is spreadsheet data at different points in time that has been automatically created by a spreadsheet creation system from raw data at different points in time, the spreadsheet data is arranged and integrated so that it can be compared. [Explanation of symbols]

[0213] 10. Integrated spreadsheet system 23 Integrated target detection means 24 Means for calculating similarity between elements 25 Means for calculating similarity between sets 26 Integration Measures 27 Division means 48 Inclusion relation storage means

Claims

1. A tally sheet integration system configured by a computer that executes a process of integrating tally sheets at different times, an inter-element similarity calculation means for calculating a table top element similarity indicating a similarity between character strings of a plurality of table top elements constituting the table tops of the summary tables at different times to be integrated and written in the summary table file, and determining table top elements having a corresponding meaning between the summary tables at different times by comparing the calculated table top element similarity with a preset threshold value, and for calculating a table side element similarity indicating a similarity between character strings of a plurality of table side elements constituting the table side parts of the summary tables at different times to be integrated, and for comparing the calculated table side element similarity with a preset threshold value to determine table side elements having a corresponding meaning between the summary tables at different times; an inter-set similarity calculation means for determining whether or not the summary tables at different times can be integrated based on whether or not both the table head element set similarity and the table side element set similarity are high; determining whether or not the summary tables at different times can be integrated based on whether or not both the table head element set similarity and the table side element set similarity are high; and an integration means for executing an integration process for creating an integrated summary table by vertically or horizontally arranging the summary data for the table head element and the table side element adjacent to each other when the semantic contents of the summary tables at different times correspond to each other for both the table head element and the table side element among the summary data of the summary tables at different times that are determined to be integrable by the inter-set similarity calculation means; A summary table integration system comprising:

2. a merging target detection means for detecting a tally table at a different time point to be merged from among a plurality of tally tables existing for each of the different time points written in the tally table file; The integrated object detection means includes: (1) For each of the summary tables at different times, a table head / side portion concatenated string is created by concatenating a table head element concatenated string obtained by concatenating each character string of a plurality of table head elements constituting the table head and / or a table head string including a character string of a table head title, and a table side element concatenated string obtained by concatenating each character string of a plurality of table side elements constituting the table side and / or a table side portion string including a character string of a table side title, and using the created table head / side portion concatenated strings of the summary tables at different times, a table head / side portion similarity indicating a similarity between the table head and the table side portion of the summary tables at different times is calculated by an algorithm different from the similarity calculation algorithm used in the inter-set similarity calculation means, and the calculated table head / side portion similarity is compared with a preset threshold value, and the summary tables at different times to be integrated are detected depending on whether the table head / side portion similarity is high or not, or (2) A table top similarity indicating the similarity of the table tops of the summary tables at different times is calculated by an algorithm different from the algorithm for calculating similarity used in the set similarity calculation means, using a table top element concatenated string obtained by concatenating the strings of each of a plurality of table top elements constituting the table tops of the summary tables at different times and / or a table top string including a string of a table top title, and the calculated table top similarity is compared with a preset threshold value, and a table side portion similarity indicating the similarity of the table side portions of the summary tables at different times is calculated by the different algorithm, using a table side element concatenated string obtained by concatenating the strings of each of a plurality of table side elements constituting the table side portions of the summary tables at different times and / or a table side portion string including a string of a table side title, and the calculated table side portion similarity is compared with a preset threshold value, and the summary tables at different times to be integrated are detected depending on whether or not both the table top similarity and the table side portion similarity are high. The element similarity calculation means The processing is executed by using the summary tables at different time points detected by the integration target detection means as integration targets.

2. The spreadsheet integrating system according to claim 1.

3. A tally sheet integration system configured by a computer that executes a process of integrating tally sheets at different times, an integration target detection means for detecting a tally table at a different time point to be integrated from among a plurality of tally tables existing for each of the different time points written in the tally table file; an inter-element similarity calculation means for calculating a table top element similarity indicating a similarity between character strings of a plurality of table top elements constituting the table tops of the summary tables at different times to be integrated detected by the integration target detection means, and determining table top elements having corresponding semantic contents between the summary tables at different times by comparing the calculated table top element similarity with a preset threshold value, and for calculating a table side element similarity indicating a similarity between character strings of a plurality of table side elements constituting the table side parts of the summary tables at different times to be integrated, and for comparing the calculated table side element similarity with a preset threshold value to determine table side elements having corresponding semantic contents between the summary tables at different times; an integration means for executing an integration process for creating an integrated summary table by vertically or horizontally arranging the summary data of a table head element and a table side element adjacent to each other when the semantic contents of the summary tables at different times of the integration target detected by the integration target detection means correspond to each other for both the table head element and the table side element, The integration target detection means (1) For each of the summary tables at different times, a table head / side portion concatenated string is created by concatenating a table head element concatenated string obtained by concatenating the character strings of each of the multiple table head elements constituting the table head, or a table head string obtained by adding a character string of a table head title to the table head element concatenated string, and a table side portion concatenated string obtained by concatenating the character strings of each of the multiple table side elements constituting the table side portion, or a table side portion string obtained by adding a character string of a table side title to the table side element concatenated string, and using the created table head / side portion concatenated strings of the summary tables at different times, a table head / side portion similarity indicating the similarity between the table heads and the table side portions of the summary tables at the different times is calculated, the calculated table head / side portion similarity is compared with a preset threshold, and the summary tables at the different times to be integrated are detected depending on whether the table head / side portion similarity is high or not; or (2) A summary table integrating system configured to calculate a top table similarity indicating a similarity of the top tables of the summary tables at different times using a top table element concatenated string obtained by concatenating character strings of a plurality of top table elements constituting the top table of each of the summary tables at different times or a top table string obtained by adding a character string of a top table title to the top table element concatenated string, and to compare the calculated top table similarity with a preset threshold value, and to calculate a side table portion similarity indicating a similarity of the side table portions of the summary tables at different times using a top table element concatenated string obtained by concatenating character strings of a plurality of top table elements constituting the top table of each of the summary tables at different times or a side table portion string obtained by adding a character string of a top table title to the top table element concatenated string, and to compare the calculated side table portion similarity with a preset threshold value, and to detect the summary tables at different times to be integrated depending on whether the top table similarity and the side table portion similarity are both high or not.

4. A group of tally tables consisting of a plurality of tally tables for different time points is written in the tally table file, and as table of contents information for each of the tally table groups, a table head title constituting the table head, a table side title constituting the table side portion, and tally table identification information are written in association with each other for each of the tally tables constituting the tally table group, a division means for dividing, when the group of summary tables includes a combined summary table as a divisible summary table, the combined summary table into unit summary tables as divisible summary tables by using the table of contents information, This division means: a determination is made as to whether the combined tally table is included in the tally table group by determining whether a plurality of table side titles are associated with the same table top title or a plurality of table top titles are associated with the same table side title in the table of contents information in a state where the combined tally table is associated with the same tally table identification information, and if it is determined that the combined tally table is included, the combined tally table portion of the table of contents information is divided into the unit tally tables associated with one table top title and one table side title in the table of contents information, A table of contents title similarity indicating a similarity of the unit summary table at a title level in the table of contents information is calculated using a concatenated title string obtained by concatenating the character strings of the top title and the side title of the unit summary table obtained by dividing the combined summary table included in the summary table group at one of the different time points in the table of contents information, and a concatenated title string obtained by concatenating the character strings of the top title and the side title of the unit summary table obtained by dividing the combined summary table included in the summary table group at the other time point in the table of contents information or a summary table that is in the state of the unit summary table from the beginning, and the calculated table of contents title similarity is compared with a preset threshold value to determine whether or not the unit summary table at the other time point has a high table of contents title similarity, and if there is a unit summary table at the other time point, the data of the combined summary table included in the summary table group at one time point is divided into data of the unit summary table in accordance with the division of the combined summary table into the unit summary table in the table of contents information.

4. The summary table integration system according to claim 1,

5. The element similarity calculation means When it is determined by the inter-set similarity calculation means that the summary tables at different times cannot be integrated, The method is configured to calculate the similarity between table top elements and the similarity between table side elements even when the table top and the table side are vertically and horizontally swapped for one of the summary tables at different times, and to determine table top elements and table side elements whose semantic contents correspond to each other between the summary tables at different times by comparing the calculated similarity between table top elements and the similarity between table side elements with respective preset thresholds.

3. The spreadsheet integration system according to claim 1 or 2.

6. a storage unit for storing inclusion relationships that stores character strings that can be table top elements or table side elements and that are inclusive of each other and correspond to each other; The integration means includes: In the case where both upper and lower character strings in a mutually inclusive relationship stored in the inclusion relationship storage means exist in the table head element or table side element of the integrated summary table, the upper and lower character strings are arranged adjacent to each other in the horizontal direction as the table head element of the integrated summary table, or arranged adjacent to each other in the vertical direction as the table side element.

4. The summary table integration system according to claim 1,

7. A program for causing a computer to function as the spreadsheet integrating system according to any one of claims 1 to 3.

Images