JP6805109B2 - Analysis program - Google Patents

Description

The present invention relates to an analytical program.

DB (database) store data of the current system is input and analyzed for each data table, and the rules (regularity) possessed by the pattern that actually occurs due to the combination of data values between columns in the data table are automatically detected. Think about it. The rules here mainly refer to business rules. For example, the business rules held by the amusement park ticket management system include "ticket prices are 2600 yen for adults and 1300 yen for children."

Conventionally, to detect business rules related to business using the current system, there have been a method in which a person reads and understands documents related to the current system and extracts them, or a method in which a person concerned conducts a hearing and asks.

Japanese Unexamined Patent Publication No. 2011-154653

However, the method of understanding a document has a problem that the correct rule cannot be extracted unless the document is constantly updated, and a problem that it cannot be automatically extracted and a large operation is required. In addition, even with the hearing method, there is a problem that rules that the hearing target person has forgotten or do not know cannot be detected, and there is also an operation problem that automatic extraction cannot be performed.

On the other hand, there is also a method of analyzing the program of the current system. In this method, it is possible to know what kind of regularity there is by tracking conditional branches, etc., but it takes a lot of operation to analyze, and the analysis result extracts only the rules described in the program. There is a problem that it cannot be done. For example, even if there is a rule that "the system is used on all days of the week except Monday" as a rule that is not described in the program, it is difficult to extract this rule by program analysis.

Further, in general, there is a method of creating a system database by creating a logical model from a conceptual data model and then creating a physical model from the logical model (see, for example, Patent Document 1). Therefore, if the reverse method of this method is followed, that is, if the physical model is used as an input, a logical model can be generated from the physical model, and a conceptual data model can be generated from the logical model. That is, it is possible to extract the specifications of the current system as a concept. The specifications of the current system that can be extracted here are what kind of meaning (entity) data table the database has, what kind of attribute (data item) has in the data table, and what is the relationship between each table? Is it?

However, even with such a method, it is difficult to extract the regularity in each data table, particularly the regularity corresponding to the business rule.

The present invention has been made in view of the above points, and an object of the present invention is to enable extraction of rules from a data group.

Therefore, in order to solve the above problem, the analysis program sorts a plurality of data having a plurality of items based on the value in the first item among the plurality of items, and in the sorting result, the first item. A computer is made to execute an extraction procedure for extracting rules regarding the data based on the arrangement state of the values of one item and the arrangement state of the values of other items among the plurality of items .

It is possible to extract rules from a data group.

It is a figure which shows the hardware configuration example of the analyzer 10 in embodiment of this invention. It is a figure which shows the functional structure example of the analyzer 10 in embodiment of this invention. It is a flowchart for demonstrating an example of the processing procedure executed by the analyzer 10. It is a figure which shows an example of the input data group. It is a figure which shows an example of the sort result of the data group based on the label of a certain item. It is a figure for demonstrating an example of the rule which can be derived in this embodiment.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a diagram showing a hardware configuration example of the analyzer 10 according to the embodiment of the present invention. The analyzer 10 of FIG. 1 includes a drive device 100, an auxiliary storage device 102, a memory device 103, a CPU 104, an interface device 105, and the like, which are connected to each other by a bus B, respectively.

The program that realizes the processing in the analyzer 10 is provided by a recording medium 101 such as a CD-ROM. When the recording medium 101 storing the program is set in the drive device 100, the program is installed in the auxiliary storage device 102 from the recording medium 101 via the drive device 100. However, the program does not necessarily have to be installed from the recording medium 101, and may be downloaded from another computer via the network. The auxiliary storage device 102 stores the installed program and also stores necessary files, data, and the like.

The memory device 103 reads and stores the program from the auxiliary storage device 102 when the program is instructed to start. The CPU 104 executes the function related to the analyzer 10 according to the program stored in the memory device 103. The interface device 105 is used as an interface for connecting to a network.

FIG. 2 is a diagram showing a functional configuration example of the analyzer 10 according to the embodiment of the present invention. In FIG. 2, the analyzer 10 includes a data input unit 11, a labeling unit 12, a sorting unit 13, a rule extraction unit 14, a rule output unit 15, and the like. Each of these parts is realized by a process of causing the CPU 104 to execute one or more programs installed in the analyzer 10.

The data input unit 11 inputs a plurality of data to be analyzed. Each data has a plurality of items (columns) in common. Therefore, the plurality of data may be formatted into table data (table data) and input. In this case, one row corresponds to one data. When a plurality of table data having a dependency relationship are input, the data input unit 11 may derive the dependency relationship based on the relationship between the primary key and the foreign key and integrate them into one table data. The value of each item may be any of a numerical value, a symbol, a character, and the like.

The label assigning unit 12 assigns a label for each item to each data for each type of value in the item. Therefore, for some data and other data, the same value of the same item is given the same label.

The sort unit 13 sorts (sorts) the data group for each item so that the same label is continuous for the item.

The rule extraction unit 14 extracts a rule (for example, a business rule) indicating regularity regarding the input data group based on the sort result (sort result) by the sort unit 13. The rule refers to, for example, a business or a business-related rule from which the input data group is generated.

The rule output unit 15 outputs the rule extracted by the rule extraction unit 14.

Hereinafter, the processing procedure executed by the analyzer 10 will be described. FIG. 3 is a flowchart for explaining an example of a processing procedure executed by the analyzer 10.

In step S101, the data input unit 11 inputs a data group (hereinafter referred to as "input data group").

FIG. 4 is a diagram showing an example of an input data group. In FIG. 4, each data (each record or each row) shows the purchase history of a specific product A (for example, a ticket for a certain amusement park a), and each item such as gender, age group, member flag, price, and address. Contains the value of. Gender is the gender of the purchaser of the goods. The age group is a concept that distinguishes buyers by the difference in the tens place of age. The member flag is information indicating whether or not the member is a member related to the product A. The price is the amount paid by the purchaser to purchase the product A. The price of the product A is not uniform for all purchasers, and changes according to the attributes of the purchasers and the like. The address is the buyer's address.

Subsequently, the labeling unit 12 counts the number of items C of the input data group (S102). In the present embodiment, since each data has the same item, the number of items of one data may be counted. When the data group is input as table data, the number of items in the table may be counted.

Subsequently, the labeling unit 12 substitutes 1 for the variable n (S103). The variable n is a variable indicating the order of the items to be processed.

Subsequently, the labeling unit 12 counts the number of types S of the value of the nth item (hereinafter, referred to as “item n”) in the input data group (S104). For example, if the nth item is an item indicating gender, the number of types S of values is basically two of "male" and "female". However, if there is data in which the nth item is blank, the number of value types S is 3. That is, blanks (states without values) are also counted as one type.

Subsequently, the labeling unit 12 determines whether or not the type number S is the same as the number of data (the number of rows in the case of table data) (S105). When the number of types S is the same as the number of data (that is, when the values of all the data are different for the item n) (YES in S105), the labeling unit 12 has the label "Labern-" common to the items n of all the data. 1 ”is given (S106). That is, the same label is given to all data.

On the other hand, when the number of types S is different from the number of data (NO in S105), the labeling unit 12 attaches a label (Labern-1 to Label-) corresponding to the type of the value of the item n of the data to the item n of each data. S) is added (S107). That is, the same label is given to the data having the same value of the item n for the item n.

The purpose of step S105 is to determine whether or not the value of item n is selective from a finite number of candidates (whether or not it is not an unspecified majority). Therefore, for example, if there is an item such as "name", it is expected that the item is determined to be YES in step S105. However, there is a possibility that the "name" has the same surname and the same name, and if the same surname and the same name are included, the determination in step S105 will be NO. Therefore, in step S105, it may be determined whether or not the ratio or ratio of the number of types S to the number of data (that is, the number of types S / the number of data) is equal to or greater than the threshold value. If the ratio or ratio is equal to or greater than the threshold value, step S106 may be executed, and if not, step S107 may be executed.

Subsequently, the labeling unit 12 determines whether or not the value of the variable n is equal to or greater than the number of items C (S108). When the value of the variable n is less than the number of items C (NO in S108), the labeling unit 12 adds 1 to the variable n (S109), and repeats step S104 and subsequent steps. That is, step S104 and subsequent steps are executed for all items. As a result, labels are given to all items of all data.

When the value of the variable n becomes the number of items C or more (YES in S108), the sort unit 13 substitutes 1 for the variable n (S110). Subsequently, the sort unit 13 sorts the input data group based on the label of the item n (S111). That is, the sort unit 13 sorts the input data group so that the same label is continuous for the item n.

FIG. 5 is a diagram showing an example of the sorting result of the data group based on the label of a certain item. In FIG. 5, the horizontal direction corresponds to the arrangement of items, and the vertical direction corresponds to the number of data (arrangement of data). That is, # 1 to # 7 are item identifiers, respectively. Further, in item # 1, the height of "Label1-1" corresponds to the number of data to which "Label1-1" is assigned to item # 1.

Here, a set of data with the same label for each item is called a "group". FIG. 5 shows the sorting result based on the label of item # 1. Therefore, for item # 1, the groups are clearly distinguished. On the other hand, since the other items are arranged according to the item # 1 (that is, the relationship between the items of each data is maintained), one group is divided into a plurality of items, for example, as in the item # 3. It is distributed. Note that FIG. 5 is not exactly consistent with FIG. 4 for convenience of explanation.

Subsequently, the rule extraction unit 14 determines that the item n has an inclusion relationship with any other item based on the arrangement state of the label of the item n and the arrangement information of the label of the other item in the sort result. It is determined whether or not the pattern is present (S112).

Here, the inclusion relation pattern refers to the relations (1) to ( 7 ) below, and is a concept for distinguishing the height of the correlation between items, for example. In the present embodiment, as an example, the following seven inclusion relation patterns 1 to 7 are defined.
(1) Pattern 1: Complete synchronization between multiple groups (that is, the number of groups is the same between one item and the other item, and the set of data belonging to each group matches).
(2) Pattern 2: Synchronize between some groups (that is, the sets of data belonging to some groups match)
(3) Pattern 3: There are multiple groups in both items, and the overall inclusion relationship (each group of one is included in one of the other groups, and each group of the other group is the one. (Including one or more groups of) is established (4) Pattern 4: A combination of one plurality of groups is synchronized with the other group (that is, a set of data belonging to one of the plurality of groups is combined with the other one group). Matches the set of data to which it belongs.)
(5) Pattern 5: a plurality of groups of one item, are included in one group of the other items (i.e., a set of data belonging to a plurality of groups of hand item, one of the other items A true subset of the group.)
(6) Pattern 6: An inclusion relationship is established for some groups (a set of data belonging to one of the groups is a true subset of a set of data belonging to any of the other groups).
(7) Pattern 7: An inclusion relationship is established because one item is a single group. Thus, in the sort result, a set of data having the same value for the item n and the same value for the other item are set. The inclusion relationship pattern (high correlation) between the item n and other items is determined based on the commonality with the set of data to be possessed or the presence or absence of the inclusion relationship. The height of the correlation is such that pattern 1> pattern 2> pattern 3> pattern 4> pattern 5> pattern 6> pattern 7.

For example, the set of data belonging to each group of item # 1 and the set of data belonging to each group of item # 2 completely match. Therefore, item # 1 has an inclusive relation pattern of pattern 1 with item # 2.

Further, the set of data belonging to the group of "Label1-1" of item # 1 and the set of data belonging to the group of "Label4-1" of item # 4 are completely matched (synchronized), but other than that. No synchronization is allowed between groups. Therefore, item # 1 has an inclusion relation pattern of pattern 2 with item # 4 .
Further, both items # 1 and item # 7 have a plurality of groups, and all of the respective groups of both are included in any of the other groups. Specifically, the data set composed of the data set belonging to the group of "Label 1-1" of item # 1 and the data set belonging to the group of "Label 1-2" of item # 7 is the data set of "Label 7-1" of item # 7. Matches a dataset that belongs to a group. Further, the data set composed of the data set belonging to the group of "Label 1-3" of item # 1 and the data set belonging to the group of "Label 1-4" belongs to the group of "Label 7-2" of item # 7. Matches the data set. Therefore, item # 1 has an inclusion relationship pattern of pattern 3 with item # 7. Regarding the presence or absence of the inclusion relationship, whether any group of item n includes a group of any item of other items, and any group of item n is any item of other items. Only one of them included in the group may be determined. This is because the other item can be determined when the other item becomes the item n.

Further, the data set belonging to the group of " Label 1-1" of item # 1 and the data set belonging to the group of "Label 1-2" are the true parts of the data set belonging to the group of "Label 8-1" of item # 8. It is a set. Therefore, item # 1 has an inclusion relationship pattern of pattern 4 with item # 8.

Further, the data set composed of the data set belonging to the group of "Label 1-1" of item # 1 and the data set belonging to the group of "Label 1-2" belongs to the group of "Label 6-1" of item # 6. Matches the data set, but no inclusion relationships are found for other groups. Therefore, item # 1 has an inclusion relationship pattern of pattern 5 with item # 6.

In addition, the group of item # 5 is one. Therefore, all groups of item # 1 are included in a single group of item # 5. Therefore, item # 1 has an inclusion relation pattern of pattern 7 with item # 5.

Note that item # 1 and item # 3 are not synchronized or included in any of the groups. Therefore, item # 1 has no inclusion relationship pattern with item # 3.

Further, regarding item # 1, there is no other item having an inclusion relationship pattern of pattern 6, but for example, item # 4 has an inclusion relationship pattern of pattern 6 with item # 6. That is, the data group belonging to the group of "Label 4-1" of item # 4 is a true subset of the data group belonging to the group of "Label 6-1" of item # 6.

If the item n does not have an inclusion relationship pattern with all the items (NO in S112), the process proceeds to step S114. On the other hand, when the item n has at least one inclusion relationship pattern with any of the other items (YES in S112), the rule extraction unit 14 sets the rule based on the inclusion relationship pattern of the item n. Extract (S113).

According to the example of FIG. 5, the following rules are extracted.

Based on the pattern 1, if the item # 1 is "Label1-1", the item # 2 is "Label2-1" (the rule in the reverse direction is also established), and the items # 1 "Label1-2" to "Label1-" Rule R1 is extracted such that the same rule is established between "4" and "Label2-2" to "Label2-4" of item # 2.

Based on the pattern 2, rule R2 such that item # 1 is "Label1-1" and item # 4 is "Label4-1" (the rule in the reverse direction is also established) is extracted.

Based on pattern 3, if item # 1 is "Label 1-1" or "Label 1-2", item # 7 is "Label 7-1" and item # 1 is "Label 1-3" or "Label 1-4". If, the rule R3 such that the item # 7 is “Label7-2” (the rule in the reverse direction is also established) is extracted. That is, the included group is a selective condition.

Based on the pattern 4, if the item # 1 is "Label 1-1" or "Label 1-2", the rule R4 such that the item # 6 is Label 6-1 (the rule in the reverse direction is also established) is extracted.

Based on the pattern 5, if the item 1 is "Label 1-1" or "Label 1-2", the rule R5 such that the item # 8 is "Label 8-1" (the rule in the reverse direction does not hold) is extracted.

Based on the pattern 7, the rule R7 such that the item # 5 is "Label 5-1" regardless of the value of the item # 1 (the rule in the reverse direction is also established) is extracted.

When item # 4 is item n, for example, if item # 4 is "Label 4-1" based on pattern 6, item # 6 is "Label 6-1" (rule in the reverse direction). R6 is extracted.

The process proceeds to step S114 following step S113. In step S114, the sort unit 13 determines whether or not the value of the variable n is equal to or greater than the number of items C. When the value of the variable n is less than the number of items C (NO in S114), the sort unit 13 adds 1 to the variable n (S115), and repeats step S111 and subsequent steps. That is, steps S111 and subsequent steps are executed for all items. At this time, in step S112, it may be determined whether or not there is an inclusion relationship pattern with the item n for the items before the item n-1. As described above, when the inclusion relationship is determined in only one direction (included or included), the inclusion relationship in the other direction can be extracted.

When the value of the variable n becomes the number of items C or more (YES in S114), the rule output unit 15 outputs the rule extracted in step S112 for each item (S116). For example, each rule may be displayed on the display device, or may be stored in the auxiliary storage device 102 or the like. When the number of output rules is enormous, for example, the output rules can be narrowed down based on the relationship of the priority of rule R1> rule R2> rule R3> rule R4> rule R5> rule R6> rule R7. The difference in the importance of the output rules may be expressed.

The user can derive the following rules by combining the output rules, for example.

FIG. 6 is a diagram for explaining an example of a rule that can be derived in the present embodiment. FIG. 6 is not completely consistent with FIG. 5 for convenience. Further, regarding item # 1, "Null" indicates a blank.

For example, paying attention to the data group d1 in FIG. 6, it is possible to derive a rule (business rule) that "the price is 200 yen for a purchaser who is a man, in his 40s, and is a member". ..

As described above, according to the present embodiment, it is possible to extract rules from the data group. The extraction of the rule is an array state of the values of each item in the sort result based on the value type of each item (that is, a set of data having the same value and a set of data having the same value for other items. Since it can be performed based on the commonality or the presence / absence of the inclusion relationship), it is possible to extract the rule with a small amount of calculation.

Further, for an item having a different value for each data (or an item in which the ratio of the type number S to the number of data is equal to or more than the threshold value), one value (label) is given to a specific value of the item (or For example, it is possible to prevent a specific address (such as a specific address) from being included in the business rule (that is, a rule that does not contribute to the business rule is included in the business rule).

Further, for example, in a system (server / client system) for performing a certain business (business), DB store data representing a trace of the business (business) performed by the system user is accumulated in the server. There are cases where the business rules (business rules) of the business that uses this system are unknown, and in that case, it hinders the renewal development of improving the current system and recreating it into a new system. In such a case, the present embodiment can be utilized to estimate the business rules related to the business using the current system.

For example, DB store data is often stored in a table data format, and in that case, the rules of each data record (row unit) in one table data are estimated. At this time, since the rule to be estimated is composed of the combination of the value types (for example, labeled data groups) held by each item (column) in the table, the combination pattern of these data groups is automatically set. The rule can be automatically estimated by detecting it. As a result, business rules (fact-based business rules) that are in line with the actual situation can be extracted. Therefore, even if a business rule is described in an existing document such as a system specification or a user manual and it is not properly maintained and managed and is not the latest business rule, it is extracted from the system DB store data. It can be used for system renewal as the business rule is the correct rule.

Further, even if there is a rule that is not described in the program constituting the system, it can be extracted from the DB store data.

Although the embodiments of the present invention have been described in detail above, the present invention is not limited to such specific embodiments, and various aspects are within the scope of the gist of the present invention described in the claims. It can be transformed and changed.

10 Analytical unit 11 Data input unit 12 Labeling unit 13 Sorting unit 14 Rule extraction unit 15 Rule output unit 100 Drive device 101 Recording medium 102 Auxiliary storage device 103 Memory device 104 CPU
105 Interface device B bus

Claims (5)

A sorting procedure for sorting a plurality of data having a plurality of items based on a value in the first item of the plurality of items, and a sorting procedure.
In the sort result, an extraction procedure for extracting a rule regarding the data based on the arrangement state of the values of the first item and the arrangement state of the values of other items among the plurality of items, and
An analysis program characterized by having a computer execute. The extraction procedure is based on the presence or absence of a commonality or inclusion relationship between a set of data having the same value for the first item and a set of data having the same value for the other items. To extract,
The analysis program according to claim 1. In the extraction procedure, when the set of data having the first value for the first item is a subset of the set of data having the second value for the second item, the first item is said. When it is the first value, the rule that the second item is the second value is extracted.
The analysis program according to claim 1 or 2. In the extraction procedure, a set of data having a third value for the first item and a set of data having a fourth value are a subset of a set of data having a fifth value for the second item. If the first item is the third value or the fourth value, the rule that the second item is the fifth value is extracted.
The analysis program according to any one of claims 1 to 3, wherein the analysis program is characterized in that. For each item, the number of types of values is specified, and for items in which the ratio of the number of types to the number of data is equal to or greater than a threshold value, the computer is made to execute an assignment procedure for assigning the same value to all the data.
For items to which the same value is assigned to all the data, the computer is made to execute the sorting procedure and the extraction procedure based on the values assigned in the assigning procedure.
The analysis program according to any one of claims 1 to 4, wherein the analysis program is characterized in that.

Images