Multidimensional model automatic merging method and system based on intelligent SQL analysis

Abstract

The embodiment of the present invention discloses a method and system for automatic merging of multidimensional models based on intelligent SQL parsing. Specifically, the following methods are included: converting the query statements into multiple logical execution plans through syntax parsing and semantic parsing in turn; performing a first classification on the multiple fact data tables corresponding to the multiple logical execution plans; performing a second classification on the association relationships of the fact data tables grouped as a group based on a greedy algorithm, and dividing all compatible logical execution plans into the same execution plan subgraph; appending a left outer join to the association relationships of the fact data tables in the execution plan subgraph, and clustering the execution plan subgraphs into the same existing model based on the left outer join; optimizing the existing cube set based on the clustered existing model. The present application solves the problem that the existing method cannot generate a unified model from semantically compatible SQL, that is, the association relationship of the tables in the model cannot be edited when recommending the model.

Description

Technical Field

[0001] The present invention relates to the technical field of multidimensional models, and in particular to a multidimensional model automatic merging method and system based on intelligent SQL parsing. Background Art

[0002] Apache Kylin's pre-calculation technology is an implementation of OLAP. It stores query views in the form of Cubes. Kyligence Enterprise enhances Kylin's Cubes by arranging dimensions in an orderly manner so that the pre-calculated data itself is consistent with the query view. Figure 1 One-to-one correspondence allows for more accurate search of pre-calculated data. In addition, Kyligence Enterprise also completes reverse engineering from query statements to Apache Kylin models, greatly reducing the threshold for users to use Kylin.

[0003] For Kyligence Enterprise, users do not need to invest too much energy in modeling knowledge, but only need to focus on their own domain knowledge, such as using SQL and commercial BI tools, and the rest will be done by the Kyligence Enterprise system. In most cases, users only need to import SQL and confirm the recommended model to get the required model. Before that, users need to add tables to the Kylin system, establish table associations, define dimensions, define measures, define aggregation groups, and other operations to get the required model. In addition, when the model already exists, users only need to execute the query statement and click one-click acceleration to generate the model that can match this query statement. On the one hand, reverse engineering from SQL to model greatly improves production efficiency, and on the other hand, due to the precise creation of Cube, it reduces the risk of Cube explosion caused by improper aggregation group definition. In the case of user-defined models and Cubes, users can know which models are used to manage which scenarios, but with model recommendation, users give up some of their control over the model to improve efficiency. However, the current existing methods cannot generate a unified model from semantically compatible SQL, that is, when recommending models, the association between tables in the model cannot be edited.

[0004] In summary, the existing methods have the problem of being unable to generate a unified model from semantically compatible SQL, that is, the association relationship between tables in the model cannot be edited when recommending the model. Summary of the invention

[0005] The purpose of the embodiments of the present invention is to provide a method and system for automatic merging of multidimensional models based on intelligent SQL parsing, so as to solve the problem that the existing methods cannot generate a unified model from semantically compatible SQL, that is, the association relationship between tables in the model cannot be edited when recommending the model.

[0006] To achieve the above object, an embodiment of the present invention provides a multidimensional model automatic merging method based on intelligent SQL parsing, the method comprising the following steps:

[0007] Convert the query statements into multiple logical execution plans through syntax parsing and semantic parsing in sequence;

[0008] Firstly classifying the multiple fact data tables corresponding to the multiple logical execution plans, and the fact data tables with the same content are grouped together;

[0009] Based on the greedy algorithm, the association relationships of the fact data tables that are grouped together are classified for the second time to obtain all compatible logical execution plans, and all compatible logical execution plans are divided into the same execution plan subgraph;

[0010] Add a left outer join to the association relationship of the fact table of the execution plan subgraph, and cluster the execution plan subgraphs into the same existing model based on the left outer join;

[0011] Based on the existing clustering model, the existing cube set is optimized.

[0012] Furthermore, the conditions satisfied by all compatible logical execution plans are that the association relationship of the fact data table is empty and the obtained fact data table contents are the same.

[0013] Furthermore, all compatible logical execution plans satisfy the conditions that the association relationship of the fact data table is a left join, the associated dimension table has the same join key, and the obtained fact data table content is the same.

[0014] Furthermore, based on the greedy algorithm, the association relationship of the fact data tables divided into a group is classified for the second time to obtain all compatible logical execution plans, and the specific process of dividing all compatible logical execution plans into the same execution plan subgraph is as follows:

[0015] When the first fact data table in the first logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t2, and the first fact data table in the second logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t3, then the first logical execution plan and the second logical execution plan are compatible, and the first logical execution plan and the second logical execution plan are grouped into the same execution plan subgraph.

[0016] Furthermore, a left outer join is added to the association relationship of the fact data table of the execution plan subgraph. The specific process of clustering the execution plan subgraph into the same existing model based on the left outer join is as follows:

[0017] When the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer joined to table t3, and the association column t1.c2=t3.c2, and the association relationship between the fact data tables of the execution plan subgraph in the existing model is table t1 left outer joined to table t2, and the association column t1.c1=t2.c1, then table t1, table t2 and table t3 are merged into the existing model.

[0018] Furthermore, a left outer join is added to the association relationship of the fact table of the execution plan subgraph, and the specific process of clustering the execution plan subgraph into the same existing model based on the left outer join is as follows:

[0019] When the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer joined to table t4, and the association column t1.c3=t4.c3, and the association relationship between the fact data tables of the execution plan subgraph in the existing model is table t1 left outer joined to table t2, and the association column t1.c1=t2.c1, and table t2 inner joined to table t3, and the association column t2.c2=t3.c2, then tables t1, t2, t3 and t4 are merged into the existing model.

[0020] A multidimensional model automatic merging system based on intelligent SQL parsing, the system comprising:

[0021] The first module is used to convert the query statement into multiple logical execution plans through syntax analysis and semantic analysis in sequence;

[0022] The second module is used to perform a first classification on the multiple fact data tables corresponding to the multiple logical execution plans, and the fact data tables with the same content are divided into one group;

[0023] The third module is used to perform a second classification on the association relationship of the fact data tables that are grouped together based on a greedy algorithm to obtain all compatible logical execution plans, and divide all compatible logical execution plans into the same execution plan subgraph;

[0024] The fourth module is used to add a left outer join to the association relationship of the fact data table of the execution plan subgraph, and cluster the execution plan subgraphs into the same existing model based on the left outer join;

[0025] The fifth module is used to optimize the existing cube set based on the existing model of clustering.

[0026] Furthermore, the conditions satisfied by all compatible logical execution plans are that the association relationship of the fact data table is empty and the obtained fact data table contents are the same;

[0027] The conditions satisfied by all compatible logical execution plans are that the association relationship of the fact table is a left join, the associated dimension table has the same join key, and the obtained fact table content is the same.

[0028] Furthermore, the third module also includes a third processing unit:

[0029] The third processing unit is used for, when the first fact data table in the first logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t2, and the first fact data table in the second logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t3, then the first logical execution plan and the second logical execution plan are compatible, and the first logical execution plan and the second logical execution plan are divided into the same execution plan subgraph.

[0030] Further, the fourth module includes a fourth processing unit;

[0031] The fourth processing unit is used for merging table t1, table t2 and table t3 into the existing model when the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer join table t3, and the association column t1.c2=t3.c2, and the association relationship of the fact data tables in the execution plan subgraph in the existing model is table t1 left outer join table t2, and the association column t1.c1=t2.c1;

[0032] The fourth processing unit is also used for merging tables t1, t2, t3 and t4 into the existing model when the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer connected to table t4, and the association column t1.c3=t4.c3, and the association relationship of the fact data tables of the execution plan subgraph in the existing model is table t1 left outer connected to table t2, and the association column t1.c1=t2.c1, and table t2 inner connected to table t3, and the association column t2.c2=t3.c2.

[0033] The embodiments of the present invention have the following advantages:

[0034] The present invention converts the query statements into multiple logical execution plans through grammatical analysis and semantic analysis in sequence; performs a first classification on the multiple fact data tables corresponding to the multiple logical execution plans, and the fact data tables with the same content are divided into a group; performs a second classification on the association relationship of the fact data tables divided into a group based on the greedy algorithm to obtain all compatible logical execution plans, and divides all compatible logical execution plans into the same execution plan subgraph; adds a left outer join to the association relationship of the fact data table in the execution plan subgraph, and clusters the execution plan subgraph into the same existing model based on the left outer join; optimizes the existing cube set based on the clustered existing model. The present application merges the fact data table into the existing model through a left join, thereby effectively controlling the growth of the number of models, avoiding data dispersion in different models and repeated calculations during construction to the greatest extent, and ultimately saving system resources and improving system stability. The present application solves the problem that the existing method cannot generate a unified model from semantically compatible SQL, that is, the association relationship of the table in the model cannot be edited when recommending the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings required for the embodiments or the description of the prior art are briefly introduced below. Obviously, the drawings described below are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0036] The structures, proportions, sizes, etc. illustrated in this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with the technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantial technical significance. Any structural modification, change in proportion or adjustment of size shall still fall within the scope of the technical contents disclosed in the present invention without affecting the effects and purposes that can be achieved by the present invention.

[0037] Figure 1 A method flow chart of a multidimensional model automatic merging method based on intelligent SQL parsing provided by an embodiment of the present invention;

[0038] Figure 2 A framework diagram of a multidimensional model automatic merging system based on intelligent SQL parsing provided by an embodiment of the present invention;

[0039] Figure 3 A framework diagram of a third processing unit in a multidimensional model automatic merging system based on intelligent SQL parsing provided by an embodiment of the present invention;

[0040] Figure 4A framework diagram of the fourth processing unit in the multidimensional model automatic merging system based on intelligent SQL parsing provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The following is a description of the implementation of the present invention by specific embodiments. People familiar with the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] Example 1

[0043] Combination Figure 1 The present embodiment is described as follows. The present embodiment is a multidimensional model automatic merging method based on intelligent SQL parsing, and the method comprises the following steps:

[0044] S1: Convert the query statements into multiple logical execution plans through syntax parsing and semantic parsing in sequence;

[0045] S2: performing a first classification on the multiple fact data tables corresponding to the multiple logical execution plans, and the fact data tables with the same content are grouped together;

[0046] S3: Based on the greedy algorithm, the association relationships of the fact data tables that are grouped together are classified for the second time to obtain all compatible logical execution plans, and all compatible logical execution plans are divided into the same execution plan subgraph;

[0047] S4: Add a left outer join to the association relationship of the fact table of the execution plan subgraph, and cluster the execution plan subgraphs into the same existing model based on the left outer join;

[0048] S5: Based on the existing clustering model, optimize the existing cube set.

[0049] In this implementation, the query statements are converted into multiple logical execution plans through syntax parsing and semantic parsing in turn; multiple fact data tables corresponding to the multiple logical execution plans are first classified, and fact data tables with the same content are divided into a group; the association relationship of the fact data tables divided into a group is secondly classified based on the greedy algorithm to obtain all compatible logical execution plans, and all compatible logical execution plans are divided into the same execution plan subgraph; a left outer join is added to the association relationship of the fact data table in the execution plan subgraph, and the execution plan subgraph is clustered into the same existing model based on the left outer join; based on the clustered existing model, the existing cube set is optimized. The present application merges the fact data table into the existing model through a left join, thereby effectively controlling the growth of the number of models, minimizing the dispersion of data in different models and repeated calculations during construction, and ultimately saving system resources and improving system stability. The present application solves the problem that the existing method cannot generate a unified model from semantically compatible SQL, that is, the association relationship of the table in the model cannot be edited when recommending the model.

[0050] In this implementation, the number of models is reduced by modifying the association relationship of the original model table, so that the Cube data of the same business can be managed by the same model.

[0051] In the process of modifying the association relationship of the original model table in this implementation, because it is an incremental addition under the premise of semantic correctness, the existing table association relationship, dimension and measurement information on the model are destroyed, so the existing Cube will not be affected.

[0052] The logical execution plan in this implementation is similar to the logical execution plan of Spark and Calcite, and it encapsulates the context information of the subquery.

[0053] The execution plan subgraph is a graph representation obtained by classifying and arranging compatible logical execution plans using a greedy algorithm.

[0054] Left outer join is the same as the left outer join in standard SQL.

[0055] Inner Join is the same as the inner join in standard SQL

[0056] Cube collection is a description collection of Apache Kylin's materialized views.

[0057] In a preferred embodiment, the conditions satisfied by all compatible logic execution plans of this implementation mode are that the association relationship of the fact data table is empty and the obtained fact data table contents are the same.

[0058] In a preferred embodiment, all compatible logical execution plans of this implementation method satisfy the following conditions: the association relationship of the fact data table is a left join, the associated dimension table has the same join key, and the obtained fact data table has the same content.

[0059] In this implementation, if the association relationship of the fact data table is an inner join, it must be incompatible, and they will be divided into two different execution plan subgraphs.

[0060] In a preferred embodiment, the present implementation performs a second classification on the association relationship of the fact data tables divided into a group based on a greedy algorithm to obtain all compatible logical execution plans, and divides all compatible logical execution plans into the same execution plan subgraph. The specific process is as follows:

[0061] When the first fact data table in the first logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t2, and the first fact data table in the second logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t3, then the first logical execution plan and the second logical execution plan are compatible, and the first logical execution plan and the second logical execution plan are grouped into the same execution plan subgraph.

[0062] In a preferred embodiment, this implementation adds a left outer join to the association relationship of the fact data table of the execution plan subgraph, and the specific process of clustering the execution plan subgraph into the same existing model based on the left outer join is as follows:

[0063] When the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer joined to table t3, and the association column t1.c2=t3.c2, and the association relationship between the fact data tables of the execution plan subgraph in the existing model is table t1 left outer joined to table t2, and the association column t1.c1=t2.c1, then table t1, table t2 and table t3 are merged into the existing model.

[0064] In this implementation mode, before the matching logic is modified in this application, this is considered incompatible, resulting in the selected model being empty, and a new model will be created when optimizing the model. The improved matching logic will consider that the execution plan subgraph and the existing model are matchable, so the model will be reused. During the model optimization process, the association relationship between the tables in the subgraph will be used to enhance the existing model, so that the association relationship changes from one group to two groups.

[0065] In this implementation, a stable sorting algorithm is used for the existing model, so that the results of multiple executions of the system are stable.

[0066] In a preferred embodiment, this implementation adds a left outer join to the association relationship of the fact data table of the execution plan subgraph, and the specific process of clustering the execution plan subgraph into the same existing model based on the left outer join is as follows:

[0067] When the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer joined to table t4, and the association column t1.c3=t4.c3, and the association relationship between the fact data tables of the execution plan subgraph in the existing model is table t1 left outer joined to table t2, and the association column t1.c1=t2.c1, and table t2 inner joined to table t3, and the association column t2.c2=t3.c2, then tables t1, t2, t3 and t4 are merged into the existing model.

[0068] In this implementation mode, before this application, only a new model can be created. The improved logic will consider that the execution plan subgraph and the existing model are compatible, so the model will be reused, and the subsequent association relationship will change from two groups to three groups. Table t2 also has an inner join relationship with other tables, but it has no effect on table t1, so it can still be determined to be compatible.

[0069] Example 2

[0070] Combination Figures 2 to 4 The present embodiment is a multidimensional model automatic merging system based on intelligent SQL parsing, the system comprising:

[0071] The first module 10 is used to convert the query statement into multiple logical execution plans through syntax analysis and semantic analysis in sequence;

[0072] The second module 20 is used to first classify the multiple fact data tables corresponding to the multiple logical execution plans, and the fact data tables with the same content are divided into one group;

[0073] The third module 30 is used to perform a second classification on the association relationship of the fact data tables grouped together based on a greedy algorithm to obtain all compatible logical execution plans, and to group all compatible logical execution plans into the same execution plan subgraph;

[0074] The fourth module 40 is used to add a left outer join to the association relationship of the fact data table of the execution plan subgraph, and cluster the execution plan subgraph into the same existing model based on the left outer join;

[0075] The fifth module 50 is used to optimize the existing cube set based on the existing clustering model.

[0076] This implementation method obtains multiple query statements and parses the corresponding query statements into one or more logical execution plans; groups the first table traversed by the logical execution plan, and groups the association relationship of the first table in the logical execution plan based on a greedy algorithm, and divides all compatible logical execution plans into the same execution plan subgraph, and adds a left outer join to the association relationship of the table in the execution plan subgraph to expand the existing model; updates the Cube collection using the association relationship information of the table, the added dimension information and the measurement information to obtain the final Cube collection. The existing method has the problem that it is impossible to generate a unified model from semantically compatible SQL, that is, the association relationship of the table in the model cannot be edited when recommending the model.

[0077] In a preferred embodiment, the conditions satisfied by all compatible logic execution plans of this implementation mode are that the association relationship of the fact data table is empty and the content of the obtained fact data table is the same;

[0078] The conditions satisfied by all compatible logical execution plans are that the association relationship of the fact table is a left join, the associated dimension table has the same join key, and the obtained fact table content is the same.

[0079] In a preferred embodiment, the third module 30 of this embodiment further includes a third processing unit 301:

[0080] The third processing unit 301 is used for when the first fact data table in the first logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t2, and the first fact data table in the second logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t3, then the first logical execution plan and the second logical execution plan are compatible, and the first logical execution plan and the second logical execution plan are divided into the same execution plan subgraph.

[0081] In a preferred embodiment, the fourth module 40 of this embodiment includes a fourth processing unit 401;

[0082] The fourth processing unit 401 is used for merging table t1, table t2 and table t3 into the existing model when the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer join table t3, and the association column t1.c2=t3.c2, and the association relationship of the fact data tables in the execution plan subgraph in the existing model is table t1 left outer join table t2, and the association column t1.c1=t2.c1;

[0083] The fourth processing unit 401 is also used to merge table t1, table t2, table t3 and t4 into the existing model when the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer connected to table t4, and the association column t1.c3=t4.c3, and the association relationship of the fact data tables of the execution plan subgraph in the existing model is table t1 left outer connected to table t2, and the association column t1.c1=t2.c1, and table t2 inner connected to table t3, and the association column t2.c2=t3.c2.

[0084] Although the present invention has been described in detail above by general description and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. A multidimensional model automatic merging method based on intelligent SQL parsing, characterized in that: The method comprises the following steps: Convert the query statements into multiple logical execution plans through syntax parsing and semantic parsing in sequence; Firstly classifying the multiple fact data tables corresponding to the multiple logical execution plans, and the fact data tables with the same content are grouped together; Based on the greedy algorithm, the association relationships of the fact data tables that are grouped together are classified for the second time to obtain all compatible logical execution plans, and all compatible logical execution plans are divided into the same execution plan subgraph; Add a left outer join to the association relationship of the fact table of the execution plan subgraph, and cluster the execution plan subgraphs into the same existing model based on the left outer join; Based on the existing clustering model, the existing cube set is optimized.

2. The multidimensional model automatic merging method based on intelligent SQL parsing according to claim 1, characterized in that: The conditions satisfied by all compatible logical execution plans are that the association relationship of the fact table is empty and the content of the obtained fact table is the same.

3. The multidimensional model automatic merging method based on intelligent SQL parsing according to claim 1, characterized in that: The conditions satisfied by all compatible logical execution plans are that the association relationship of the fact table is a left join, the associated dimension table has the same join key, and the obtained fact table content is the same.

4. The multidimensional model automatic merging method based on intelligent SQL parsing according to claim 1, characterized in that: The specific process of performing a second classification on the association relationships of the fact data tables that are divided into a group based on the greedy algorithm to obtain all compatible logical execution plans and divide all compatible logical execution plans into the same execution plan subgraph is as follows: When the first fact data table in the first logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t2, and the first fact data table in the second logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t3, then the first logical execution plan and the second logical execution plan are compatible, and the first logical execution plan and the second logical execution plan are grouped into the same execution plan subgraph.

5. The multidimensional model automatic merging method based on intelligent SQL analysis according to claim 1, characterized in that: The specific process of adding a left outer join to the association relationship of the fact table of the execution plan subgraph and clustering the execution plan subgraph into the same existing model based on the left outer join is as follows: When the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer joined to table t3, and the association column t1.c2=t3.c2, and the association relationship between the fact data tables of the execution plan subgraph in the existing model is table t1 left outer joined to table t2, and the association column t1.c1=t2.c1, then table t1, table t2 and table t3 are merged into the existing model.

6. The multidimensional model automatic merging method based on intelligent SQL analysis according to claim 1, characterized in that: The specific process of adding a left outer join to the association relationship of the fact table of the execution plan subgraph and clustering the execution plan subgraph into the same existing model based on the left outer join is as follows: When the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer joined to table t4, and the association column t1.c3=t4.c3, and the association relationship between the fact data tables of the execution plan subgraph in the existing model is table t1 left outer joined to table t2, and the association column t1.c1=t2.c1, and table t2 inner joined to table t3, and the association column t2.c2=t3.c2, then tables t1, t2, t3 and t4 are merged into the existing model.

7. A multidimensional model automatic merging system based on intelligent SQL parsing, characterized in that: The system comprises: The first module is used to convert the query statement into multiple logical execution plans through syntax analysis and semantic analysis in sequence; The second module is used to perform a first classification on the multiple fact data tables corresponding to the multiple logical execution plans, and the fact data tables with the same content are divided into one group; The third module is used to perform a second classification on the association relationship of the fact data tables that are grouped together based on a greedy algorithm to obtain all compatible logical execution plans, and divide all compatible logical execution plans into the same execution plan subgraph; The fourth module is used to add a left outer join to the association relationship of the fact data table of the execution plan subgraph, and cluster the execution plan subgraphs into the same existing model based on the left outer join; The fifth module is used to optimize the existing cube set based on the existing model of clustering.

8. The multidimensional model automatic merging system based on intelligent SQL analysis as claimed in claim 7, characterized in that: The conditions that all compatible logical execution plans meet are that the association relationship of the fact data table is empty and the content of the obtained fact data table is the same; The conditions satisfied by all compatible logical execution plans are that the association relationship of the fact table is a left join, the associated dimension table has the same join key, and the obtained fact table content is the same.

9. The multidimensional model automatic merging system based on intelligent SQL analysis according to claim 7, characterized in that: The third module also includes a third processing unit: The third processing unit is used for, when the first fact data table in the first logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t2, and the first fact data table in the second logical execution plan is t1, and the association relationship between the tables in the first fact data table is table t1 left outer join table t3, then the first logical execution plan and the second logical execution plan are compatible, and the first logical execution plan and the second logical execution plan are divided into the same execution plan subgraph.

10. The multidimensional model automatic merging system based on intelligent SQL analysis according to claim 7, characterized in that: The fourth module includes a fourth processing unit; The fourth processing unit is used for merging table t1, table t2 and table t3 into the existing model when the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer join table t3, and the association column t1.c2=t3.c2, and the association relationship of the fact data tables in the execution plan subgraph in the existing model is table t1 left outer join table t2, and the association column t1.c1=t2.c1; The fourth processing unit is also used for merging tables t1, t2, t3 and t4 into the existing model when the association relationship between the fact data tables in the execution plan subgraph is table t1 left outer connected to table t4, and the association column t1.c3=t4.c3, and the association relationship of the fact data tables of the execution plan subgraph in the existing model is table t1 left outer connected to table t2, and the association column t1.c1=t2.c1, and table t2 inner connected to table t3, and the association column t2.c2=t3.c2.

Images