Medical Data Assembly System
By designing a medical data assembly system, using the coordinated work of the processor and database, data assembly rules are automatically generated, which solves the problem of messy and difficult assembly of original business data, and realizes efficient and high-quality special disease data assembly, and improves scientific research efficiency.
Patent Information
- Application Number
- CN202410416859.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-08
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2044-04-08
AI Technical Summary
In medical data processing, the original business data is messy and difficult to assemble into specific structures and data for specialized diseases, resulting in inefficiency of scientific researchers in research on specialized patients.
A medical data assembly system was designed, through the coordinated work of processor, memory and database, and the query keywords and assembly rule description text entered by the user are used to perform data preprocessing, semantic annotation, and grammatical analysis to automatically generate data assembly rules. Achieve efficient assembly of medical data.
It improves the efficiency and quality of data assembly, and can automatically generate data assembly rules based on the assembly rules entered by users, simplifying the acquisition process of special disease data and improving research efficiency.
Smart Images

Figure CN118170792B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data processing, and in particular to a medical data assembly system. Background Art
[0002] In some application scenarios, for example, for research needs, doctors need to assemble and process general clinical medical data into disease-specific structures and data to help them efficiently conduct research on patients with specific diseases. However, the original business data is large and complex, and how to assemble and process the original business data in the clinical system into the data required for a specific disease will be a challenging task. Summary of the invention
[0003] In view of the above technical problems, the technical solution adopted by the present invention is:
[0004] According to a first aspect of the present invention, a medical data assembly system is provided, the system comprising a processor, a memory and a database connected to the processor in communication, the memory storing a computer program, the database storing an original medical data record table I={I1, I2, ..., I i , ..., I n};I i is the i-th medical service domain information in I, i ranges from 1 to n, and n is the number of medical data records in I; I i =(US i , DS i ), US i For I i Corresponding medical business domain ID, DS i For I i The corresponding field information.
[0005] When the computer program is executed by the processor, it is used to implement the following steps:
[0006] S100, based on the query keyword input by the user, obtain the corresponding medical data from I, and obtain the query data record table A = {A1, A2, ..., A j , ..., A m}; A j is the jth query data in A, A j =(B j , C j ), B j A j The ID of the corresponding object, C j For B j The medical data records corresponding to the corresponding objects, j ranges from 1 to m, and m is the number of query data in A; C i ={Ci1 , C i2 , ..., C ih , ..., C if(j)}, C ih C i The hth medical service domain information in B, where h ranges from 1 to f(j), and f(j) is j The number of medical business domains corresponding to the corresponding objects; C ih =(UM jh ,DM jh ), UM ih C ih The ID of the corresponding medical business domain, DM ih For UM ih The corresponding field information.
[0007] S200, obtaining the target metadata structure table T input by the user and the corresponding assembly rule description text set Q; wherein T={T1, T2, ……, T u , ..., T p}, T u is the u-th medical service domain information in T, T u =(UT u , DT u ), UT u is the ID of the u-th medical service domain in T, DT u is the field information corresponding to the u-th medical service domain in T; the value of u ranges from 1 to p, where p is the number of medical service domains in T; Q = {Q1, Q2, ..., Q r , ..., Q h}, Q r The assembly rule description text corresponding to the rth field corresponding to T, r ranges from 1 to M, and M is the number of fields corresponding to T; the assembly rule description text includes at least a first description text, and the first description text includes at least a target field, a source field, a condition description, and a calculation logic description.
[0008] S300, for any assembly rule description text Q r The first description text in is preprocessed to obtain the corresponding first processed text PT1 r ={(TF:TFC r ), (SF:SFC r ), (CT:CTC r ), (CL:CLC r )}, where TF represents the target field, TFC r The field corresponding to the target field, SF represents the source field, SFC r The field corresponding to the source field, CT represents the condition, CTCr is the sentence after word segmentation corresponding to CT, CL represents the calculation logic, CLC r is the sentence after word segmentation corresponding to CL.
[0009] S400, for PT1 r Semantically annotate each word and phrase in the text to obtain the corresponding second processed text PT2 r .
[0010] S500, based on PT2 r Generate the corresponding syntax analysis tree PTE r , and based on PTE r , T and I get AT r The corresponding query statement QS r .
[0011] S600, based on QS r , get the corresponding data from T as the field value FV corresponding to the rth field in T r .
[0012] The present invention has at least the following beneficial effects:
[0013] The medical data assembly system provided by the embodiment of the present invention can automatically generate corresponding data assembly rules based on the assembly rule description text input by the user, thereby improving the data assembly efficiency and the quality of assembled data.
[0014] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0016] Figure 1 A flow chart of a method implemented when a medical data assembly system according to an embodiment of the present invention executes a computer program;
[0017] Figure 2 This is a schematic diagram of a grammatical analysis tree obtained according to an embodiment of the present invention. DETAILED DESCRIPTION
[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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 those skilled in the art without creative work are within the scope of protection of the present invention.
[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.
[0020] It should be noted that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the steps as sequential processes, many of the steps can be implemented in parallel, concurrently or simultaneously. In addition, the order of the steps can be rearranged. The process can be terminated when its operation is completed, but it can also have additional steps not included in the accompanying drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.
[0021] An embodiment of the present invention provides a medical data assembly system, which may include a processor and a memory and a database that are communicatively connected to the processor.
[0022] The memory stores a computer program. The database stores an original medical data record table I = {I1, I2, ..., I i , ..., I n};I i is the i-th medical service domain information in I, i ranges from 1 to n, and n is the number of medical data records in I. i =(US i , DS i ), US i For I i Corresponding medical business domain ID, DS i For I i Corresponding field information, DS i ={DS i1 , DS i2 , ..., DS ik , ..., DS ig(i)}, DS ik For DS i The kth field information in DS ik =(CS ik , E.S.ik , G.S. ik , R.S. ik ), CS ik For DS i The Chinese name of the kth field in ES ik For DS i The English name of the kth field in GS ik For DS i The field description of the kth field in RS ik For DS i The field value corresponding to the kth field in ; the value of k ranges from 1 to g(i), where g(i) is DS i The number of fields in .
[0023] In an embodiment of the present invention, I may be all medical data generated by a designated hospital. The ID of the medical service domain may be the name of the medical service domain, and the medical service domain may include basic patient information, basic medical information, outpatient registration information table, outpatient medical record, diagnosis, outpatient prescription drug information, outpatient transfer association table, emergency observation medical record, test application, immunohistochemistry, etc. The field information of each medical service domain may be set based on actual needs. For example, the fields corresponding to the basic patient information may include the medical institution code, patient number, place of birth, date of birth, nationality, native place, patient name, medical insurance card number, certificate type, certificate number, resident health card number, patient phone number, marital status, ethnicity, blood type, data source, etc. The field description may be set based on actual needs. For example, it may include the meaning and purpose of the field, data type, whether it is an enumeration type, primary and foreign key identifiers, primary and foreign key association tables, etc. The field value is the value corresponding to the field. For example, the field value of the patient's name is Zhang Moumou, and the date of birth is 19850302, etc.
[0024] Furthermore, in an embodiment of the present invention, when the computer program is executed by the processor, it is used to implement the following steps:
[0025] S100, based on the query keyword input by the user, obtain the corresponding medical data from I, and obtain the query data record table A = {A1, A2, ..., A j , ..., A m}.
[0026] Among them, A j is the jth query data in A, A j =(B j , C j ), B j A j The ID of the corresponding object, C j For B jThe medical data records corresponding to the corresponding objects, j ranges from 1 to m, and m is the number of query data in A; C i ={C i1 , C i2 , ..., C ih , ..., C if(j)}, C ih C i The hth medical service domain information in B, where h ranges from 1 to f(j), and f(j) is j The number of medical business domains corresponding to the corresponding objects; C ih =(UM jh ,DM jh ), UM ih C ih The ID of the corresponding medical business domain, DM ih For UM ih Corresponding field information;DM ih ={DM 1 ih ,DM 2 ih , ..., DM t ih , ..., DM z(j,h) ih},DM t ih For DM ih The tth field information in , the value of t is 1 to z(j, h), z(j, h) is DM ih The number of fields in DM t ih =(CM t jh , E.M. t jh , GM t jh , R.M. t jh ), CM t jh For DM ih The Chinese name of the tth field in EM t jh For DM ih The English name of the tth field in GM t jh For DM ih The field description of the tth field in RM t jh For DM ih The field value of the tth field in .
[0027] In an embodiment of the present invention, the query keyword can be set based on the disease type required by the user, for example, the query keyword is "diagnosed as lung malignant tumor", etc. By querying the keyword, medical data related to the disease type required by the user can be obtained from the full amount of data.
[0028] S200, obtaining the target metadata structure table T and the corresponding assembly rule description text set Q input by the user.
[0029] Where, T = {T1, T2, ..., T u , ..., T p}, T u is the u-th medical service domain information in T, T u =(UT u , DT u ), UT u is the ID of the u-th medical service domain in T, DT u is the field information corresponding to the u-th medical business domain in T; the value of u ranges from 1 to p, and p is the number of medical business domains in T. The field value corresponding to each field in T is a null value.
[0030] Those skilled in the art know that the user can set a corresponding target metadata structure table based on the required disease type.
[0031] Furthermore, Q = {Q1, Q2, ..., Q r , ..., Q h}, Q r The assembly rule description text corresponding to the rth field corresponding to T, r ranges from 1 to M, and M is the number of fields corresponding to T; the assembly rule description text includes at least a first description text, and the first description text includes at least a target field, a source field, a condition description, and a calculation logic description.
[0032] In an embodiment of the present invention, the assembly rule description text may be a natural language description text. In an illustrative embodiment, the first description text may be "generate the first operation time, the source is the operation time, the condition is the first operation time whose diagnosis includes lung malignant tumor and the operation name includes lung resection."
[0033] S300, for any assembly rule description text Q r The first description text in is preprocessed to obtain the corresponding first processed text PT1 r ={(TF:TFC r ), (SF:SFC r ), (CT:CTC r ), (CL:CLC r )}, where TF represents the target field, TFC rThe field corresponding to the target field, SF represents the source field, SFC r The field corresponding to the source field, CT represents the condition, CTC r is the sentence after word segmentation corresponding to CT, CL represents the calculation logic, CLC r is the sentence after word segmentation corresponding to CL.
[0034] In an embodiment of the present invention, the preprocessing may include performing Chinese word segmentation, keyword standardization, and clause segmentation on the first description text in sequence. The format of keyword standardization is: target field: (field corresponding to the target field), source field: (field corresponding to the source field), condition: (statement corresponding to the condition), calculation logic: (statement corresponding to the calculation logic).
[0035] In an embodiment of the present invention, an existing word segmentation tool such as IK Analyzer 3.0 Chinese word segmenter can be used for word segmentation processing. Taking the first description text "generating the first operation time, the source is the operation time, the condition is the first operation time when the diagnosis includes lung malignant tumor and the operation name includes lung resection" as an example, after preprocessing the text, the first processed text obtained is: "target field: first operation time", "source field: operation time", "condition: diagnosis name includes lung malignant tumor and operation name includes lung resection", "calculation logic: first operation time".
[0036] S400, for PT1 r Semantically annotate each word and phrase in the text to obtain the corresponding second processed text PT2 r .
[0037] In an embodiment of the present invention, the semantic tag definition includes:
[0038] source: source, expressing the location where the data value is taken.
[0039] target: target, which expresses the fields to be assembled and generated.
[0040] condition: condition, expressing the filtering condition.
[0041] eva: value, such as "seek", "calculate", "get", etc.
[0042] expression: expression, which expresses a complete statement.
[0043] comp_op: comparison operator, the comparison relationship between expressions, such as "greater than", "equal to", "greater than or equal to", "less than or equal to", "not equal to", "earlier than", and "later than".
[0044] range_op: range operator, range relationship between expressions, such as "between...", "not between...".
[0045] math_op: Mathematical operators, mathematical operation relationships between expressions, such as addition, subtraction, multiplication, division, cube root, square root, absolute value, etc.
[0046] pattern_op: pattern operator, character pattern relationship between expressions, such as contains, has, match, regular expression, etc.
[0047] logic_op: logical operator, the logical relationship between expressions, such as and, and, or, not, etc.
[0048] func: function, the operation function of expression, such as maximum value, minimum value, number of times, sum, average, all values, any value, number of times, etc.
[0049] field: field, a special type of expression, which is a field in the expression table, such as diagnosis name, operation name, operation time, drug name, chief complaint, examination findings, etc.
[0050] Statement modes include:
[0051] Binocular infix: expression op expression
[0052] Single prefix: op expression
[0053] Unary suffix: expression op
[0054] Function call: func([expression[,expression...]])
[0055] In an embodiment of the present invention, a trained deep neural network model is used for semantic annotation. The trained deep neural network model can be obtained by the following steps:
[0056] S401, obtaining a sample data set, including Q sample data, each sample data including a sample sentence with a defined semantic label. The format of the sample sentence is the same as the format of the first processed text.
[0057] In the embodiment of the present invention, the sample data may be manually labeled data.
[0058] In an embodiment of the present invention, the deep neural network used can be built through the deep learning library provided by TensorFlow.
[0059] S402, input the current batch of training sample data into the current deep neural network model for training to obtain corresponding prediction results; the prediction results include semantic labels for each word and phrase in the sample sentence.
[0060] S403, based on the prediction results of the current batch and the corresponding actual results, obtain the current loss function value of the current deep neural network model, and determine whether the current loss function value meets the preset model training end condition. If yes, execute step S405, otherwise, execute step S404.
[0061] S404: Update the parameters of the current deep neural network model based on the current loss function value, and use the next batch of sample data as the training sample data of the current batch, and execute S402.
[0062] S405: Use the current deep neural network model as the trained deep neural network model.
[0063] In an embodiment of the present invention, the loss function value can be calculated based on an existing loss function. The preset model training end condition can be set based on actual needs. For example, the loss is less than or equal to a set loss threshold and remains unchanged within a set time period.
[0064] S500, based on PT2 r Generate the corresponding syntax analysis tree PTE r , and based on PTE r , T and I get AT r The corresponding query statement QS r .
[0065] In the embodiment of the present invention, the rule parsing engine can be implemented by ANTLR4. ANTLR4 can generate a top-down syntax analyzer called ALL(*). ALL(*) means that it can use all the remaining input text to make decisions. The root node of each subtree of the syntax analysis tree uses an abstract name to name the other contained elements, that is, the root node of the subtree corresponds to the name of the grammar rule, and the leaf nodes of the tree are symbols or lexical symbols in the sentence. Taking the first processed text as an example, the syntax analysis tree generated by using ANTLR4 can be as follows Figure 2 shown.
[0066] Further, in the embodiment of the present invention, the PTE-based r , T and preset medical metadata structure to obtain AT r The corresponding query statement QS r , which may specifically include:
[0067] S501, using accessors from PTEr The visit() function is called at the root node of .
[0068] S502, PTE r All child nodes in are passed as parameters to the visit() function, and each node of the syntax analysis tree is visited based on the depth-first traversal method to obtain the corresponding visit result V r ; V r Based on T and I.
[0069] In the embodiment of the present invention, the access result is the English name of the leaf node of each subtree, and the corresponding field English name can be obtained by looking up T and I through the Chinese field name corresponding to the node.
[0070] S503, based on V r and a preset query statement expression to generate the query statement.
[0071] In an embodiment of the present invention, the query statement may be a SQL statement, and the preset query statement expression may be set based on actual needs. Figure 2 Taking the syntax analysis tree shown in the figure as an example, the query statement obtained can be:
[0072] select first_date(operation_record.OPERATION_TIME) as first_op_date,patient_info.PATIENT_SN as SN from cdm_patiens where diagnose.DIAGNOSIS contains 'Lung Malignancy' and operation_record.OPERATION_NAME contains Lung Resection groupby patient_info.PATIENT_SN.
[0073] S600, based on QS r , get the corresponding data from T as the field value FV corresponding to the rth field in T r .
[0074] Further, in the embodiment of the present invention, S600 may specifically include:
[0075] S601, for QS r Perform optimization processing to obtain the optimized QS r .
[0076] In the embodiment of the present invention, the optimization process at least includes simplifying the conditional expressions in the query statement and performing equivalent replacement on the logical expressions in the query statement.
[0077] The simplification of the conditional expression in the query statement may specifically include:
[0078] (1) If there is no group by condition expression or no aggregate function in the query statement, merge the having condition expression into the where condition expression;
[0079] (2) Delete redundant parentheses in the conditional expressions in the query statement to reduce the level of and and or trees. For example, (A and B) and (C and D) can be simplified to A and B and C and D.
[0080] (3) Extract and calculate the operations corresponding to the conditional expressions in the query statement, for example, where field = 3*10 is transformed into where field = 30.
[0081] (4) Transform the inequalities in the query. For example, where f1>0.1and f2=3.0andf1>0.2 is simplified to where f1>0.2and f2=3.0.
[0082] Furthermore, performing equivalent replacement processing on the logical expression in the query statement may include:
[0083] 1) Like predicate replacement. Like is a pattern matching comparison operation, which is usually time-consuming. Rewriting the like rule can effectively improve query performance. For example, f1 like'infection' has no wildcard and is equivalent to =. It can be replaced with f1 = 'infection'.
[0084] 2) Between-and range operation. For example, f1 between 0.1and 1.2 can be replaced with f1>=0.1and f1<=1.2, which can effectively use the index and avoid full table scan.
[0085] 3) in is converted to or. For example, f1 in('trastuzumab','apatinib mesylate','pembrolizumab') is replaced with f1='trastuzumab'or f1='apatinib mesylate'or f1='pembrolizumab'.
[0086] 4) Equivalent replacement of not predicates. For example, not(f1>0.2) is replaced by f1<=0.2; not(f1!=1.0) is replaced by f1=1.0; not(f1<0.2) is replaced by f1>=0.2; not(f1=1.0) is replaced by f1!=1.0. S602, based on the optimized QS r , perform database operations on T to obtain the corresponding data.
[0087] In the embodiment of the present invention, by performing the above optimization processing on the query statement, the query can be made more accurate and more efficient.
[0088] Furthermore, in an embodiment of the present invention, the assembly rule description text also includes at least one second description text, and the second description text at least includes a condition description field.
[0089] The number of second description texts can be set based on actual needs. If there are multiple second description texts, the relationship between the multiple second description texts can be an and or or relationship. The second description text is a filter condition for filtering the query results of the query statement obtained based on the first description text, which can be set based on the actual needs of the user. For example, the second description text can be "the medication time is greater than the first surgery time" and so on.
[0090] Furthermore, when the computer program is executed by the processor, it is also used to implement the following steps:
[0091] S700: If Q includes a second description text, based on the second description text corresponding to the rth field, perform a r Processing is performed so that the processed FV r Meet the conditions corresponding to the corresponding second description text, and process the FV r As the target field value corresponding to the rth field in T.
[0092] It is known to those skilled in the art that any second description text corresponding to the rth field is used to describe FV r Processing is performed so that the processed FV r Methods that meet the conditions corresponding to the corresponding second description texts all belong to the protection scope of the present invention. For example, if the second description texts are in an or relationship, then FV can be determined based on each second description text. r If the second description texts are in an and relationship, the corresponding second description texts are used to filter FV in a progressive relationship. r Perform filtering processing to obtain corresponding processing results.
[0093] In the embodiment of the present invention, by setting the second description text, the acquired data is filtered to obtain the data required by the user, which can make the configuration of the assembly rule more flexible and improve the user experience.
[0094] In summary, the data assembly system provided by the embodiment of the present invention is not limited to specific diseases or specific data structure models, and can flexibly define the field generation rules to be assembled. Moreover, it can automatically convert the field generation rules described in natural language into a computer-recognizable calculation logic rule method, with high automation, simple and fast assembly and production of disease-specific data, and provides the advantages of efficient assembly performance and relatively high data quality.
[0095] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution disclosed in the present invention can be achieved, and this document does not limit this.
[0096] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A medical data assembly system, characterized in that: The system includes a processor, a memory and a database connected to the processor in communication, wherein the memory stores a computer program, and the database stores an original medical data record table I={I1, I2, . . . , I i , ..., I n };I i is the i-th medical service domain information in I, i ranges from 1 to n, and n is the number of medical data records in I; I i =(US i , DS i ), US i For I i Corresponding medical business domain ID, DS i For I i Corresponding field information; When the computer program is executed by the processor, it is used to implement the following steps: S100, based on the query keyword input by the user, the corresponding medical data is obtained from I, and the query data record table A={A1, A2, ..., A j , ..., A m }; A j is the jth query data in A, A j =(B j , C j ), B j A j The ID of the corresponding object, C j For B j The medical data records corresponding to the corresponding objects, j ranges from 1 to m, and m is the number of query data in A; C i ={C i1 , C i2 , ..., C ih , ..., C if(j) }, C ih C i The hth medical service domain information in B, where h ranges from 1 to f(j), and f(j) is j The number of medical business domains corresponding to the corresponding objects; C ih =(UM jh ,DM jh ), U.M. ih C ih The ID of the corresponding medical business domain, DM ih For UM ih Corresponding field information; S200, obtaining the target metadata structure table T input by the user and the corresponding assembly rule description text set Q; wherein T={T1, T2, ..., T u , ..., T p }, T u is the u-th medical service domain information in T, T u =(UT u , DT u ), UT u is the ID of the u-th medical service domain in T, DT u is the field information corresponding to the u-th medical service domain in T; the value of u ranges from 1 to p, where p is the number of medical service domains in T; Q={Q1, Q2, ..., Q r , ..., Q h }, Q r is the assembly rule description text corresponding to the rth field corresponding to T, where r is 1 to M, and M is the number of fields corresponding to T; the assembly rule description text includes at least a first description text, which includes at least a target field, a source field, a condition description, and a calculation logic description; S300, for any assembly rule description text Q r The first description text in is preprocessed to obtain the corresponding first processed text PT1 r ={(TF:TFC r ), SF: SFC r ), (CT: CTC r ), (CL:CLC r )}, where TF represents the target field, TFC r The field corresponding to the target field, SF represents the source field, SFC r The field corresponding to the source field, CT represents the condition, CTC r is the sentence after word segmentation corresponding to CT, CL represents the calculation logic, CLC r is the sentence after word segmentation corresponding to CL; S400, for PT1 r Semantically annotate each word and phrase in the text to obtain the corresponding second processed text PT2 r ; S500, based on PT2 r Generate the corresponding syntax analysis tree PTE r , and based on PTE r , T and I get AT r The corresponding query statement QS r ; S600, based on QS r , get the corresponding data from T as the field value FV corresponding to the rth field in T r ; The root node of each subtree of the grammar analysis tree corresponds to the name of the grammar rule, and the leaf nodes of the tree are symbols or lexical symbols in the sentence; The PTE-based r , T and preset medical metadata structure to obtain AT r The corresponding query statement QS r , including: S501, using accessors from PTE r Call the visit() function at the root node; S502, PTE r All child nodes in are passed as parameters to the visit() function, and each node of the syntax analysis tree is visited based on the depth-first traversal method to obtain the corresponding visit result V r The access result is the English name of the leaf node of each subtree. The access result is obtained by consulting T and I through the Chinese field name corresponding to the node; S503, based on V r and a preset query statement expression to generate the query statement.
2. The system according to claim 1, characterized in that The assembly rule description text also includes at least one second description text, and the second description text at least includes a condition description.
3. The system according to claim 2, characterized in that When the computer program is executed by the processor, it is also used to implement the following steps: S700: If Q includes a second description text, based on the second description text corresponding to the rth field, perform a r Processing is performed so that the processed FV r Meet the conditions corresponding to the corresponding second description text, and process the FV r As the target field value corresponding to the rth field in T.
4. The system according to claim 1, characterized in that S600 specifically includes: S601, for QS r Perform optimization processing to obtain the optimized QS r ; S602, based on optimized QS r , perform database operations on T to obtain the corresponding data.
5. The system according to claim 4, characterized in that The optimization process at least includes simplifying the conditional expressions in the query statement and performing equivalent replacement on the logical expressions in the query statement.
6. The system according to claim 5, characterized in that The simplification of the conditional expression in the query statement specifically includes: If there is no group by condition expression or no aggregate function in the query statement, merge the having condition expression into the where condition expression; Delete the redundant brackets in the conditional expression in the query statement; Extract and calculate the operations corresponding to the conditional expressions in the query statement; Transform the inequalities in the query statement.
7. The system according to claim 1, characterized in that Use a trained deep neural network model for semantic annotation.
8. The system according to claim 1, characterized in that In S300, IK Analyzer 3.0 Chinese word segmenter is used for word segmentation.
