CN104834700A - Method for capturing movement data increment based on track change - Google Patents

Abstract

The invention discloses a mobile data incremental capture method based on trajectory change, which uses a trigger to capture the operation log, records the operation change process of the operation data table, and uses a purification method to merge the operation log to realize the capture of the net increment and the entire operation A record of the change process. The invention makes full use of the existing trigger mechanism of the mobile database and the low frequency of mobile node write operations in the mobile computing environment, significantly improves the efficiency of increment capture, and further adopts a purification operation when extracting increments to ensure that the increments The amount of data is the least, which greatly reduces the amount of synchronous data communication, relaxes the demand for network bandwidth of the synchronous system, and reduces the cost of communication. At the same time, the operation change process recorded by the present invention can be used as an effective credential for the rollback operation of the synchronization system when the synchronization system is unexpectedly terminated or severely timed out, which affects the final consistency of data, thereby increasing the robustness of the synchronization system.

Description

A Method of Incremental Capture of Mobile Data Based on Trajectory Change

technical field

The invention relates to an incremental capture method, which belongs to the application field of mobile computing and database intersecting technologies, in particular to a trajectory change-based mobile data incremental capture method.

Background technique

With the development and popularization of mobile computing technology and wireless communication technology, mobile applications are becoming increasingly complex and diverse. In order to meet mobile users' requirements for reliable processing of mobile service data, mobile computing systems need to allow users to process data offline and support a distributed storage mechanism with multiple data copies. At present, Optimistic Replication is commonly used in distributed storage mechanisms to ensure the availability of business data and control the consistency of business data. Due to the weak connectivity of the mobile network and the relatively limited communication delay and bandwidth, it is difficult for the data synchronization mechanism in the mobile computing environment to achieve efficient data consistency.

In order to reduce the amount of synchronization data, improve synchronization efficiency, and reduce the demand for synchronization on mobile network bandwidth, mobile computing systems usually adopt an incremental capture method, that is, only the modified data is exchanged during each data synchronization. Typical incremental data capture methods include snapshot method (Snapshot), trigger method (Trigger), log method (Log) and timestamp method (Timestamp). They have their own advantages and disadvantages. Snapshot compares the old and new snapshots of the data table to extract information increments. This method is widely applicable, but requires high storage space. With the increase of data tuples, the comparison detection algorithm is easy It will become the performance bottleneck of the system. Trigger uses triggers to capture incremental data, and the incremental capture efficiency is high, but it is only used in data management systems with a trigger mechanism, and when there is a large amount of data in the data set for operations, triggers have a greater impact on system performance . Log uses the operation log of the analysis database itself to extract increments, which does not increase system overhead and is highly efficient. However, databases and management systems generally have strict permission restrictions on log access; in addition, most of the log formats of data management systems are different. Not the same, resulting in many restrictions on the use of the log method. Timestamp sets a timestamp field in the tuple. When synchronizing, extract all fields greater than the last synchronization time to obtain the increment. It is simple and easy to implement, but this method requires changing the structure of the business data set, and it is difficult to capture the delete operation , limiting its scope of application.

Most of the existing incremental data capture methods are oriented to stable distributed computing environments, and are not suitable for mobile computing environments. Simple transplantation can easily lead to low efficiency and high overhead.

Contents of the invention

The purpose of the present invention is to provide a mobile data incremental capture method, which can further reduce the traffic of synchronization data, reduce the requirement of synchronization for mobile network bandwidth, and improve the synchronization efficiency under the premise of ensuring that each mobile node realizes the final consistency of data.

The technical scheme that the present invention adopts is as follows:

A mobile data incremental capture method based on track change uses triggers provided by relational databases to capture operation logs, record the operation change process of the operation data table, and use the purification method to merge the operation logs to realize the capture of net increments and the entire operation change A record of the process. The steps are as follows:

Step 1: Define three triggers for adding, modifying, and deleting on the operation data table, respectively set the trigger conditions of the triggers, and then define an operation log table to store the captured data change track, the operation log table In addition to having all the fields of the operation data table, it also includes the primary key and the "operation type" field;

Step 2: To operate (add, modify or delete) the operation data table will trigger the corresponding pre-defined trigger, and the trigger will automatically record the operation record and operation type into the operation log table;

Step 3: group all the records in the operation log table according to the ID numbers of the operated records;

Step 4: apply the purification algorithm to compress and merge these groups, that is, purify, and obtain the purification results of each operated record;

Step 5: Combine the purification results of each operated record to obtain the purification result of the entire operation log table, that is, complete the operation of purifying the operation log table.

The situation of the trigger in the step 1 is: there are three types of operations in the relational database, which are respectively adding, modifying and deleting. According to these three types of operations, the triggers of these three types of operations are respectively defined on the operation data table , such as Add_trigger, Modify_trigger and Delete_trigger, and set their trigger conditions respectively. For example, when a new record is added on the operation data table, the new trigger Add_trigger will be triggered, and Add_trigger will record the newly added record data and the new type into the operation log table for purification processing. The situation of the operation log table in step 1 is: the operation log table must contain all the fields in the operation data table, its own primary key and the "operation type" field, so as to be able to record the complete operation process and this operation type.

The grouping of operation records in step three is: since the operation log table records the ID numbers of the operated records, a series of records in the operation log table are divided into several groups according to the ID numbers of the operated records.

Table 1 Operation type

serial number basic operation type significance serial number Composite operation type significance 1 A Add 5 A-M Modified after adding 2 m Revise 6 A-D Delete after adding 3 D. delete 7 M-M edit after edit 4 N no action 8 M-D delete after modification

The purification and merging operations in steps 4 and 5 are as follows: in a relational database, any operation sequence on the same record can be decomposed into 8 basic operation types and composite operation types, as shown in Table 1. Split the series of operations represented by each group into basic operation types and composite operation types, and then use the corresponding purification formulas from (1)-(5) for purification according to the types of operation sequences in each group. Finally, formula (6) is used to combine the purification results of each group, that is, the purification operation of the operation log table is completed.

f ^* _i (L _i )＝OP _A (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _M (l _n )＝OP _A (l _n ) (1)

f ^* _i (L _i )＝OP _A (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _D (l _n )＝NULL (2)

f ^* _i (L _i )＝OP _M (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _M (l _n )＝OP _M (l _n ) (3)

f ^* _i (L _i )＝OP _M (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _D (l _n )＝OP _D (l _n ) (4)

f ^* _i (L _i ) = OP _D (l ₁ ) = OP _D (l ₁ ) (5)

${\mathrm{<span\; class="notranslate">\; f</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{<span\; class="notranslate">\; \&cup;</span>}}{{\mathrm{<span\; class="notranslate">\; f</span>}}^{<span\; class="notranslate">\; *</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$

The meanings of the symbols in formulas (1)-(6) are: A, M, and D, as shown in Table 1, respectively represent the operation types of adding, modifying, and deleting; OP represents an operation on a record; the operation merge operator "∪" means to find the operation union of the two operands before and after it, where NULL means the operation union is empty; we regard the operation log table and the operation data table as the operation log set and the operation data set respectively, then for the operation log set The operation set L _i (1≤i≤n, i∈N) of the i-th record in the operation data set in L, assuming L _i ={l ₁ ,l ₂ ,…l _k …l _n }(1≤k≤ n, k∈N), OP _Γ (l _k ), Γ∈{A,M,D} means that the k-th operation type on the i-th record is Γ, and the result is l _k ; f ^* _i (L _i ) Represents the optimal equivalent operation function of L _i .

The present invention has the following beneficial effects:

1. The existing incremental data capture method is oriented to a stable distributed computing environment and is not suitable for a mobile computing environment. Simple transplantation can easily lead to low efficiency and high overhead. The invention adopts the trigger to capture the operation log, records the operation change process of the operation data table, makes full use of the existing trigger mechanism of the mobile database and the low frequency of the mobile node write operation in the mobile computing environment, and significantly improves the incremental capture s efficiency.

2. Due to the weak connectivity of the mobile network and the relatively limited communication delay and bandwidth, it is required to reduce the transmitted data as much as possible in the process of mobile data synchronization. The present invention will further purify the captured incremental data to ensure that the incremental data At least, it greatly reduces the amount of synchronous data communication, relaxes the requirements of the synchronous system for network bandwidth, and reduces the cost of communication.

3. In the process of mobile data synchronization, the synchronization server is responsible for the main responsibility of synchronization, and the transmission of too much synchronization data will greatly increase the burden on the synchronization server. The present invention ensures the least incremental data by adding purification operations. When the data is transmitted to the synchronization server, it is more conducive to the processing of the synchronization server and reduces the burden on the synchronization server.

4. The time complexity of the incremental capture method adopted by the present invention is O(n), which is simple and efficient, and can greatly shorten the synchronization response time and improve the synchronization efficiency.

5. The present invention uses a trigger to capture the operation log while recording the operation change process of the operation data table. This operation log can be used as the rollback operation of the synchronization system when the synchronization system is unexpectedly terminated or seriously timed out, which affects the final consistency of the data. valid credentials, increasing the robustness of the synchronization system.

Description of drawings

Figure 1 is a diagram of the mobile data synchronization architecture.

Figure 2 is a graph of trigger capture increments.

Figure 3 is a diagram of the cleaning operation process.

Detailed ways

The technical problem to be solved by the invention is to realize the incremental capture of mobile data in the mobile computing environment. Under the mobile distributed storage mechanism, the optimistic replication method is used to achieve the final consistency of multi-mobile terminal business data. However, the weak connectivity of the mobile network and the relatively limited communication delay and bandwidth make the data synchronization mechanism in the mobile computing environment difficult. Achieve efficient data consistency. In order to further reduce the amount of synchronous data communication, the incremental synchronization method is usually used. However, the existing incremental data capture method is oriented to a stable distributed computing environment and is not suitable for a mobile computing environment. Simple transplantation may easily lead to low efficiency and high overhead. The invention proposes a mobile data incremental capture method based on track change, which uses a trigger to capture operation logs, and uses a purification method to merge the operation logs to obtain net incremental data, which is suitable for data synchronization in a mobile environment. The present invention not only makes full use of the existing trigger mechanism in the mobile database and the characteristics of low frequency of writing operations of the mobile terminal, but also compresses the incremental data to the minimum, reduces the synchronous data traffic to the greatest extent, and reduces the demand for communication bandwidth. Improved synchronization efficiency.

Below in conjunction with legend and example the technical scheme of the present invention is described in detail:

The incremental acquisition method of mobile data based on track change proposed by the invention is applicable to the synchronization mechanism of mobile data. The mobile data synchronization architecture is shown in Figure 1. Mobile terminals (such as mobile phones, PADs, tablets, etc.) usually integrate their own mobile databases (such as SQLite, SQL Server CE, etc.) within the system. When offline, The mobile terminal stores the operation data in the local mobile database; when it is online or can access the synchronization server, the mobile terminal will take out the incremental data from the local mobile database, obtain the net incremental data through net incremental processing, and then access the The Access Point (AP) connects to the Internet and transmits the net incremental data to the synchronization server, which is processed by the synchronization server and updated to the main database, thereby realizing the final consistency of data between the mobile terminal and the synchronization server.

The mechanism of the mobile data incremental capture method based on trajectory change proposed by the present invention is shown in Figure 2. First, an operation data table and an operation log table are defined in the mobile service database of the mobile terminal, and then triggers of related operations are defined on the operation data table. At this time, when the user operates the mobile terminal service, the system automatically converts the user operation into Data Manipulation Language (Data Manipulation Language, DML) to operate the operation data table, and then triggers the corresponding trigger. The server records related operation logs into the operation log table. When data synchronization is initiated, the system extracts incremental data from the operation log table, and sends it to the synchronization server after cleaning operation.

As shown in Figure 3, the system extracts the incremental data from the operation log table, and the purification operation mechanism groups the operation log according to the ID of the operated record, purifies each group separately, and then merges each purified group Combined, you can get the net incremental data of the operation log.

In order to facilitate the understanding of the technical solution of the present invention, some concepts are defined below:

Definition 1 Net incremental processing Net incremental processing refers to the equivalent combination of a series of operations acting on the operation data set, compressing the operation steps, so that

The final state of the operation sequence under the action of the compressed operation is consistent with that of the original operation.

Net incremental processing can not only reduce the actual size of incremental data and data transmission time, but also reduce the data loading and maintenance time of the synchronization server, making the entire synchronization process more stable and efficient.

In order to better express the various operation change processes and their result relationships in relational databases, the present invention introduces the concept of relational set, then the relationship of a table can be represented by relational set R, and the records in the table are the elements in the relational set . If there are n records in a table, it is expressed as a relational set R={r ₁ ,r ₂ ,…,r _n },n∈N ^* , where r ₁ ,r ₂ ,…,r _n represent records.

Definition 2. The operation function relationship set X becomes X' after a series of operations. We can describe that X becomes X' after being acted on by the function Ψ(X), that is, Ψ(X)=X', and the function Ψ is X to An operation function of X'. From the definition of the operation function, it can be seen that a series of operations of a certain record in the operation log is an operation function of the record.

Definition 3 Equivalent operation function For relational set A, A' is obtained under the action of operation function f ₁ , that is, f ₁ (A)=A', if there is another operation function f ₂ such that f ₂ (A)=A' , then the operation functions f ₁ and f ₂ are equivalent operation functions with respect to A, denoted as f ₁ (A)≈f ₂ (A).

Definition 4. The optimal equivalent operation function is not unique to the operation function equivalent to f _1. If f ₂ is the least operation step among all the equivalent operation functions of f ₁ , then f ₂ is the optimal equivalent operation function of f ₁ .

The net incremental processing of incremental data is to find its optimal equivalent operation function f ^* according to f ₁ (X)=X' such that:

f ^* (X) = X' (1)

The present invention takes the "synchronized address book" system as an application example, installs it on the Android smart phone, and adopts the SQLite3 mobile database. The incremental capture process of the present invention is:

Step 1: Define three different types of triggers for adding, modifying, and deleting on the operation data table, respectively set the trigger conditions of the triggers, and then define an operation log table to store the captured data change track, In addition to all the fields of the operation data table, the operation log table also includes the primary key and the "operation type" field. Its specific description is as follows:

① Define a contact operation data table contact in the mobile database, including LUID, name, number, email, birth fields, LUID is the primary key;

② Define an operation log table log_contact in the same mobile database as ①, including all fields in contact, the primary key ID and the "operation type" operaType field. The operaType field takes one of three types: A (new), M (modified), and D (deleted);

③ Three triggers are defined on the contact, namely the new trigger Add_trigger, the modification trigger Modify_trigger and the deletion trigger Delete_trigger. These triggers are used to capture the data change track.

Step 2: Operation (addition, modification or deletion) on the operation data table will trigger the corresponding pre-defined trigger, and the trigger will automatically record the operation record, that is, record the change track and operation type, into the operation log table. For example:

①Add two pieces of contact information to the contact table. The new contact information in the contact table is shown in Table 1:

Table 1 New contact information in the contact table

LUID name number email birth abcdwwwefgruila Zhang San 15112345678 zhansanyx.com 2000/11/9 afkltyoskdffgtr Li Si 18112345678 lisinj.com 1999/2/13

②At this time, when the new trigger Add_trigger is triggered, the operation log is written to log_contact, as shown in Table 2:

Table 2 Operation logs recorded by log_contact

ID LUID name number email birth operaType 1 abcdwwwefgruila Zhang San 15112345678 zhansanyx.com 2000/11/9 A 2 afkltyoskdffgtr Li Si 18112345678 lisinj.com 1999/2/13 A

③Modify the contact information. For the first time, modify the number of the contact person Zhang San to "18309876543"; for the second time, modify the number of Zhang San to "15311176543", and modify the email to zhansan0000yx.com; for the third time, change the number of Li Si Contact record deletion. The modification trigger Modify_trigger will be triggered for the first time and the second time respectively; the delete trigger Delete_trigger will be triggered for the third time; then three new log records will be added to the log_contact content, as shown in Table 3:

Table 3 log_contact table information after modifying records

ID LUID name number email birth operaType 1 abcdwwwefgruila Zhang San 15112345678 zhansanyx.com 2000/11/9 A 2 afkltyoskdffgtr Li Si 18112345678 lisinj.com 1999/2/13 A 3 abcdwwwefgruila Zhang San 18309876543 zhansanyx.com 2000/11/9 m 4 abcdwwwefgruila Zhang San 15311176543 zhansan0000yx.com 2000/11/9 m 5 afkltyoskdffgtr Li Si 18112345678 lisinj.com 1999/2/13 D.

Step 3: Group all the records in the operation log table according to the ID numbers of the operated records.

A sequence of operations on a dataset is essentially an operation on individual records in the dataset. The operation log is represented by a relational set L, and the operation set of the i-th record in the operation data set is represented by L _i (1≤i≤n, i∈N), so the relational set of the operation log table can be expressed as L=(L ₁ ∪L ₂ ∪…∪L _n ). Thus, the purification process of the relationship set L can be converted into the purification process of the operation set L _i of each record, namely:

f ^* (L) = f ^* (L ₁ ∪L ₂ ∪…∪L _n )

(2)

＝f ^* ₁ (L ₁ )∪f ^* ₂ (L ₂ )∪…∪f ^* _n (L _n )

When extracting increments, first take out all data change tracks from the operation log table. These change tracks themselves are a series of records and are composed of the operation sequence of each operated record in the operation data table, so the operation log table All the records in are grouped according to the ID number of the operated record. According to the formula (2), the purification operation of the operation log table is the purification operation of each group.

The log records in the log_contact table are divided into two groups according to the keyword LUID, as shown in Table 4:

Table 4 Grouping records recorded in log_contact

Step 4: Apply the purification algorithm to compress and merge these groups, that is, purify, to obtain the purification results of each operated record.

In a relational database, any operation sequence on the same record can be decomposed into the following eight basic operation types or composite operation types, as shown in Table 5:

Table 5 Operation Type

serial number basic operation type significance serial number Composite operation type significance 1 A Add 5 A-M Modified after adding 2 m Revise 6 A-D Delete after adding 3 D. delete 7 M-M edit after edit 4 N no action 8 M-D delete after modification

For example, if a record undergoes a series of operations "Add-Modify-Modify-Delete", it can be composed of the basic operation type "1-2-2-3" in the table; it can also be composed of the composite operation type "5-7-8 "combined to.

It can be seen from this that a series of operations of each record can be split into basic operation types and composite operation types, and finally combined operations. The operation rules are shown in Table 6:

Table 6 Operation rules for merging operations on the same record

"/" in Table 6 means that there is no such merge operation, and NULL means that the merge result of the operation is empty. The operation rules can be described as follows:

①OP _A ∪OP _M ＝OP _A , ②OP _A ∪OP _D ＝NULL, ③OP _M ∪OP _M ＝OP _M ,

④ OP _M ∪ OP _D ＝ OP _D , ⑤ OP _J ∪ OP _N ＝ OP _N ∪ OP _J ＝ OP _J , J ∈ {A, M, D, N}.

OP represents an operation on the record, and the "∪" operator represents the operation union of the two operands before and after it.

For the operation set L _i (1≤i≤n,i∈N) of the i-th record in the operation log relational set L, suppose L _i ={l ₁ ,l ₂ ,…l _k …l _n } (1≤k≤n, k∈N), OP _Γ (l _k ), Γ∈{A,M,D} means that the kth operation type on the i-th record is Γ, and the result is l _k . Then the set of recorded operation sequences can be divided into five types, and the following five purification formulas are used respectively:

f ^* _i (L _i )＝OP _A (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _M (l _n )＝OP _A (l _n ) (3)

f ^* _i (L _i )＝OP _A (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _D (l _n )＝NULL (4)

f ^* _i (L _i )＝OP _M (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _M (l _n )＝OP _M (l _n ) (5)

f ^* _i (L _i )＝OP _M (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _D (l _n )＝OP _D (l _n ) (6)

f ^* _i (L _i ) = OP _D (l ₁ ) = OP _D (l ₁ ) (7)

According to the type of the recorded operation sequence, the purification result of the recorded operation sequence can be obtained by using the corresponding purification formula from (3)-(7).

According to the grouping situation in Table 4, choose (3) to purify group 1, choose (4) to purify group 2, and the purification results are as follows:

Table 7 Purification results of Group 1 and Group 2

ID LUID name number email birth operaType group 4 abcdwwwefgruila Zhang San 15311176543 zhansan0000yx.com 2000/11/9 A one

Step 5: The purification result of the entire operation log table can be obtained by merging the purification results of each operated record, that is, the purification operation of the operation log table is completed.

Therefore, the optimal equivalent operation function for the operation log relationship set is:

${\mathrm{<span\; class="notranslate">\; f</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{<span\; class="notranslate">\; \&cup;</span>}}{{\mathrm{<span\; class="notranslate">\; f</span>}}^{<span\; class="notranslate">\; *</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>$

After the operation log is obtained, the operation log can be purified by formula (8), redundant records are removed, and the net incremental synchronization data is obtained, so as to reduce the amount of synchronization data that needs to be transmitted during the synchronization process, shorten the synchronization response time, and improve the synchronization efficiency.

The net incremental synchronization data can be obtained by combining the purification results of group 1 and group 2. It can be seen from Table 7 that group 2 becomes empty after the purification operation, that is, all operation log records of group 2 are deleted. Therefore, the net increment of the operation log in the log_contact table of the "synchronized address book" system is the same as that shown in Table 7.

Claims (4)

1. A mobile data incremental capture method based on trajectory changes, which uses triggers provided by relational databases to capture operation logs, records the operation change process of the operation data table, and uses a purification method to merge operation logs to achieve net incremental capture and A record of the entire operational change process; the steps are as follows: Step 1: Define three triggers for adding, modifying, and deleting on the operation data table, respectively set the trigger conditions of the triggers, and then define an operation log table to store the captured data change track, the operation log table In addition to having all the fields of the operation data table, it also includes the primary key and the "operation type" field; Step 2: To operate (add, modify or delete) the operation data table will trigger the corresponding pre-defined trigger, and the trigger will automatically record the operation record and operation type into the operation log table; Step 3: group all the records in the operation log table according to the ID numbers of the operated records; Step 4: apply the purification algorithm to compress and merge these groups, that is, purify, and obtain the purification results of each operated record; Step 5: Combine the purification results of each operated record to obtain the purification result of the entire operation log table, that is, complete the operation of purifying the operation log table. 2. a kind of mobile data incremental capture method based on track change according to claim 1, the situation of trigger in the described step 1 is: there are three kinds of operation types in the relational database, are respectively adding, revising and deleting According to these three types of operations, define the triggers of these three types of operations on the operation data table, such as Add_trigger, Modify_trigger and Delete_trigger, and set their trigger conditions respectively, such as when adding a new record on the operation data table. The new trigger Add_trigger will be triggered, and Add_trigger will record the newly added record data and the newly added type into the operation log table for purification. In the case of the operation log table in step 1, the operation log table must contain the operation All fields in the data table, its own primary key and the "operation type" field, so that the complete operation process and the type of this operation can be recorded. 3. A kind of mobile data incremental capture method based on trajectory change according to claim 1, the operation record grouping situation in the said step 3 is: because the operation log table has recorded the ID number of the operated record, so the operation log A series of records in the table are divided into several groups according to the ID numbers of the records being operated. 4. A method for capturing incremental mobile data based on trajectory changes according to claim 1, the purification and merging operations in steps 4 and 5 are as follows: in a relational database, any sequence of operations on the same record is It can be decomposed into 8 basic operation types and composite operation types, as shown in the following table: serial number basic operation type significance serial number Composite operation type significance 1 A Add 5 A-M Modified after adding 2 m Revise 6 A-D Delete after adding 3 D. delete 7 M-M edit after edit 4 N no action 8 M-D delete after modification Split the series of operations represented by each group into basic operation types and compound operation types, and then use the corresponding purification formulas from (1)-(5) to perform purification according to the type of operation sequence in each group. Finally, formula (6) is used to combine the purification results of each group, that is, the purification operation of the operation log table is completed. f ^* _i (L _i )＝OP _A (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _M (l _n )＝OP _A (l _n ) (1) f ^* _i (L _i )＝OP _A (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _D (l _n )＝NULL (2) f ^* _i (L _i )＝OP _M (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _M (l _n )＝OP _M (l _n ) (3) f ^* _i (L _i )＝OP _M (l ₁ )∪OP _M (l ₂ )∪…∪OP _M (l _k )∪…∪OP _D (l _n )＝OP _D (l _n ) (4) f ^* _i (L _i ) = OP _D (l ₁ ) = OP _D (l ₁ ) (5) ${\mathrm{<span\; class="notranslate">\; f</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&cup;</span>}}}{{\mathrm{<span\; class="notranslate">\; f</span>}}^{<span\; class="notranslate">\; *</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$ The meanings of the symbols in formulas (1)-(6) are: A, M, and D, as shown in Table 1, respectively represent the operation types of adding, modifying, and deleting; OP represents an operation on a record; the operation merge operator "∪" means to find the operation union of the two operands before and after it, where NULL means the operation union is empty; we regard the operation log table and the operation data table as the operation log set and the operation data set respectively, then for the operation log set The operation set L _i (1≤i≤n, i∈N) of the i-th record in the operation data set in L, assuming L _i ={l ₁ ,l ₂ ,…l _k …l _n }(1≤k≤ n, k∈N), OP _Γ (l _k ), Γ∈{A,M,D} means that the k-th operation type on the i-th record is Γ, and the result is l _k ; f ^* _i (L _i ) Represents the optimal equivalent operation function of L _i .