WO2023236114A1

WO2023236114A1 - Industrial test script generation method and apparatus, and storage medium

Info

Publication number: WO2023236114A1
Application number: PCT/CN2022/097686
Authority: WO
Inventors: 刘宁
Original assignee: 西门子股份公司; 西门子（中国）有限公司
Priority date: 2022-06-08
Filing date: 2022-06-08
Publication date: 2023-12-14

Abstract

Provided in the embodiments of the present application are an industrial test script generation method and apparatus, and a storage medium. The method comprises: receiving a natural language test case, which is written in a natural language and is used for testing an industrial application; performing sentence segmentation on the natural language test case, so as to obtain a plurality of natural language commands, and representing each natural language command as a word sequence to be processed; by taking each word sequence to be processed as an input of a pre-trained sequence-to-sequence model, obtaining a target word sequence, which is output by the sequence-to-sequence model and is associated with an industrial test script function; determining corresponding industrial test script functions on the basis of target word sequences; and according to an order consistent with the sorting of the natural language commands, connecting the obtained industrial test script functions corresponding to the natural language commands, so as to obtain an industrial test script corresponding to the natural language test case. By means of the technical solution in the embodiments of the present application, an industrial test script can be automatically generated.

Description

Industrial test script generation method, device and storage medium

Technical field

The present application relates to the field of computers, and in particular to an industrial test script generation method, device and computer-readable storage medium.

Background of the invention

Testing is an important part of ensuring product quality in the industrial product development process. In system testing of industrial equipment, test engineers want to generate test cases that cover all combinations of each input variable of the equipment, verifying that a specific industrial application works as expected. This is a huge project.

To this end, there are currently methods in the field of industrial technology that assist in generating test cases through graphical methods, which can generate test cases by dragging and dropping input variable buttons in the scene graph view. Although this automated testing infrastructure can reduce some coding work, it still requires test engineers to write test scripts, and the quality of test scripts depends largely on the experience of the engineers. Furthermore, from a broader perspective of testing, there are currently some automated test case generation tools for unit testing, such as EvoSuite, Randoop, and Agitar. But obviously, these unit testing methods are not suitable for system level testing of industrial equipment.

Therefore, those skilled in the art are still working on finding other industrial test script generation solutions.

Contents of the invention

In view of this, embodiments of the present application provide, on the one hand, an industrial test script generation method, and on the other hand, an industrial test script generation device and a computer-readable storage medium, which can realize automatic generation of industrial test scripts and improve industrial testing. flexibility and efficiency.

In order to solve the above technical problems, the technical solution of this application is implemented as follows:

An industrial test script generation method, including: receiving a natural language test case written in natural language for testing an industrial application; the natural language test case includes a natural language describing an operation of testing the industrial application in natural language command; perform sentence segmentation on the natural language test case to obtain multiple natural language commands, and represent each natural language command as a sequence of words to be processed; for each sequence of words to be processed, use it as a pre-trained Good sequence-to-sequence model input, and obtain the target word sequence output by the sequence-to-sequence model associated with an industrial test script function; the sequence-to-sequence model takes a word sequence corresponding to a historical natural language command as an input sample , obtained by training with the word sequence of the industrial test script function corresponding to the historical natural language command as the output sample; determining the corresponding industrial test script function based on the target word sequence; using the obtained industrial test script function corresponding to each natural language command Connect in a sequence consistent with the ordering of each natural language command to obtain an industrial test script corresponding to the natural language test case.

An industrial test script generation device, including: a first module for receiving natural language test cases written in natural language for testing an industrial application; the natural language test cases include testing the industrial application described in natural language The natural language command of the application operation; the second module is used to segment the natural language test case into sentence segments, obtain multiple natural language commands, and represent each natural language command as a sequence of words to be processed; the third module A module configured to use each word sequence to be processed as an input to a pre-trained sequence-to-sequence model, and obtain a target word sequence output by the sequence-to-sequence model that is associated with an industrial test script function; The sequence-to-sequence model is trained using the word sequence corresponding to the historical natural language command as an input sample and the word sequence of the industrial test script function corresponding to the historical natural language command as the output sample; the fourth module is used to train based on the The target word sequence determines the corresponding industrial test script function; the fifth module is used to connect the obtained industrial test script functions corresponding to each natural language command in an order consistent with the order of each natural language command to obtain the corresponding natural language command. Industrial test scripts for test cases.

An industrial test script generation device includes at least one memory and at least one processor, wherein: the at least one memory is used to store a computer program; the at least one processor is used to call the computer program stored in the at least one memory, Execute the industrial test script generation method as described in any of the above embodiments.

A computer-readable storage medium has a computer program stored thereon; the computer program can be executed by a processor and implement the industrial test script generation method as described in any of the above embodiments.

It can be seen from the above technical solution that in this application, a sequence-to-sequence model capable of deep learning is obtained by pre-training by using historical natural language commands and their corresponding industrial test script functions. After that, the test cases can be segmented into statements, and we get For each natural language command, the corresponding industrial test script function can then be obtained based on the sequence-to-sequence model reasoning. Then, the obtained industrial test script functions can be connected according to the order of each natural language command in the test case, and the corresponding industrial test script function can be obtained. The industrial test script for test cases realizes a fully automatic workflow from test cases to test scripts, saving the workload of test engineers.

Brief description of the drawings

In order to better understand the present application, the embodiments of the present application will be described in detail below with reference to the accompanying drawings, so that the above and other features and advantages of the present application will be clearer to those of ordinary skill in the art. In the accompanying drawings:

Figure 1 is an exemplary flow chart of an industrial test script generation method in an embodiment of the present application.

Figure 2 is a schematic structural diagram of a sequence-to-sequence model in the embodiment of the present application.

Figure 3 is an exemplary structural diagram of an industrial test script generating device in an embodiment of the present invention.

Figure 4 is an exemplary structural diagram of another industrial test script generation device in an embodiment of the present application.

Among them, the reference signs are as follows:

标号label	含义meaning
101～105101～105	步骤 step
301301	第一单元The first unit
302302	第二单元 Second unit
303303	第三单元Unit 3
304304	第四单元Unit 4
305305	第五单元Unit 5
4141	存储器 memory
4242	处理器 processor
4343	总线bus

How to implement this application

In the embodiment of the present application, in order to further reduce the coding work and improve the efficiency of industrial test script generation, it is considered to provide an automatic generation solution for industrial test scripts.

In order to make the purpose, technical solutions and advantages of the present application more clear, the technical solutions of the present application will be described in detail below with reference to the accompanying drawings and examples.

Figure 1 is an exemplary flow chart of an industrial test script generation method in an embodiment of the present application. As shown in Figure 1, the method may include the following steps:

Step 101: Receive natural language test cases written in natural language for testing an industrial application. The natural language test cases include natural language commands describing operations of the industrial application described in natural language.

For example, an example of a natural language test case for panel alarm LED light inspection includes the following:

Test Case-Panel Alarm LED Check; Test Case-Panel Alarm LED Check

Precondition: precondition

1.Panel connect with all kinds of devices full with full load; panel connects with various devices at full load

Steps:; steps

1.Check the panel is normal, check the panel Alarm LED is off; check whether the panel is normal, check whether the panel alarm LED is off

2.Make an alarm from any type of device,check the panel Alarm LED should be kept on;

3.Remove the alarm event, reset panel; check the panel is normal, check the Alarm LED is off; remove the alarm event, reset the panel; check whether the panel is normal, check whether the alarm LED is off

In the above example, the natural language is in English, and the right side of the semicolon is its Chinese translation. Of course, natural language can also use other languages, such as Chinese. Among them, step 1 includes two natural language commands, namely: 1.1 Check the panel is normal (check whether the panel is normal), 1.2 Check the panel Alarm LED is off (check whether the panel alarm LED light is off). Step 2 includes two natural language commands, namely: 2.1 Make an alarm from any type of device (issue an alarm from any type of device), 2.2 Check the panel Alarm LED should be kept on (check whether the panel alarm LED light should be stay lit). Step 3 includes four natural language commands, namely: 3.1 Remove the alarm event (remove the alarm event), 3.2 Reset panel (reset the panel); 3.3 Check the panel is normal (check whether the panel is normal), 3.4 Check the Alarm LED is off (check whether the alarm LED light is off).

In specific implementation, if the test case is a document D, the document D can be expressed as D={s ₁ , s ₂ ,..., s _m }, where s ₁ , s ₂ ,..., s _m are natural language commands expressed in natural language sentences.

Step 102: Segment the natural language test case into sentence segments to obtain multiple natural language commands, and represent each natural language command as a sequence of words to be processed.

For example, for the above test case, several natural language commands 1.1 to 3.4 mentioned above can be obtained. During specific implementation, in order to facilitate sentence segmentation, specific separators can be added between adjacent natural language commands in the test case, and then the statement can be segmented according to the specific separators in this step; or, in this step, you can also Segment statements directly based on commas and periods in test cases. The specifics can be determined according to the actual situation and are not limited here.

After obtaining each natural language command, each natural language command can be expressed as a sequence of words to be processed, that is, _si = {w ₁ , w ₂ ,..., w _s }, where w ₁ , w ₂ , ...,w _s is each word in the natural language command. Taking the natural language command "Check the panel is normal" as an example, the corresponding word sequence to be processed is {Check,the,panel,is,normal}.

Of course, if the natural language is another language, the "word" here refers to the smallest component unit in the sentence of that language. For example, if the natural language is Chinese, the "word" here refers to each Chinese character.

Step 103: For each word sequence to be processed, use it as the input of a pre-trained sequence-to-sequence model, and obtain the target word sequence output by the sequence-to-sequence model associated with an industrial test script function; so The sequence-to-sequence model is trained by taking the word sequence corresponding to the historical natural language command as an input sample, and using the word sequence of the industrial test script function corresponding to the historical natural language command as the output sample.

Figure 2 shows a schematic structural diagram of a sequence-to-sequence model in an embodiment of the present invention. As shown in Figure 2, the sequence-to-sequence model is a sequence-to-sequence model with a bidirectional long short-term memory network (LSTM, Long Short-Term Memory) encoder and LSTM decoder.

Among them, all words in the word sequence to be processed are input into the sequence-to-sequence model in order. For example, all words in the word sequence to be processed are input into the LSTM encoder of the model at each timestamp, and the LSTM encoder is embedded After processing and multiplying the hidden states by their attention weights, a context-aware representation vector is obtained. Then, the LSTM decoder selects a word w _i at time t. After selecting w _i , the hidden layer state h i at time _i will be fed into the LSTM decoder at time t+1. The sequence-to-sequence model then outputs a target word sequence associated with an industrial test script function.

In Figure 2,

The degree of attention paid to the j-th word of the input word sequence to be processed (hereinafter referred to as the input sequence) at the moment when the second word is generated;

is the value of the hidden layer of the j-th word in the input sequence; J is the length of the input sequence word;

is the value of the hidden layer of the second word in the output sequence;

is the value of the hidden layer of the second word in the output sequence after adding the attention mechanism. where ⊕ is the vector splicing symbol, and W and b are learnable parameters.

When implemented, the sequence-to-sequence model can extract target words from a vocabulary based on inference results. Here, the vocabulary list may include all words obtained by deduplicating all words in the function name (or function name and parameters) of each industrial test script function. If the inference result indicates that the required target word is not in the vocabulary, the target word can be directly copied from the input sequence (ie, the input sequence of words to be processed). That is to say, the words composing the target word sequence may be words extracted from a vocabulary, or may be words copied from the word sequence to be processed.

During specific implementation, the sequence-to-sequence model determines the words output at each moment in the target word sequence based on the following operations:

1) Combine the words in the vocabulary and the words in the word sequence to be processed to obtain an extended vocabulary V _ext , that is, the extended vocabulary V _ext is added to the input sequence on the basis of the words in the original vocabulary. A vocabulary made up of words.

2) Calculate the output probability distribution of each word in an extended vocabulary V _ext at time t according to the following formula (1):

P _Vert (y _t )＝p _gen,t P _g (y _t )+(1-p _gen,t )P _c (y _t )(1)

Among them, p _gen,t is the choice at time t whether to generate a word from the vocabulary or select a word from the input sequence.

Among them, σ is the sigmoid function;

It is the weighted sum of the attention and hidden layer values of all words in the input sequence at the moment when the t-th word is generated;

is the value of the hidden layer of the t-th word in the output sequence; E _y,t-1 is the value of the hidden layer that produced the word at the previous moment (t-1), W _s,z , W _s,h , W _{s, E} , b _s are learnable parameters. The output probability distribution P _g (y _t ) of each word in the vocabulary at time t and the attention distribution P _c (y _t ) of each word in the input sequence at time t are determined according to the following formulas (2) and (3) respectively. :

Among them, y _t is the word that needs to be output generated at time t, P _vocab,t (y _t ) is the probability of generating y _t from the vocabulary at time t,

is the attention paid to the j-th word in the input sequence at time t when the word is generated, ν is the vocabulary, ∪ is the union symbol, unk is the defined unknown character symbol, and χ is the vocabulary of the input sequence.

3) After obtaining the output probability distribution of each word in the extended vocabulary V _ext at time t, the word with the highest probability is used as the output word y t at time _t .

In addition, some industrial test script functions may only have function names, while some industrial test script functions may have both function names and parameters. Correspondingly, the target word sequence output by the sequence-to-sequence model may include: function name words, or include function names. words and parameter words. During specific implementation, for the situation where the target word sequence includes both function name words and parameter words, in order to separate the function name words and parameter words, the target word sequence output by the sequence-to-sequence model may include function name words, parameter words, and The delimiter between the function name word and the argument word. Correspondingly, when training the sequence-to-sequence model, the word sequence of the industrial test script function as the output sample also includes: function name words, parameter words, and words between the function name words and the parameter words. separator.

Step 104: Determine the corresponding industrial test script function based on the target word sequence.

In this embodiment, when the sequence-to-sequence model based on deep learning has been trained with a large number of samples and the inference results have become sufficiently accurate, the target word sequence it outputs will be close enough to the word sequence of the industrial test script function or even the same as the industrial test. The word sequence of the script function is consistent. In this case, the corresponding industrial test script function can be determined directly according to the target word sequence. Or, further, in order to improve the mapping accuracy of the target word sequence to the corresponding industrial test script function, in this step, the target word sequence can be matched with the stored word sequence of each industrial test script function to obtain the corresponding Industrial test script functions.

For example, for each natural language command, after reasoning through the sequence-to-sequence model in step 103, a function name word W _gen ={w _1a ,w _2a ,...,w _pa }, delimiters, and Function parameter word W _{gen_aguments} = target word sequence of {w _1b , w _2b ,..., w _qb }.

Each industrial test script function name funcName _i can be expressed as a name list W _ground,i = {w ₁ , w ₂ ,...}. During specific implementation, the function name list W _gen generated by ROUGE metric* can be used to search for the most similar W _ground,i . For example, the word-level number of single-word overlaps, the number of two-word overlaps, and the longest common sequence between the generated W _gen and W _ground,i can be measured using the weighted sum of ROUGE-1, ROUGE-2, and ROUGE-L, respectively. The number of words in . Specifically, the most similar funcNamet can be found through the following formula (4):

funcName _t ＝argmax _i (ROUGE-L _F1 (W _ground,i ; W _gen )+α*ROUGE-2 _F1 (W _ground,i ; W _gen )+β*ROUGE-1 _F1 (W _ground,i ; W _gen ))(4)

Among them, α and β are preset experience values

For example, for the natural language command "Check the panel is normal", the industrial test script function FC30_AI_Normal check can be obtained, and its parameters can be shown in Table 1 below:

Table 1

Step 105: Connect the obtained industrial test script functions corresponding to each natural language command in an order consistent with the order of each natural language command to obtain an industrial test script corresponding to the natural language test case.

For example, the obtained complete test script T can be expressed as: T={funcName ₁ (W _{gen_aguments,1} ), funcName ₂ (W _{gen_aguments,2} ),..., funcName _q (W _{gen_aguments,q} )}.

In this embodiment, it is assumed that the industrial test script function has been written. In this embodiment, the industrial test script function is selected and the input parameters of the function are filled in.

The industrial test script generating method in the embodiment of the present invention has been described in detail above, and the industrial test script generating device in the embodiment of the present invention will be described in detail below. The industrial test script generation device in the embodiment of the present invention can be used to implement the industrial test script generation method in the embodiment of the present invention. For details not disclosed in the device embodiment of the present invention, please refer to the corresponding description in the method embodiment of the present invention.

Figure 3 is an exemplary structural diagram of an industrial test script generating device in an embodiment of the present invention. As shown in FIG. 3 , the device may include: a first module 301 , a second module 302 , a third module 303 , a fourth module 304 and a fifth module 305 .

Among them, the first module 301 is used to receive a natural language test case written in natural language for testing an industrial application; the natural language test case includes a natural language command described in natural language to test the operation of the industrial application.

The second module 302 is used to segment the natural language test case into sentence segments, obtain multiple natural language commands, and represent each natural language command as a sequence of words to be processed.

The third module 303 is configured to use each to-be-processed word sequence as an input to a pre-trained sequence-to-sequence model, and obtain the target word output by the sequence-to-sequence model associated with an industrial test script function. Sequence; the sequence-to-sequence model is trained by taking the word sequence corresponding to the historical natural language command as an input sample, and using the word sequence of the industrial test script function corresponding to the historical natural language command as the output sample. In one embodiment, the sequence-to-sequence model is a sequence-to-sequence model with a bidirectional long short-term memory network LSTM encoder and LSTM decoder. Wherein, when the industrial test script function has a function name and parameters, the target word sequence output by the sequence-to-sequence model may include: function name words, parameter words, and words between the function name words and the parameter words. separator between. In one embodiment, the target word sequence may include: words extracted from a vocabulary and/or words copied from the word sequence to be processed; wherein, the words in the vocabulary include: words related to each industrial test The word associated with the script function.

The fourth module 304 is used to determine the corresponding industrial test script function based on the target word sequence. During specific implementation, the fourth module 304 may match the target word sequence with the stored word sequence of each industrial test script function to obtain the corresponding industrial test script function.

The fifth module 305 is used to connect the obtained industrial test script functions corresponding to each natural language command in an order consistent with the order of each natural language command, to obtain an industrial test script corresponding to the natural language test case.

In fact, the industrial test script generation device provided by this embodiment of the present application can be implemented in various ways. For example, the industrial test script generation device can be compiled into a plug-in installed in a smart terminal by using an application programming interface that conforms to specific rules, or it can be packaged into an application program for users to download and use.

When compiled as a plug-in, the industrial test script generation device can be implemented in a variety of plug-in forms, such as ocx, dll and cab. The industrial test script generation device provided by this implementation of the present application can also be implemented by using specific technologies, such as Flash plug-in technology, RealPlayer plug-in technology, MMS plug-in technology, MIDI plug-in technology or ActiveX plug-in technology.

The industrial test script generation method provided by this implementation of the present application can be stored in various storage media in an instruction storage mode or an instruction set storage mode. These storage media include but are not limited to: floppy disk, optical disk, DVD, hard disk, flash memory, USB flash memory, CF card, SD card, MMC card, SM card, memory stick and xD card.

In addition, the industrial test script generation method provided by this embodiment of the present application can also be applied to flash memory (Nand-flash)-based storage media, such as USB flash drives, CF cards, SD cards, SDHC cards, MMC cards, SM cards , memory stick and xD card.

It should be clear that the operating system operating in the computer can not only implement the program code read by the computer from the storage medium, but also implement part or all of the actual operations by using instructions based on the program code to implement the above embodiments. function of any embodiment.

For example, FIG. 4 is an exemplary structural diagram of another industrial test script generating device in an embodiment of the present application. The device can be used to perform the method shown in Figure 1, or to implement the device in Figure 3. As shown in FIG. 4 , the device may include at least one memory 41 and at least one processor 42 . In addition, some other components can be included, such as communication ports, input/output controllers, network communication interfaces, etc. These components communicate via bus 43 and so on.

At least one memory 41 is used to store computer programs. In one example, the computer program can be understood as including various modules of the device shown in FIG. 3 . In addition, at least one memory 41 may store an operating system and the like. Operating systems include but are not limited to: Android operating system, Symbian operating system, windows operating system, Linux operating system, etc.

At least one processor 42 is used to call a computer program stored in at least one memory 41 to execute the industrial test script generation method described in the example of this application. The processor 42 may be a CPU, a processing unit/module, an ASIC, a logic module or a programmable gate array, etc., and it may receive and send data through a communication port.

The input/output controller has a display and an input device for inputting, outputting and displaying relevant data as a human-computer interaction module.

It will be understood that as used herein, "and/or" is intended to include any and all possible combinations of one or more of the associated listed items.

It can be seen from the above technical solution that in this application, a sequence-to-sequence model capable of deep learning is obtained by pre-training by using historical natural language commands and their corresponding industrial test script functions. After that, the test cases can be segmented into statements, and we get For each natural language command, the corresponding industrial test script function can then be obtained based on the sequence-to-sequence model reasoning. Then, the obtained industrial test script functions can be connected according to the order of each natural language command in the test case, and the corresponding industrial test script function can be obtained. Industrial test scripts for test cases realize a fully automated workflow from test cases to test scripts, saving the workload of test engineers.

In addition, the technical solution based on deep learning in the embodiment of the present invention has higher accuracy than the rule-based method. In addition, the technical solution in the embodiment of the present invention can save the work of rule summary.

The above are only preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the present application. within the scope of protection.

Claims

An industrial test script generation method is characterized by including:

Receive natural language test cases written in natural language for testing an industrial application; the natural language test cases include natural language commands describing the operation of the industrial application in natural language;

Perform sentence segmentation on the natural language test case to obtain multiple natural language commands, and represent each natural language command as a sequence of words to be processed;

For each word sequence to be processed, use it as the input of a pre-trained sequence-to-sequence model, and obtain the target word sequence output by the sequence-to-sequence model associated with an industrial test script function; the sequence to The sequence model uses the word sequence corresponding to the historical natural language command as the input sample, and is trained with the word sequence of the industrial test script function corresponding to the historical natural language command as the output sample; the corresponding industrial test script function is determined based on the target word sequence. ;

The obtained industrial test script functions corresponding to each natural language command are connected in a sequence consistent with the sorting of each natural language command to obtain an industrial test script corresponding to the natural language test case.
The method for generating industrial test scripts according to claim 1, wherein determining the corresponding industrial test script function based on the target word sequence includes: combining the target word sequence with the stored words of each industrial test script function. Match the sequence and obtain the corresponding industrial test script function.
The method for generating industrial test scripts according to claim 1, wherein the sequence-to-sequence model is a sequence-to-sequence model with a bidirectional long short-term memory network LSTM encoder and an LSTM decoder.
The industrial test script generation method according to any one of claims 1 to 3, characterized in that the industrial test script function has a function name and parameters; the target word sequence output by the sequence-to-sequence model includes: function name words , parameter words, and separators between the function name words and the parameter words.
The industrial test script generation method according to any one of claims 1 to 3, characterized in that the target word sequence includes: words extracted from a vocabulary and/or words copied from the word sequence to be processed ; Wherein, the words in the vocabulary include: words associated with each industrial test script function.
The industrial test script generation method according to claim 5, wherein the sequence-to-sequence model determines the words output at each moment in the target word sequence based on the following operations:

Combining the words in the vocabulary and the words in the word sequence to be processed to obtain an extended vocabulary V ext ;

The output probability distribution of each word in the extended vocabulary V ext at time t is calculated according to the following formula:

P Vert (y t )=p gen,t P g (y t )+(1-p gen,t )P c (y t )

In the above formula, p gen,t is the choice at time t whether to generate a word from the vocabulary or select a word from the input sequence.
Among them, σ is the sigmoid function;
It is the weighted sum of the attention and hidden layer values of all words in the input sequence at the moment when the t-th word is generated;
is the value of the hidden layer of the t-th word in the output sequence; E y,t-1 is the value of the hidden layer that produced the word at the previous moment (t-1), W s,z , W s,h , W s, E , b s are learnable parameters. The output probability distribution P g (y t ) of each word in the vocabulary at time t and the attention distribution P c (y t ) of each word in the input sequence at time t are determined according to the following formulas (2) and (3) respectively. :

Among them, y t is the word that needs to be output generated at time t, P vocab,t (y t ) is the probability of generating y t from the vocabulary at time t,
is the attention paid to the j-th word in the input sequence at time t when the word is generated, ν is the vocabulary, ∪ is the union symbol, unk is the defined unknown character symbol, and χ is the vocabulary of the input sequence.

After obtaining the output probability distribution of each word in the extended vocabulary V ext at time t, the word with the highest probability is used as the output word y t at time t.
An industrial test script generation device, characterized by including:

The first module is used to receive natural language test cases written in natural language for testing an industrial application; the natural language test cases include natural language commands describing the operation of the industrial application in natural language;

The second module is used to segment the natural language test cases into sentence segments, obtain multiple natural language commands, and represent each natural language command as a sequence of words to be processed;

The third module is used to treat each word sequence to be processed as an input to a pre-trained sequence-to-sequence model, and obtain the target word output by the sequence-to-sequence model associated with an industrial test script function. Sequence; the sequence-to-sequence model is trained by using the word sequence corresponding to the historical natural language command as an input sample, and using the word sequence of the industrial test script function corresponding to the historical natural language command as the output sample;

The fourth module is used to determine the corresponding industrial test script function based on the target word sequence;

The fifth module is used to connect the obtained industrial test script functions corresponding to each natural language command in an order consistent with the order of each natural language command, to obtain an industrial test script corresponding to the natural language test case.
The industrial test script generation device according to claim 7, wherein the target word sequence includes: words extracted from a vocabulary and/or words copied from the word sequence to be processed; wherein, the vocabulary The words in the table include: words associated with each industrial test script function;

The sequence-to-sequence model determines the words output at each moment in the target word sequence based on the following operations:

Combining the words in the vocabulary and the words in the word sequence to be processed to obtain an extended vocabulary V ext ;

The output probability distribution of each word in the extended vocabulary V ext at time t is calculated according to the following formula:

P Vert (y t )=p gen,t P g (y t )+(1-p gen,t )P c (y t )

In the above formula, p gen,t is the choice at time t whether to generate a word from the vocabulary or select a word from the input sequence.
Among them, σ is the sigmoid function;
It is the weighted sum of the attention and hidden layer values of all words in the input sequence at the moment when the t-th word is generated;
is the value of the hidden layer of the t-th word in the output sequence; E y,t-1 is the value of the hidden layer that produced the word at the previous moment (t-1), W s,z , W s,h , W s, E , b s are learnable parameters. The output probability distribution P g (y t ) of each word in the vocabulary at time t and the attention distribution P c (y t ) of each word in the input sequence at time t are determined according to the following formulas (2) and (3) respectively. :

Among them, y t is the word that needs to be output generated at time t, P vocab,t (y t ) is the probability of generating y t from the vocabulary at time t,
is the attention paid to the j-th word in the input sequence at time t when the word is generated, ν is the vocabulary, ∪ is the union symbol, unk is the defined unknown character symbol, and χ is the vocabulary of the input sequence.

After obtaining the output probability distribution of each word in the extended vocabulary V ext at time t, the word with the highest probability is used as the output word y t at time t.
An industrial test script generation device is characterized by including at least one memory and at least one processor, wherein:

The at least one memory is used to store a computer program;

The at least one processor is configured to call a computer program stored in the at least one memory to execute the industrial test script generation method according to any one of claims 1 to 6.
A computer-readable storage medium on which a computer program is stored; characterized in that the computer program can be executed by a processor and implement the industrial test script generation method as claimed in any one of claims 1 to 6.