A method, system, terminal and medium for detecting machine-generated Chinese text
By segmenting Chinese text and decoding it using a T5 model mask, and calculating the negative log-likelihood function value and confidence score, the problem of low accuracy in detecting machine-generated text in existing technologies is solved, achieving fast and accurate detection of machine-generated text.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHONGQING JUEXIAO TECH CO LTD
- Filing Date
- 2023-03-28
- Publication Date
- 2026-05-29
AI Technical Summary
Existing methods for detecting machine-generated Chinese text have low accuracy and their effectiveness decreases as the realism of text generated by large language models increases, making it difficult to effectively distinguish between machine-generated and human-written text.
By segmenting the Chinese text to be detected into multiple paragraphs, using the T5 model for mask decoding, calculating the negative log-likelihood function value and confidence score, and combining this with a set threshold, it is determined whether the text was generated by a machine.
It enables rapid and accurate detection of whether Chinese text is machine-generated, avoiding the complexity of building sample training models and improving detection efficiency and accuracy.
Smart Images
Figure CN116468022B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer software technology, and specifically to a method, system, terminal, and medium for detecting machine-generated Chinese text. Background Technology
[0002] With the rapid development of natural language processing technology, large language models are increasingly used in people's lives and work. For example, people use ChatGPT to generate advertising copy, students use ChatGPT to generate essays, and journalists use it to generate news articles. Text generation based on large language models is becoming increasingly common. However, this has also caused some problems for society. For example, teachers cannot determine whether students' assignments are machine-generated, and reviewers cannot determine whether submitted papers are machine-generated. Therefore, it is necessary to invent a method to detect whether Chinese text is machine-generated. Currently available detection techniques are as follows:
[0003] 1. Statistical learning methods are used to analyze the differences between machine-generated text and human-written text, such as statistical perplexity and TF-IDF. The differences in feature indicators between machine-generated text and human-written text are analyzed and compared to draw conclusions. However, most of these methods have low accuracy, and it is very difficult to find an effective feature.
[0004] 2. A neural network modeling approach is used, constructing a dataset based on collected machine-generated and human-written text. A binary classification model is then built using neural networks or traditional machine learning classification models, such as support vector machines or decision trees. However, this method gradually becomes less effective as machine-generated text becomes increasingly realistic.
[0005] In summary, neural network modeling requires the construction of samples, and as the text generated by large language models becomes increasingly similar to human-written text, the effectiveness of this method will deteriorate. While statistical analysis does not require the construction of samples for training the model, it does require finding effective factors to distinguish between machine-generated text and human-written text. Summary of the Invention
[0006] To address the shortcomings of existing technologies, embodiments of the present invention provide a method, system, terminal, and medium for detecting machine-generated Chinese text, which can accurately and efficiently detect whether Chinese text is generated by a machine.
[0007] In a first aspect, embodiments of the present invention provide a method for detecting machine-generated Chinese text, comprising:
[0008] The Chinese text to be detected is divided into N text segments according to a set step size;
[0009] Iterate through a list of N text segments, sample each text segment at a set sampling rate, sample M times and mask to obtain M masked text segments, then input the M masked text segments into the T5 model sequentially for decoding to obtain a list of unmasked text segments.
[0010] Calculate the negative log-likelihood score for each text segment, calculate the negative log-likelihood score for each element in the list of unmasked text segments, and calculate the confidence score for each text segment based on the negative log-likelihood score for each text segment and the negative log-likelihood score for each element.
[0011] The confidence score is compared with a set threshold. If it is greater than the threshold, the text paragraph is determined to be machine-generated; if it is less than or equal to the threshold, the text paragraph is determined to be human-written. The average confidence score of N text paragraphs is calculated and compared with the set threshold. If the average confidence score is greater than the threshold, the Chinese text to be tested is determined to be machine-generated; if it is less than or equal to the threshold, the Chinese text to be tested is determined to be human-written.
[0012] Secondly, an embodiment of the present invention provides a system for detecting machine-generated Chinese text, comprising: a text segmentation module, a mask decoding module, a confidence score calculation module, and a discrimination module, wherein the text segmentation module is used to segment the Chinese text to be detected according to a set step size to obtain a list of N text segments;
[0013] The mask decoding module is used to traverse a list of N text segments, sample each text segment at a set sampling rate, sample M times and mask to obtain M masked text segments, and input the M masked text segments into the T5 model in sequence for decoding to obtain a list of unmasked text segments.
[0014] The confidence score calculation module is used to calculate the negative log-likelihood function score of each text segment, calculate the negative log-likelihood function score of each element corresponding to the unmasked text segment list, and calculate the confidence score of each text segment based on the negative log-likelihood function score of each text segment and the negative log-likelihood function score of each element.
[0015] The discrimination module compares the confidence score with a set threshold. If the confidence score is greater than the threshold, the text paragraph is determined to be machine-generated; if it is less than or equal to the threshold, the text paragraph is determined to be human-written. The module calculates the average confidence score of N text paragraphs and compares the average confidence score with the set threshold. If the average confidence score is greater than the threshold, the Chinese text to be tested is determined to be machine-generated; if it is less than or equal to the threshold, the Chinese text to be tested is determined to be human-written.
[0016] Thirdly, an embodiment of the present invention provides a smart terminal, including a processor, an input device, an output device, and a memory. The processor is connected to the input device, the output device, and the memory, respectively. The memory is used to store a computer program, the computer program including program instructions, and the processor is configured to call the program instructions to execute the method described in the above embodiments.
[0017] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, cause the processor to perform the methods described in the above embodiments.
[0018] The beneficial effects of this invention are:
[0019] This invention provides a method for detecting machine-generated Chinese text. By calculating the negative log-likelihood function value of the Chinese text, it can determine whether the Chinese text is generated by a machine. This method does not require building a sample training model to detect whether the Chinese text is generated by a machine. The method is simple to implement and can quickly and accurately detect whether the Chinese text is generated by a machine.
[0020] This invention provides a system, terminal, and medium for detecting machine-generated Chinese text, which can quickly and accurately detect whether Chinese text is generated by a machine. Attached Figure Description
[0021] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.
[0022] Figure 1 A flowchart of a method for detecting machine-generated Chinese text provided in the first embodiment of the present invention is shown;
[0023] Figure 2 This invention provides a schematic diagram of the structure of a system for detecting machine-generated Chinese text, according to another embodiment of the present invention.
[0024] Figure 3 A schematic diagram of the structure of a smart terminal provided in another embodiment of the present invention is shown. Detailed Implementation
[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0026] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.
[0027] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.
[0028] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.
[0029] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrases "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."
[0030] It should be noted that, unless otherwise stated, the technical or scientific terms used in this application should have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.
[0031] like Figure 1 The diagram shows a flowchart of a method for detecting machine-generated Chinese text provided in the first embodiment of the present invention. The method includes the following steps:
[0032] S1: Divide the Chinese text to be detected into a list of N text segments according to the set step size. N is a positive integer.
[0033] Specifically, the Chinese text to be detected, `text`, is divided into N text segments according to a set stride, where `stride` = 25, `stride` is the maximum length of the segment, and N is the number of segments to be divided into. The resulting list of N segments, `text_`, is then generated. list =[text1,text2,text3,…,text n ].
[0034] S2: Iterate through the list of N text segments, sample each text segment at a set sampling rate, sample M times, mask to obtain M masked text segments, and input the M masked text segments into the T5 model sequentially for decoding to obtain a list of unmasked text segments. M is a positive integer.
[0035] Specifically, iterate through the list of N text paragraphs, and for each text paragraph... i If the sampling rate is ratio = 0.2, then the number of sampled text characters is: K = int(ratio * len(text) i (), where int represents integer subtraction. Replace these K characters with extra0, extra1, ..., extraK to form mask_text. i_1 For text paragraphs, the "uer / t5-base-chinese-cluecorpussmall" T5 model from the transformer is needed to transform the mask_text. i_1 When input into the T5 model, the T5 model will output the text that should be filled in at the corresponding extra0, extra1, ..., extraK positions. Replace the mask_text with the output text in sequence. i_1 The extra0, extra1, ..., extraK in the text form a new unmasked text segment. i_1 For text paragraphs i Repeated sampling M times yields M masked text segments. These M masked text segments are then sequentially input into the T5 model for decoding, resulting in a list of unmasked text segments. i_list =[text i_1 text i_2 text i_3 , ..., text i_m In this embodiment, M = 50.
[0036] S3: Calculate the negative log-likelihood score for each text segment, calculate the negative log-likelihood score for each element in the list of unmasked text segments, and calculate the confidence score for each text segment based on the negative log-likelihood scores for each text segment and each element.
[0037] Specifically, the negative log-likelihood score of N text segments is calculated. One text segment i is selected sequentially, and its negative log-likelihood score is calculated. The GPT2 model is used here, specifically "uer / gpt2-chinese-cluecorpussmall". The negative log-likelihood score is:
[0038] score = -loss(GPT2(text) i label = text i )) (1)
[0039] Equation (1) means that the text i After encoding, the label also uses text. i The encoding of text i The label and the input are both fed into the GPT2 model, and the backpropagation error value (loss) is obtained. The negative loss value is taken as the score. At this point, the text is calculated. i The negative log-likelihood function value score is score i Similarly, the list of unmasked text segments can be calculated. i_list The negative log-likelihood function value of each corresponding element (score text) i_score =[score_text i_1 score_text i_2 score_text i_3 ..., score_text i_m ], determine the text paragraph i The formula for confidence scores generated manually or by machine is:
[0040] confidence_score i = (score) i -mean(text i_score )) / std(text i_score (2)
[0041] In equation (2), mean represents the average, std represents the standard deviation, and mean(text) i_scoreThe `std(text)` function calculates the average of the negative log-likelihood function values of each element in a list of unmasked text segments. i_score The value is the standard deviation of the negative log-likelihood function score of each element in the list of unmasked text paragraphs.
[0042] S4: Compare the confidence score with a set threshold. If it is greater than the threshold, the text paragraph is determined to be machine-generated. If it is less than or equal to the threshold, the text paragraph is determined to be human-written. Calculate the average confidence score of N text paragraphs and compare the average confidence score with a set threshold. If it is greater than the threshold, the Chinese text to be tested is determined to be machine-generated. If it is less than or equal to the threshold, the Chinese text to be tested is determined to be human-written.
[0043] Specifically, to determine whether text segment i is machine-generated or manually written, the method is as follows: Based on the method in step S3, the confidence scores of N text segments can be calculated:
[0044] confidence_score=[confidence_score1,confidence_score2,
[0045] confidence_score3,…,confidence_score n (3)
[0046] The calculated confidence score of a text segment is compared with a set threshold to determine whether the text segment was written manually or generated by a machine. In this embodiment, the set threshold is 0.7. If the confidence score of text segment i is less than or equal to 0.7, the segment is recorded as written manually. Conversely, if the confidence score of text segment i is greater than 0.7, the segment is recorded as machine-generated text.
[0047] Simultaneously, the probability that text segment i was written manually can be calculated. The specific method is as follows: calculate a set threshold and subtract the confidence score of text segment i to obtain the difference. Then, calculate the cumulative density function (CDF) of the standard normal distribution based on this difference. The CDF value is then used as the probability of determining that the text was written manually, denoted as prob. i .
[0048] Similarly, to determine whether the Chinese text to be tested is machine-generated or manually written, the specific method is to calculate the average confidence score (score_mean) of N text paragraphs:
[0049] score_mean = mean(confidence_score), where mean represents the average value.
[0050] The average confidence score of N text paragraphs is calculated. This average confidence score is then compared to a set threshold to determine whether the Chinese text being tested is machine-generated or human-written. If the average confidence score is greater than 0.7, the text is classified as machine-generated; otherwise, if it is less than or equal to 0.7, it is considered human-written.
[0051] Simultaneously, the probability that the Chinese text under test was written by a human can be calculated. The specific method is as follows: subtract the set threshold from the average confidence score to obtain the difference; use the difference to calculate the cumulative density function of the standard normal distribution; the cumulative density function of the standard normal distribution is used as the probability of determining that the text paragraph was written by a human. The calculation method is as follows:
[0052] prob=normcdf(abs(0.7-mean_score))*100
[0053] Where normcdf is the cumulative density function of the standard normal distribution, abs is the absolute value function, and 0.7 is the set threshold.
[0054] This invention provides a method for detecting machine-generated Chinese text. By calculating the negative log-likelihood function value of the Chinese text, it can determine whether the Chinese text is generated by a machine. This method does not require building a sample training model to detect whether the Chinese text is generated by a machine. The method is simple to implement and can quickly and accurately detect whether the Chinese text is generated by a machine.
[0055] In the first embodiment described above, a method for detecting machine-generated Chinese text is provided. Correspondingly, this application also provides a system for detecting machine-generated Chinese text. Please refer to... Figure 2 This is a structural block diagram of a system for detecting machine-generated Chinese text provided in the second embodiment of the present invention. Since the device embodiment is basically similar to the method embodiment, it is described simply; relevant details can be found in the description of the method embodiment. The device embodiment described below is merely illustrative.
[0056] like Figure 2The diagram illustrates a system for detecting machine-generated Chinese text according to another embodiment of the present invention. The system includes: a text segmentation module, a mask decoding module, a confidence score calculation module, and a discrimination module. The text segmentation module segments the Chinese text to be detected into a list of N text segments according to a set step size. The mask decoding module iterates through the list of N text segments, samples each text segment at a set sampling rate, performs masking M times to obtain M masked text segments, and sequentially inputs these M masked text segments into a T5 model for decoding to obtain a list of unmasked text segments. The confidence score calculation module calculates the negative logarithm of each text segment. The likelihood function score is calculated by assigning a negative log-likelihood function score to each element in the list of unmasked text segments. Based on the negative log-likelihood function score of each text segment and each element, a confidence score is calculated for each text segment. The discrimination module compares the confidence score with a set threshold. If the confidence score is greater than the threshold, the text segment is determined to be machine-generated; if it is less than or equal to the threshold, the text segment is determined to be human-written. The average confidence score of N text segments is calculated and compared with a set threshold. If the average confidence score is greater than the threshold, the Chinese text to be tested is determined to be machine-generated; if it is less than or equal to the threshold, the Chinese text to be tested is determined to be human-written.
[0057] The discrimination module includes a first probability calculation unit, which subtracts the set threshold from the confidence score to obtain the difference, and uses the difference to calculate the cumulative density function of the standard normal distribution. The cumulative density function of the standard normal distribution is used as the probability of determining that the text paragraph is manually written.
[0058] The discrimination module includes a second probability calculation unit. The second probability calculation unit subtracts the set threshold from the average confidence score to obtain the difference, and uses the difference to calculate the cumulative density function of the standard normal distribution. The cumulative density function of the standard normal distribution is used as the probability of determining that the text paragraph is manually written.
[0059] The confidence calculation module includes a first score calculation unit, which simultaneously inputs the text paragraph and the label into the GPT2 model, calculates the backpropagation error value, and takes the negative value of the error value to obtain the negative log-likelihood function score.
[0060] The present invention provides a system for detecting machine-generated Chinese text, which is based on the same inventive concept and has the same beneficial effects as the method for detecting machine-generated Chinese text provided in the first embodiment, and will not be described again here.
[0061] like Figure 3The diagram shows a schematic of a smart terminal, which includes a processor, an input device, an output device, and a memory. The processor is connected to the input device, the output device, and the memory. The memory stores a computer program, which includes program instructions. The processor is configured to call the program instructions to execute the method described in the above embodiments.
[0062] It should be understood that, in the embodiments of the present invention, the processor may be a Central Processing Unit (CPU), but it may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.
[0063] Input devices may include touchpads, fingerprint sensors (for collecting the user's fingerprint information and fingerprint orientation information), microphones, etc., while output devices may include displays (LCDs, etc.), speakers, etc.
[0064] The memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory may also include non-volatile random access memory. For example, the memory may also store information about the device type.
[0065] In specific implementations, the processor, input device, and output device described in the embodiments of the present invention can execute the implementation methods described in the method embodiments of the present invention, or they can execute the implementation methods described in the system embodiments of the present invention, which will not be repeated here.
[0066] The present invention also provides an embodiment of a computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the method described in the above embodiments.
[0067] The computer-readable storage medium can be an internal storage unit of the terminal described in the foregoing embodiments, such as the terminal's hard drive or memory. The computer-readable storage medium can also be an external storage device of the terminal, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the terminal. Furthermore, the computer-readable storage medium can include both internal storage units and external storage devices of the terminal. The computer-readable storage medium is used to store the computer program and other programs and data required by the terminal. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.
[0068] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0069] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the terminals and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0070] In the several embodiments provided in this application, it should be understood that the disclosed terminals and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices or units, or may be electrical, mechanical or other forms of connection.
[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.
Claims
1. A method for detecting machine-generated Chinese text, characterized in that, include: The Chinese text to be detected is divided into N text segments according to a set step size; Iterate through a list of N text segments, sample each text segment at a set sampling rate, sample M times and mask to obtain M masked text segments, then input the M masked text segments into the T5 model sequentially for decoding to obtain a list of unmasked text segments. Calculate the negative log-likelihood score for each text segment, calculate the negative log-likelihood score for each element in the list of unmasked text segments, and calculate the confidence score for each text segment based on the negative log-likelihood score for each text segment and the negative log-likelihood score for each element. The confidence score is compared with a set threshold. If it is greater than the threshold, the text paragraph is determined to be machine-generated. If it is less than or equal to the threshold, the text paragraph is determined to be human-written. The average confidence score of N text paragraphs is calculated and compared with a set threshold. If the average confidence score is greater than the threshold, the Chinese text to be detected is determined to be machine-generated. If it is less than or equal to the threshold, the Chinese text to be detected is determined to be human-written. The method also includes determining the probability that a text paragraph was written by a human. Specifically, the method includes: subtracting the set threshold from the confidence score to obtain the difference, using the difference to calculate the cumulative density function of the standard normal distribution, and using the cumulative density function of the standard normal distribution as the probability corresponding to determining that the text paragraph was written by a human. The method also includes determining the probability that the Chinese text to be detected is manually written. The specific method includes: subtracting the set threshold from the average confidence score to obtain the difference, using the difference to calculate the cumulative density function of the standard normal distribution, and using the cumulative density function of the standard normal distribution as the probability corresponding to determining that the Chinese text to be detected is manually written. The specific method for calculating the negative log-likelihood function score for each text segment includes: inputting the text segment and label into the GPT2 model simultaneously, calculating the backpropagation error value, and taking the negative value of the error value to obtain the negative log-likelihood function score.
2. A system for detecting machine-generated Chinese text, characterized in that, include: The system includes a text segmentation module, a mask decoding module, a confidence score calculation module, and a discrimination module. The text segmentation module is used to segment the Chinese text to be detected into a list of N text segments according to a set step size. The mask decoding module is used to traverse a list of N text segments, sample each text segment at a set sampling rate, sample M times and mask to obtain M masked text segments, and input the M masked text segments into the T5 model in sequence for decoding to obtain a list of unmasked text segments. The confidence score calculation module is used to calculate the negative log-likelihood function score of each text segment, calculate the negative log-likelihood function score of each element corresponding to the unmasked text segment list, and calculate the confidence score of each text segment based on the negative log-likelihood function score of each text segment and the negative log-likelihood function score of each element. The discrimination module compares the confidence score with a set threshold. If the confidence score is greater than the threshold, the text paragraph is determined to be machine-generated. If the confidence score is less than or equal to the threshold, the text paragraph is determined to be human-written. The module calculates the average confidence score of N text paragraphs and compares the average confidence score with a set threshold. If the average confidence score is greater than the threshold, the Chinese text to be detected is determined to be machine-generated. If the average confidence score is less than or equal to the threshold, the Chinese text to be detected is determined to be human-written. The discrimination module includes a first probability calculation unit, which subtracts the set threshold from the confidence score to obtain a difference, and uses the difference to calculate the cumulative density function of the standard normal distribution. The cumulative density function of the standard normal distribution is used as the probability of determining that the text paragraph is manually written. The discrimination module includes a second probability calculation unit. The second probability calculation unit subtracts the set threshold from the average confidence score to obtain the difference, and uses the difference to calculate the cumulative density function of the standard normal distribution. The cumulative density function of the standard normal distribution is used as the probability of determining that the Chinese text to be detected is manually written. The confidence score calculation module includes a first score calculation unit, which simultaneously inputs the text paragraph and the label into the GPT2 model, calculates the backpropagation error value, and takes the negative value of the error value to obtain the negative log-likelihood function score.
3. A smart terminal, comprising a processor, an input device, an output device, and a memory, wherein the processor is connected to the input device, the output device, and the memory, and the memory is used to store a computer program, the computer program comprising program instructions, characterized in that, The processor is configured to invoke the program instructions to execute the method as described in claim 1.
4. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the method as described in claim 1.