Target measure determination method, electronic equipment and storage medium

Abstract

The invention provides a target measure determination method, electronic equipment and a storage medium, and relates to the technical field of text processing, and the method comprises the following steps: carrying out keyword extraction on a target paragraph of a target text, obtaining a target keyword list, matching a target keyword with a preset word in a preset word bank, and if the matching succeeds, determining a target measure; obtaining an importance degree value of the target keyword in the target paragraph, obtaining a negative correlation word corresponding to the target keyword, if the negative correlation word appears in a statement where the target keyword is located, obtaining a negative correlation probability value of the target keyword, obtaining a target length of the target paragraph, and comparing the target length with a preset length threshold value; and obtaining the target length score, and determining the target paragraph as the target measure based on the importance degree value of the target keyword in the target paragraph, the negative correlation probability value of the target keyword and the target length score, thereby improving the accuracy of the target measure in the paragraph.

Description

Technical Field

[0001] The present invention relates to the technical field of text processing, and particularly to a method for determining target measures, an electronic device, and a storage medium. Background Art

[0002] To help policymakers quickly locate the key content of policy implementation, understand the key points and difficulties in actual implementation, evaluate the operability and effectiveness of policies, etc., it is often necessary to identify the target policy, identify the target policy from numerous paragraphs, and in the document regarding the target policy, identify the specific implementation paragraphs and highlight the specific implementation paragraphs.

[0003] In the prior art, pre-trained language models such as BERT are usually used to identify paragraphs. The pre-trained model is fine-tuned on a dataset in the field of policy documents, so as to train a model capable of classifying the content related to implementation strategies in policy texts. However, the accuracy of using only the pre-trained model for identification is not high, and misjudgment is likely to occur. Manual analysis is often required again, and the steps are rather cumbersome. Summary of the Invention

[0004] In view of the above technical problems, the technical solution adopted by the present invention is as follows:

[0005] According to a first aspect of the present invention, there is provided a method for determining target measures, which is used to determine that a paragraph of target text is a target measure. The method includes the following steps:

[0006] S110, extract keywords from the target paragraph of the target text to obtain a list of target keywords, where the list of target keywords includes several target keywords;

[0007] S120, match the target keywords with the obtained preset word library, where the preset word library includes several preset words related to target measures;

[0008] S130, if the match is successful, obtain the importance value of the target keyword in the target paragraph, and obtain the negative correlation words corresponding to the target keyword. The negative correlation words are words whose simultaneous occurrence probability in a sentence corresponding to a preset measure is less than a preset correlation threshold;

[0009] S140, if the negative correlation words appear in the sentence where the target keyword is located, obtain the negative correlation probability value of the target keyword, where the negative correlation probability value is less than 0;

[0010] S150, obtain the target length of the target paragraph, compare the target length with a preset length threshold, and obtain the target length score. The smaller the absolute value of the difference between the target length and the preset length threshold, the greater the target length score;

[0011] S160. Determine that the target paragraph is a target measure based on the importance value of the target keyword in the target paragraph, the negative correlation probability value of the target keyword, and the target length score.

[0012] According to a second aspect of the present invention, there is provided a non-transitory computer-readable storage medium storing a computer program, which is loaded and executed by a processor to implement the foregoing method.

[0013] According to a third aspect of the present invention, there is provided an electronic device, including: a processor, a memory, and a computer program stored on the memory and executable on the processor, wherein the processor implements the foregoing method when executing the computer program.

[0014] The present invention has at least the following beneficial effects: In summary, extract the target keywords from the target paragraph of the target text to obtain a list of target keywords, match the target keywords with the preset words in the preset word library. If the match is successful, obtain the importance value of the target keyword in the target paragraph, and obtain the negative correlation words corresponding to the target keyword. If the negative correlation words appear in the sentence where the target keyword is located, obtain the negative correlation probability value of the target keyword, obtain the target length of the target paragraph, compare the target length with the preset length threshold to obtain the target length score, and determine that the target paragraph is a target measure based on the importance value of the target keyword in the target paragraph, the negative correlation probability value of the target keyword, and the target length score. The present invention determines that the target paragraph is a target measure based on three aspects, reducing manual participation and improving the accuracy of the target measure in the paragraph. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0016] Figure 1 It is a flowchart of a method for determining a target measure provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0018] It should be noted that the terms "first", "second", etc. in the specification, claims and the above-mentioned drawings of the present invention are used to distinguish similar tasks, and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0019] Embodiment 1

[0020] Embodiment 1 of the present invention provides a method for determining a target measure, as Figure 1 shown, the method is used to determine that a paragraph of the target text is a target measure, and the method includes the following steps:

[0021] S110, extract keywords from the target paragraph of the target text to obtain a list of target keywords, and the list of target keywords includes several target keywords.

[0022] Specifically, any method for obtaining the target paragraph of the target text in the prior art belongs to the protection scope of the present invention.

[0023] The preferred way of the present invention is: S100, identify the paragraph titles of the obtained target text, and divide the target text into several target paragraphs based on the identified paragraph titles. Among them, those skilled in the art know that any method for identifying paragraph titles of target text in the prior art belongs to the protection scope of the present invention, and will not be elaborated here.

[0024] S120, match the target keywords with the preset words in the preset word library, where the preset word library includes several preset words related to the target measure. Specifically, those skilled in the art know that any method for matching target keywords with preset words in the prior art belongs to the protection scope of the present invention, such as obtaining the similarity between the target keywords and the preset words.

[0025] S130, if the match is successful, obtain the importance value of the target keyword in the target paragraph, and obtain the negative correlation words corresponding to the target keyword, where the negative correlation words are words whose co-occurrence probability in the sentences corresponding to a preset measure is less than a preset correlation threshold.

[0026] Specifically, successful matching can be understood as that the matching degree between the target keyword and the preset word is greater than the preset matching threshold, that is, the similarity between the target keyword and a certain preset word is greater than the preset matching threshold.

[0027] Specifically, the TF-IDF value of the target keyword in the target paragraph is used as the importance value of the target keyword in the target paragraph.

[0028] In an embodiment of the present invention, a preset negative correlation word pair list is obtained. The preset negative correlation word pair list includes several preset negative correlation word pairs. Each preset negative correlation word pair includes two preset negative correlation words whose occurrence probabilities in the sentences corresponding to a preset measure are less than the preset correlation threshold. Specifically, by matching the target keyword with the preset negative correlation words, the preset negative correlation words corresponding to the successfully matched target keyword are obtained, and the other preset negative correlation word in the same preset negative correlation word pair as the preset negative correlation word corresponding to the target keyword is used as the negative correlation word corresponding to the target keyword.

[0029] S140, if the negative correlation word appears in the sentence where the target keyword is located, obtain the negative correlation probability value of the target keyword, where the negative correlation probability value is less than 0.

[0030] Specifically, the negative correlation probability value of the target keyword is the negative of the importance value of the negative correlation word in the sentence where the target keyword is located. In an embodiment of the present invention, the negative correlation probability value of the target keyword is the negative of the TF-IDF value of the negative correlation word in the sentence where the target keyword is located.

[0031] S150, obtain the target length of the target paragraph, compare the target length with the preset length threshold, and obtain the target length score. The smaller the absolute value of the difference between the target length and the preset length threshold, the greater the target length score. It can be understood that when the target length of the target paragraph is longer, especially when it exceeds the preset length threshold by a large amount, it is considered that the possibility of the target paragraph being the target measure is low. For example, the length of a single item in the issued incentive policies or guiding policies will not be too long.

[0032] S160, determine that the target paragraph is the target measure based on the importance value of the target keyword in the target paragraph, the negative correlation probability value of the target keyword, and the target length score.

[0033] In summary, extract the target keywords from the target paragraph of the target text to obtain a list of target keywords, match the target keywords with the preset words in the preset word library. If the match is successful, obtain the importance value of the target keyword in the target paragraph, and obtain the negatively correlated words corresponding to the target keyword. If the negatively correlated words appear in the sentence where the target keyword is located, obtain the negative correlation probability value of the target keyword. Obtain the target length of the target paragraph, compare the target length with the preset length threshold to obtain the target length score. Based on the importance value of the target keyword in the target paragraph, the negative correlation probability value of the target keyword, and the target length score, determine that the target paragraph is a target measure. The present invention determines that the target paragraph is a target measure based on three aspects, reducing manual participation and improving the accuracy of the target measure in the paragraph.

[0034] Specifically, S160 further includes:

[0035] S161, sum up the importance values of all successfully matched target keywords to obtain the sum of importance values.

[0036] S162, sum up the negative correlation probability values of all target keywords to obtain the sum of negative correlation probabilities.

[0037] S163, sum up the sum of importance values, the sum of negative correlation probabilities, and the target length score to obtain the final score.

[0038] S164, if the final score is greater than the preset measure threshold, determine that the target paragraph is a target measure.

[0039] In summary, sum up the importance values of all successfully matched target keywords to obtain the sum of importance values, sum up the negative correlation probability values of all target keywords to obtain the sum of negative correlation probabilities, sum up the sum of importance values, the sum of negative correlation probabilities, and the target length score to obtain the final score. If the final score is greater than the preset measure threshold, determine that the target paragraph is a target measure. By integrating the three aspects of the importance value, the negative correlation probability value, and the target length score, the final score can be obtained more accurately.

[0040] Further, before S100, it further includes: obtaining an initial file, and using a corresponding preset processing method to process the initial file based on the type of the initial file to obtain the target text.

[0041] Furthermore, when the type of the initial file is in picture format, the preset processing method corresponding to the type of the initial file is OCR technology. For example, when the type of the initial file is in audio format, use an audio recognition method to process the initial file to obtain the target file.

[0042] In an embodiment of the present invention, S150 further includes obtaining the target length score through the following steps:

[0043] Obtain the absolute value x of the difference between the target length and the preset length threshold. That is, x = |a - b|, where a is the target length and b is the preset length threshold.

[0044] Obtain the preset difference range and the preset score corresponding to the preset difference range. Among them, any two preset difference ranges do not overlap. If the maximum value in a preset difference range A1 is less than the minimum value in a preset difference range A2, then the preset score corresponding to A1 is greater than the preset score corresponding to A2.

[0045] Obtain the preset score corresponding to the preset difference range where the absolute value x is located as the target length score. That is, obtain the preset difference range where the absolute value x is located as the specified difference range, and use the preset score corresponding to the specified difference range as the target length score.

[0046] In summary, obtain the absolute value x of the difference between the target length and the preset length threshold, obtain the preset difference range and the preset score corresponding to the preset difference range, and obtain the preset score corresponding to the preset difference range where the absolute value x is located as the target length score. It can be understood that the smaller the absolute value of the difference between the target length and the preset length threshold, the greater the target length score, and the preset length threshold can be determined by the length of the target measure in the historical data.

[0047] In another embodiment of the present invention, S150 further includes obtaining the target length score through the following steps:

[0048] Obtain the absolute value x of the difference between the target length and the preset length threshold.

[0049] Obtain the target length score, where the target length score is the ratio of the preset length threshold to the absolute value x. That is, the target length score c = b / x, where b is the preset length threshold.

[0050] In summary, obtain the absolute value of the difference between the target length and the preset length threshold, and obtain the target length score, where the target length score is the ratio of the absolute value x to the target length. The smaller the absolute value x, the greater the target length score.

[0051] Further, after S160, it further includes:

[0052] S170, highlight the target measure in the target text, and divide the target paragraphs that are not the target measure into several target sentences.

[0053] Specifically, the target measure is highlighted in the target text to make it more prominent and easier to observe with the naked eye. Those skilled in the art know that any method of dividing a paragraph into sentences in the prior art falls within the protection scope of the present invention. For example, using the punctuation mark "." to divide the target paragraph that is not the target measure into target sentences.

[0054] S180, obtain the target sentence containing the target keyword and mark it as the intermediate sentence, and vectorize the intermediate sentence to obtain the intermediate vector.

[0055] Specifically, the method of obtaining the target sentence containing the target keyword belongs to any one of the prior arts. For example, by means of text matching.

[0056] S190, obtain the target similarity between the target vector and the intermediate vector.

[0057] S200, if the target similarity is greater than the preset similarity threshold, highlight the intermediate sentence corresponding to the intermediate vector. The target vector is the vector obtained by vectorizing the target measure. It can be understood that when the target similarity is greater than the preset similarity threshold, it is considered that the intermediate sentence corresponding to the intermediate vector is also the target measure, but the intermediate sentence corresponding to the intermediate vector is not successfully recognized due to reasons such as the paragraph where it is located or other reasons. Therefore, the method of S170 - S200 is used to highlight the intermediate sentence.

[0058] In summary, the target measure is highlighted in the target text, the target paragraph that is not the target measure is divided into several target sentences, the target sentence containing the target keyword is obtained and marked as the intermediate sentence, the intermediate sentence is vectorized to obtain the intermediate vector, the target similarity between the target vector and the intermediate vector is obtained, and if the target similarity is greater than the preset similarity threshold, the intermediate sentence corresponding to the intermediate vector is highlighted. The present invention further screens the target measures that are not paragraphs through the method of target keywords and similarity.

[0059] Furthermore, the present invention further includes: the preset word library also stores the initial tags corresponding to the preset words, obtain the preset tags corresponding to the target keywords based on the initial tags, and determine the target tags corresponding to the target measures based on the preset tags corresponding to the target keywords; for example, among the preset tags corresponding to the target keywords, the preset tag with the most occurrences is used as the target tag corresponding to the target measure.

[0060] Embodiment 2

[0061] Embodiment 2 of the present invention provides a method for determining a target object, and the method is used to determine the target object for implementing the target measure. The method includes the following steps:

[0062] S001. Screen the obtained initial objects based on the implementation conditions of the target measure to obtain the first intermediate object.

[0063] Specifically, in an embodiment of the present invention, the target measure is the issued target policy, and the target object is the target enterprise that can implement the target policy.

[0064] For example, the implementation conditions of the target measure include: applicable regions, types of applicable objects, etc.

[0065] S002. Obtain the first similarity degree between the target vector and the description information vector; wherein, the target vector is the vector obtained by vectorizing the target measure, and the description information vector is the vector obtained by vectorizing the description information of the first intermediate object.

[0066] Specifically, those skilled in the art know that any method for obtaining the similarity degree between two vectors in the prior art belongs to the protection scope of the present invention and will not be elaborated here. The description information of the first intermediate object includes the introduction, development history, core business, products and services, team strength, etc. of the first intermediate object.

[0067] S003. Obtain the second similarity degree between the target vector and the key sentence vector, wherein the key sentence vector is the vector obtained by vectorizing the sentences containing the target keyword in the description information, and the target keyword is the keyword contained in the target measure.

[0068] Specifically, obtain the sentences containing the target keyword in the description information, and vectorize the sentences containing the target keyword in the summary description information to obtain the key sentence vector.

[0069] S004. Obtain the quantity ratio y of the target keyword in the description information.

[0070] Specifically, those skilled in the art know that any method for obtaining the quantity ratio of the target keyword in the description information in the prior art belongs to the protection scope of the present invention. For example, based on the word count calculation, first determine the number of times the target keyword appears in the description information, and then obtain the total number of all words in the paragraph. Divide the number of times the target keyword appears in the description information by the total number of all words in the paragraph to obtain the quantity ratio y.

[0071] S005. Obtain the weighted sum of the first similarity degree and the second similarity degree as the target similarity degree, wherein the ratio of the weight of the first similarity degree to the weight of the second similarity degree is equal to 1 / (1 + y).

[0072] Specifically, the greater the proportion y of the target keyword in the description information, the greater the weight of the second similarity degree. It can be understood that when the proportion y of the target keyword in the description information is greater, the second similarity degree obtained through the key sentence vector is more accurate. Therefore, the weight of the second similarity degree is greater.

[0073] S006. Based on the target similarity degree, determine the first intermediate object as the target object for the implementation of the target measure.

[0074] In summary, based on the implementation conditions of the target measure, screen the obtained initial objects to obtain the first intermediate object, obtain the first similarity degree between the target vector and the description information vector, obtain the second similarity degree between the target vector and the key sentence vector, obtain the proportion of the number of target keywords in the description information, obtain the weighted sum of the first similarity degree and the second similarity degree as the target similarity degree, and based on the target similarity degree, determine the first intermediate object as the target object for the implementation of the target measure. The present invention determines the target object through the similarity degree between the target measure and the description information and the similarity degree between the target measure and the key sentence, and improves the accuracy of obtaining the target object of the target measure through the weights of the first similarity degree and the second similarity degree.

[0075] Specifically, S006 further includes:

[0076] S061. If the target similarity degree is greater than the preset similarity threshold, mark the first intermediate object as the second intermediate object and execute S062. It can be understood that by comparing the target similarity degree with the preset similarity threshold, preliminary screening is performed.

[0077] S062. Input the description information vector corresponding to the second intermediate object into the target classification model to obtain the target probability value, where the target probability value is the matching degree between the description information vector and the target measure.

[0078] Specifically, the target classification model is the target neural network model. Specifically, construct a neural network model, train the constructed neural network model, and if the trained neural network model meets the training requirements, use the trained neural network model as the target neural network model.

[0079] S063. Sort the target probability values in descending order to obtain the target permutation order.

[0080] S064. Obtain the target quantity p input by the user, and based on a number of historical quantities and the historical accuracy corresponding to each historical quantity, obtain the target accuracy corresponding to the target quantity p. The historical accuracy is the F1 value when the first historical quantity of historical permutation order corresponding historical objects are target objects; wherein, the historical permutation order is the order from large to small of historical probability values, and the historical probability value is the value obtained by inputting the description information of the historical object into the target classification model.

[0081] Wherein, F1 = 2×(P×R) / (P + R), where P is the accuracy rate when the first historical quantity of historical permutation order corresponding historical objects are target objects, and R is the recall rate when the first historical quantity of historical permutation order corresponding historical objects are target objects.

[0082] Specifically, when a historical quantity is equal to p, use the historical accuracy corresponding to this historical quantity as the target accuracy corresponding to p; if any historical quantity is not equal to p, obtain the historical quantity closest to p, and use the historical accuracy corresponding to this historical quantity as the target accuracy.

[0083] S065. If [(maximum accuracy - target accuracy) / maximum accuracy] is less than the preset threshold, determine that the second intermediate object corresponding to the first p matching degrees in the target permutation order is the target object, where the maximum accuracy is the maximum value among all historical accuracies. It can be understood that when the gap between the target accuracy and the maximum accuracy is not large, determine the target object according to the target quantity input by the user to meet the user's needs.

[0084] S066. If [(maximum accuracy - target accuracy) / maximum accuracy] is not less than the preset threshold, determine that the second intermediate object corresponding to the first q matching degrees in the target permutation order is the target object, where q is the historical quantity corresponding to the maximum accuracy. It can be understood that when the gap between the target accuracy and the maximum accuracy is large, still determine the target object according to the historical quantity corresponding to the maximum accuracy to avoid reducing the accuracy rate of the determined target object.

[0085] In summary, if the target similarity degree is greater than the preset similarity threshold, mark the first intermediate object as the second intermediate object, input the description information vector corresponding to the second intermediate object into the target classification model to obtain the target probability value, sort the target probability values in descending order to obtain the target permutation order, obtain the target number p input by the user, and based on a number of historical numbers and the historical accuracy corresponding to each historical number, obtain the target accuracy corresponding to the target number p. If [(maximum accuracy - target accuracy) / maximum accuracy] is less than the preset threshold, determine that the second intermediate objects corresponding to the first p matching degrees in the target permutation order are the target objects. If [(maximum accuracy - target accuracy) / maximum accuracy] is not less than the preset threshold, determine that the second intermediate objects corresponding to the first q matching degrees in the target permutation order are the target objects. This invention illustrates two cases, namely the number input by the user and the historical number corresponding to the maximum accuracy. Under the condition of meeting the preset threshold, the object is determined according to the target number. Under the condition of not meeting the preset threshold, the target object is determined based on the maximum accuracy.

[0086] An embodiment of the present invention also provides a non-transitory computer-readable storage medium, which can be set in an electronic device to store a computer program related to a method for implementing a method in the method embodiment. The computer program is loaded and executed by the processor to implement the method provided in the above embodiment.

[0087] An embodiment of the present invention also provides an electronic device, including: a processor, a memory, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method provided in the above embodiment is implemented.

[0088] An embodiment of the present invention also provides a computer program product, which includes program code. When the program product runs on an electronic device, the program code is used to cause the electronic device to execute the steps in the method according to various exemplary embodiments of the present invention described above in this specification.

[0089] Although some specific embodiments of the present invention have been described in detail by way of examples, those skilled in the art should understand that the above examples are only for illustration and not for limiting the scope of the present invention. Those skilled in the art should also understand that various modifications can be made to the embodiments without departing from the scope and spirit of the present invention.

Claims

1. A method for determining a target measure, characterized in that: The method is used to determine a paragraph of a target text as a target measure, and the method comprises the following steps: S110, extracting keywords from a target paragraph of a target text to obtain a target keyword list, wherein the target keyword list includes a plurality of target keywords; S120, matching the target keyword with the acquired preset word library, wherein the preset word library includes a plurality of preset words related to the target measure; S130, if the match is successful, obtain the importance value of the target keyword in the target paragraph, and obtain the negatively related words corresponding to the target keyword, wherein the negatively related words are words whose probability of co-appearing in a sentence corresponding to a preset measure is less than a preset correlation threshold; S140, if the negatively related words appear in the sentence where the target keyword is located, obtaining a negatively related probability value of the target keyword, wherein the negatively related probability value is less than 0; S150, obtaining a target length of the target paragraph, and comparing the target length with a preset length threshold to obtain a target length score, wherein the smaller the absolute value of the difference between the target length and the preset length threshold, the larger the target length score; S160, determining the target paragraph as a target measure based on the importance value of the target keyword in the target paragraph, the negative correlation probability value of the target keyword and the target length score.

2. The target measure determination method according to claim 1, characterized in that: The S160 also includes: S161, summing up the importance values ​​of all successfully matched target keywords to obtain an importance sum value; S162, summing up the negative correlation probability values ​​of all target keywords to obtain a negative correlation probability sum value; S163, summing the importance and value, the negative correlation probability and value, and the target length score to obtain a final score; S164: If the final score is greater than the preset measure threshold, the target paragraph is determined as the target measure.

3. The target measure determination method according to claim 1, characterized in that: The TF-IDF value of the target keyword in the target paragraph is used as the importance value of the target keyword in the target paragraph.

4. The target measure determination method according to claim 1, characterized in that: Before S100, the method further includes: obtaining an initial file, and processing the initial file using a corresponding preset processing method based on the type of the initial file to obtain a target text.

5. The target measure determination method according to claim 4, characterized in that: When the type of the initial file is a picture format, the preset processing method corresponding to the type of the initial file is the OCR technology.

6. The target measure determination method according to claim 1, characterized in that: S150 also includes obtaining the target length score by the following steps: Get the absolute value x of the difference between the target length and the preset length threshold; Obtain a preset difference range and a preset score corresponding to the preset difference range; The preset score corresponding to the preset difference range where the absolute value x is located is obtained as the target length score.

7. The target measure determination method according to claim 1, characterized in that: S150 also includes obtaining the target length score by the following steps: Get the absolute value x of the difference between the target length and the preset length threshold; A target length score is obtained, where the target length score is a ratio of a preset length threshold and an absolute value x.

8. The target measure determination method according to claim 1, characterized in that: After S160, it also includes: S170, highlighting the target measure in the target text, and dividing the target paragraph that is not the target measure into a plurality of target sentences; S180, obtaining a target sentence containing a target keyword and marking it as an intermediate sentence, and vectorizing the intermediate sentence to obtain an intermediate vector; S190, obtaining target similarity between the target vector and the intermediate vector; S200: If the target similarity is greater than a preset similarity threshold, highlight the intermediate sentence corresponding to the intermediate vector, wherein the target vector is a vector obtained by vectorizing the target measure.

9. A non-transitory computer-readable storage medium, characterized in that: The storage medium stores a computer program, which is loaded and executed by a processor to implement the target measure determination method as described in any one of claims 1 to 8.

10. An electronic device comprising: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the target measure determination method as described in any one of claims 1 to 8 when executing the computer program.

Images