Merchant comprehensive value evaluation method and system based on artificial intelligence, and terminal equipment

Abstract

The embodiment of the invention provides a merchant comprehensive value evaluation method and system based on artificial intelligence and terminal equipment, and belongs to the technical field of artificial intelligence. The method comprises the steps of obtaining historical consumption information and an initial evaluation text and initial score information under the historical consumption information; performing text emotion consistency analysis on the initial evaluation text to obtain a target analysis result; determining a target data state according to the target analysis result in combination with the initial score information; calculating an information association degree between the historical consumption information and the initial evaluation text; deleting the initial evaluation text according to the target data state and the information association degree to obtain a target evaluation text, and obtaining association score information from the initial score information; performing data clustering on the target evaluation text to obtain a target clustering result, and correcting the associated score information according to the target clustering result to obtain target score information; and determining a merchant value evaluation result according to the historical consumption information and the target evaluation text in combination with the target score information.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based merchant comprehensive value assessment method, system and terminal device. Background Art

[0002] In today's digital business environment, when recommending stores to users, the comprehensive evaluation value of the store can be used to screen out stores that meet their needs and preferences, thereby significantly improving the user's shopping efficiency and experience. However, in terms of the current technical status, in the process of comprehensive evaluation of stores, some data is often simply weighted averaged, which results in the evaluation results not being able to truly reflect the actual situation of the store, and the resulting comprehensive evaluation value of the goods is not accurate enough, which will have a serious negative impact on subsequent store recommendations and store rankings, thereby greatly reducing the user's consumption experience. Summary of the invention

[0003] The main purpose of the embodiments of the present invention is to provide a merchant comprehensive value assessment method, system and terminal device based on artificial intelligence, aiming to solve the problem that the comprehensive evaluation value of goods in related technologies is not accurate enough, which will have a serious negative impact on subsequent store recommendations and store rankings, thereby greatly reducing the user's consumption experience.

[0004] In a first aspect, an embodiment of the present invention provides a merchant comprehensive value assessment method based on artificial intelligence, comprising:

[0005] Obtaining historical consumption information corresponding to the target user's consumption at the target merchant, and obtaining an initial evaluation text corresponding to the target merchant under the historical consumption information and initial rating information corresponding to the initial evaluation text;

[0006] Performing text sentiment consistency analysis on the initial evaluation text to obtain a target analysis result corresponding to the initial evaluation text;

[0007] Determine the target data state corresponding to the initial evaluation text according to the target analysis result combined with the initial scoring information;

[0008] Calculating the information correlation between the historical consumption information and the initial evaluation text;

[0009] Deleting the initial evaluation text according to the target data state and the information relevance to obtain a target evaluation text, and obtaining the associated scoring information corresponding to the target evaluation text from the initial scoring information;

[0010] Performing data clustering on the target evaluation text to obtain a target clustering result, and performing data correction on the associated scoring information according to the target clustering result to obtain target scoring information corresponding to the target evaluation text;

[0011] The merchant value assessment result corresponding to the target merchant is determined using an artificial intelligence model based on the historical consumption information and the target evaluation text combined with the target rating information.

[0012] In a second aspect, an embodiment of the present invention provides a merchant comprehensive value assessment system based on artificial intelligence, comprising:

[0013] A data acquisition module is used to obtain the historical consumption information corresponding to the target user's consumption at the target merchant, and obtain the initial evaluation text corresponding to the target merchant under the historical consumption information and the initial rating information corresponding to the initial evaluation text;

[0014] A sentiment analysis module, used to perform text sentiment consistency analysis on the initial evaluation text to obtain a target analysis result corresponding to the initial evaluation text;

[0015] A state determination module, used to determine the target data state corresponding to the initial evaluation text according to the target analysis result combined with the initial scoring information;

[0016] A correlation calculation module, used to calculate the information correlation degree between the historical consumption information and the initial evaluation text;

[0017] A data association module, configured to delete the initial evaluation text according to the target data state and the information association degree to obtain a target evaluation text, and obtain associated scoring information corresponding to the target evaluation text from the initial scoring information;

[0018] A data correction module, used for performing data clustering on the target evaluation text to obtain a target clustering result, and performing data correction on the associated scoring information according to the target clustering result to obtain target scoring information corresponding to the target evaluation text;

[0019] The value assessment module is used to determine the merchant value assessment result corresponding to the target merchant by using an artificial intelligence model based on the historical consumption information and the target evaluation text combined with the target scoring information.

[0020] In a third aspect, an embodiment of the present invention further provides a terminal device, comprising a processor, a memory, a computer program stored in the memory and executable by the processor, and a data bus for realizing connection and communication between the processor and the memory, wherein when the computer program is executed by the processor, the steps of any one of the methods for comprehensive merchant value assessment based on artificial intelligence provided in the specification of the present invention are implemented.

[0021] The embodiment of the present invention provides a merchant comprehensive value assessment method, system and terminal device based on artificial intelligence. The method includes: obtaining historical consumption information corresponding to a target user after consumption at a target merchant, and obtaining an initial evaluation text corresponding to the target merchant under the historical consumption information and initial scoring information corresponding to the initial evaluation text, and then performing text sentiment consistency analysis on the initial evaluation text to obtain a target analysis result corresponding to the initial evaluation text, and then determining a target data state corresponding to the initial evaluation text according to the target analysis result combined with the initial scoring information, and calculating the corresponding information association degree between the historical consumption information and the initial evaluation text; thereby deleting the initial evaluation text according to the target data state and the information association degree to obtain a target evaluation text, and obtaining associated scoring information corresponding to the target evaluation text from the initial scoring information, and then performing data clustering on the target evaluation text to obtain a target clustering result, and performing data correction on the associated scoring information according to the target clustering result to obtain the target scoring information corresponding to the target evaluation text; finally, determining a merchant value assessment result corresponding to the target merchant by using an artificial intelligence model according to the historical consumption information and the target evaluation text combined with the target scoring information. This method analyzes the text sentiment consistency of the initial evaluation text, determines the target data state of the initial evaluation text in combination with the initial scoring information, and deletes the initial evaluation text according to the target data state and information relevance, which can effectively remove those evaluations that may be abnormal, false or have low relevance to consumer information, and then filter out these invalid information, thereby improving the quality of the evaluation data. Then, the target evaluation text is clustered, and the associated scoring information is corrected according to the target clustering result, so that the score can more accurately reflect the real experience of consumers. Therefore, by combining historical consumption information, target evaluation text and target scoring information, the merchant value evaluation result corresponding to the target merchant is determined by using an artificial intelligence model, which can comprehensively and objectively evaluate the value of the merchant from multiple dimensions, so that the merchant's operating conditions and development potential can be more accurately measured, so that users can make more informed decisions based on these real and accurate information when choosing merchants, reducing consumption risks. This also solves the problem that the comprehensive evaluation value of the product in the relevant technology is not accurate enough, which will have a serious negative impact on the subsequent store recommendation and store sorting, thereby greatly reducing the user's consumption experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0023] Figure 1 A flowchart of a merchant comprehensive value assessment method based on artificial intelligence provided by an embodiment of the present invention;

[0024] Figure 2 A schematic diagram of the module structure of a merchant comprehensive value assessment system based on artificial intelligence provided by an embodiment of the present invention;

[0025] Figure 3 A schematic block diagram of the structure of a terminal device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0028] It should be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0029] The embodiment of the present invention provides a merchant comprehensive value assessment method, system and terminal device based on artificial intelligence. The merchant comprehensive value assessment method based on artificial intelligence can be applied to a terminal device, which can be an electronic device such as a tablet computer, a laptop computer, a desktop computer, a personal digital assistant and a wearable device. The terminal device can be a server or a server cluster.

[0030] Some embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0031] Please refer to Figure 1 , Figure 1 A flowchart of a merchant comprehensive value assessment method based on artificial intelligence provided by an embodiment of the present invention.

[0032] like Figure 1As shown, the merchant comprehensive value assessment method based on artificial intelligence includes steps S101 to S107.

[0033] Step S101: Obtain historical consumption information corresponding to a target user's consumption at a target merchant, and obtain an initial evaluation text corresponding to the target merchant under the historical consumption information and initial rating information corresponding to the initial evaluation text.

[0034] Exemplarily, the target user's historical consumption information corresponding to consumption at the target merchant is obtained from the database, and the initial evaluation text corresponding to the target user's evaluation of the target merchant under the historical consumption information is obtained from the database or by crawling means, and the initial rating information corresponding to the target user's rating of the target merchant under the initial evaluation text is obtained. The target user is a plurality of users who consume at the target merchant.

[0035] Step S102: performing text sentiment consistency analysis on the initial evaluation text to obtain a target analysis result corresponding to the initial evaluation text.

[0036] Exemplarily, the initial evaluation text is segmented to obtain multiple segmented texts, and then each segmented text is sentimentally classified using a sentiment classification model such as an SVM model to obtain the sentiment type corresponding to each segmented text, thereby comparing all sentiment types. When all sentiment types are completely consistent, the target analysis result corresponding to the initial evaluation text is determined to be sentimentally consistent; when all sentiment types are not completely consistent, the target analysis result corresponding to the initial evaluation text is determined to be sentimentally inconsistent.

[0037] In some embodiments, the text sentiment consistency analysis of the initial evaluation text to obtain the target analysis result corresponding to the initial evaluation text includes: performing text segmentation on the initial evaluation text to obtain multiple segmented evaluation texts, and determining a first evaluation text from the multiple segmented evaluation texts and a first remaining evaluation text obtained by removing the first evaluation text from the multiple segmented evaluation texts; performing sentiment classification on the first evaluation text to obtain a first sentiment type corresponding to the first evaluation text and a first probability information corresponding to the first sentiment type; performing sentiment analysis on any second evaluation text in the first remaining evaluation text and the first evaluation text after text merging to obtain a second sentiment type corresponding to the second evaluation text and a second probability information corresponding to the second sentiment type; comparing the first sentiment type and the second sentiment type to obtain Obtain a type comparison result. When the type comparison result is that the first emotion type and the second emotion type are the same, determine the first analysis result corresponding to the first evaluation text and the second evaluation text according to the first probability information and the second probability information, wherein the first analysis result is used to characterize the analysis result of the emotion consistency between the first evaluation text and the second evaluation text; when the type comparison result is that the first emotion type and the second emotion type are not the same, determine the first analysis result corresponding to the first evaluation text and the second evaluation text as a preset result; perform text merging on the multiple segmented evaluation texts according to the first analysis result to obtain a target merging result; perform text emotion consistency analysis on the initial evaluation text according to the target merging result to obtain the target analysis result corresponding to the initial evaluation text.

[0038] Exemplarily, the initial evaluation text is segmented into a plurality of relatively independent sentences according to the grammatical rules of the language, punctuation marks and other segmentation methods to obtain a plurality of segmented evaluation texts.

[0039] Exemplarily, a plurality of segmented evaluation texts are determined as first evaluation texts in sequence, and then the selected first evaluation text is removed from all segmented evaluation texts, and the remaining segmented evaluation texts constitute the first remaining evaluation text.

[0040] Exemplarily, a machine learning model, such as a naive Bayes classifier, a support vector machine or other sentiment classification model, is used to analyze the first evaluation text to obtain the first sentiment type (such as positive, negative, neutral) corresponding to the first sentiment type and the first probability information corresponding to the first sentiment type. For example, if a machine learning model is used, the model will output the probability of each sentiment type, select the sentiment type with the highest probability as the first sentiment type, and record the probability as the first probability information.

[0041] Exemplarily, for each second evaluation text in the first remaining evaluation text, the second evaluation text and the first evaluation text are merged, that is, the text is spliced, so as to perform sentiment analysis using the same method as the sentiment classification of the first evaluation text, and obtain the second sentiment type corresponding to each second evaluation text and the second probability information corresponding to the second sentiment type. That is, the second sentiment type combines the first evaluation text and the second evaluation text. If the second evaluation text is consistent with the first evaluation text in sentiment, the second probability information will be greater than the first probability information, that is, the second evaluation text plays an enhancing role. If the second evaluation text is inconsistent with the first evaluation text in sentiment or the second evaluation text is relatively neutral in sentiment, it will cause the second probability information or the second sentiment type to be significantly different from the first probability information or the first sentiment type.

[0042] Exemplarily, the first emotion type of the first evaluation text is compared with the second emotion type of each second evaluation text one by one to obtain a type comparison result. When the type comparison result is that the first emotion type and the second emotion type are the same, the first analysis result is determined by calculating the difference between the probabilities of the two. For example, when the second probability information is greater than the first probability information, it means that the second evaluation text has strong emotional consistency with the first evaluation text, and the first analysis result can be directly determined to be strongly consistent; when the second probability information is less than or equal to the first probability information, it means that the second evaluation text may be emotionally consistent or inconsistent with the first evaluation text, and the first analysis result can be directly determined to be not necessarily consistent.

[0043] Exemplarily, when the type comparison result is that the first emotion type and the second emotion type are different, the first analysis result corresponding to the first evaluation text and the second evaluation text is determined as a preset result, and the preset result is set as emotion inconsistency.

[0044] Exemplarily, which segmented evaluation texts can be merged is determined based on the first analysis result. For example, if the first analysis result is strongly consistent, they can be directly merged; if the sentiment is inconsistent, they are not merged; if the first analysis result is not necessarily consistent, the second evaluation text needs to be classified as an independent text for sentiment, and when the classification result is consistent with the sentiment classification result of the first evaluation text, the first evaluation text and the second evaluation text are merged, so that multiple segmented evaluation texts are merged according to the above merging rules to obtain the target merged result.

[0045] For example, according to the target merging result, the sentiment consistency of each merged text block is comprehensively considered to perform an overall text sentiment consistency analysis on the initial evaluation text. When there is only one merging result in the target merging result, it means that the target analysis result corresponding to the initial evaluation text is sentiment consistent; when there are two or more merging results in the target merging result, it means that the target analysis result corresponding to the initial evaluation text is sentiment inconsistent.

[0046] Specifically, when determining the first analysis result, not only whether the emotion type is the same is considered, but also the first probability information and the second probability information are combined. This makes the analysis result more objective and accurate. When the first emotion type and the second emotion type are different, the first analysis result is determined as the preset result, which can effectively identify the contradictory emotions that may exist in the text, thereby discovering some emotional information hidden in the text through text segmentation and emotion consistency analysis, and then merging multiple segmented evaluation texts according to the first analysis result to obtain the target merging result. Accurate text emotion consistency analysis results can help merchants better understand the true feelings of consumers, and thus provide good support for subsequent merchant value evaluation.

[0047] In some embodiments, the text merging of the multiple segmented evaluation texts according to the first analysis result to obtain a target merging result includes: obtaining a first consistency text corresponding to the first evaluation text from the second evaluation text according to the first analysis result; removing the first consistency text from the first remaining evaluation text to obtain a second remaining evaluation text; performing sentiment classification on the first consistency text to obtain a third emotion type corresponding to the first consistency text and third probability information corresponding to the third emotion type; performing sentiment analysis on any third evaluation text in the second remaining evaluation text and the first consistency text after merging them to obtain a fourth emotion type corresponding to the third evaluation text and fourth probability information corresponding to the fourth emotion type; determining the corresponding second analysis result between the first consistency text and the third evaluation text according to the third emotion type, the third probability information, the fourth emotion type and the fourth probability information; determining the fourth evaluation text from the second remaining evaluation text, and obtaining the first evaluation text and the fourth evaluation text from the first analysis result. a third analysis result corresponding to the first consistency text and a fourth analysis result corresponding to the first consistency text and the fourth evaluation text obtained from the second analysis result; when the third analysis result and the fourth analysis result are consistent, the first consistency text, the first evaluation text and the fourth evaluation text are merged according to the third analysis result to obtain a first merged result; when the third analysis result and the fourth analysis result are inconsistent, the first consistency text and the first evaluation text are merged to obtain a first merged text, and the sentiment consistency between the first merged text and the fourth evaluation text is calculated to obtain a fifth analysis result corresponding to the first merged text; according to the fifth analysis result, the first consistency text, the first evaluation text and the fourth evaluation text are merged to obtain the first merged result; the first merged result is removed from the initial evaluation text to obtain other evaluation texts, and texts are merged according to the other evaluation texts to obtain a second merged result; the target merged results corresponding to the multiple segmented evaluation texts are determined according to the first merged result and the second merged result.

[0048] Exemplarily, a first analysis result is obtained between each second evaluation text and the first evaluation text. If the first analysis result shows that the emotions of the two are consistent, these second evaluation texts are determined as the first consistency texts corresponding to the first evaluation text. For example, if the first analysis result shows that a second evaluation text has the same emotion type as the first evaluation text and the second probability information is greater than the first probability information, then the second evaluation text belongs to the first consistency text.

[0049] Exemplarily, the first consistent text is removed from the first remaining evaluation text, and the remaining text constitutes the second remaining evaluation text. This step is to separate the part that is consistent with the first evaluation text in sentiment, so as to facilitate subsequent separate processing.

[0050] Exemplarily, the first consistent text is analyzed by a machine learning model such as a naive Bayes classifier, a support vector machine, etc. to obtain the corresponding sentiment type (such as positive, negative, neutral) and the probability information corresponding to the sentiment type, i.e., the third sentiment type and the third probability information, so that for each third evaluation text in the second remaining evaluation text, the third evaluation text and the first consistent text are concatenated to perform sentiment analysis using the same method as the sentiment classification of the first consistent text, and the fourth sentiment type corresponding to each third evaluation text and the fourth probability information corresponding to the fourth sentiment type are obtained. That is, the fourth sentiment type combines the first consistent text and the third evaluation text. If the third evaluation text is consistent with the first consistent text in sentiment, the fourth probability information will be greater than the third probability information, that is, the third evaluation text plays an enhancing role. If the third evaluation text is inconsistent with the first consistent text in sentiment or the third evaluation text is relatively neutral in sentiment, the fourth probability information or the fourth sentiment type will be significantly different from the third probability information or the third sentiment type.

[0051] Exemplarily, the third emotion type and the fourth emotion type are compared to obtain a comparison result. When the comparison result is that the third emotion type and the fourth emotion type are the same, then when the third probability information is greater than the fourth probability information, then it is determined that the second analysis result corresponding to the first consistency text and the third evaluation text is emotionally consistent. When the third probability information is less than or equal to the fourth probability information, then it is determined that the second analysis result corresponding to the first consistency text and the third evaluation text is emotionally inconsistent. When the comparison result is that the third emotion type and the fourth emotion type are different, then the second analysis result corresponding to the first consistency text and the third evaluation text is determined as a preset result, that is, it is determined to be emotionally inconsistent.

[0052] Exemplarily, a fourth evaluation text is randomly selected from the second remaining evaluation text multiple times, and then the third analysis result corresponding to the first evaluation text and the fourth evaluation text is found from the first analysis result obtained previously, and the fourth analysis result corresponding to the first consistency text and the fourth evaluation text is obtained from the second analysis result. When the third analysis result and the fourth analysis result are consistent, it means that the first evaluation text, the first consistency text and the fourth evaluation text have a high consistency in emotion, and they can be merged into a first merged result. When the third analysis result and the fourth analysis result are inconsistent, the first consistency text and the first evaluation text are first merged to obtain a first merged text. Then the emotional consistency between the first merged text and the fourth evaluation text is calculated to obtain a fifth analysis result. Finally, according to the fifth analysis result, it is decided whether to merge the fourth evaluation text with the first merged text to obtain the first merged result.

[0053] Exemplarily, the first merged result is removed from the initial evaluation text, and the remaining text is the other evaluation text. The above-mentioned similar sentiment analysis and merging steps are repeated for the other evaluation texts to finally obtain the second merged result.

[0054] Exemplarily, the first merging result and the second merging result are integrated to obtain a target merging result corresponding to the multiple segmented evaluation texts.

[0055] For example, multiple segmented evaluation texts include text a, text b, text c and text d, then text a is taken as the first evaluation text, and text b, text c and text d are determined as the first remaining evaluation text. If text b is determined to be the first consistency text of the first evaluation text, text b is determined as the first consistency text, and then text c and text d are determined as the second remaining evaluation text, and then sentiment classification is performed on text b to obtain a third sentiment type and third probability information corresponding to the third sentiment type, and text b is respectively merged with text c and text d in the second remaining evaluation text to obtain text b text c and text b text d, so as to obtain the corresponding fourth sentiment type and fourth probability information corresponding to the fourth sentiment type, so as to judge the analysis result between text b and text c and the analysis result between text b and text d, and then combine to obtain the second analysis result corresponding to the first consistency text and the third evaluation text, and then arbitrarily select a text from the second remaining evaluation text to determine as the fourth evaluation text, for example, determine text c as the fourth evaluation text, and then obtain the third analysis result between text a and text c from the first analysis result and the fourth probability information from the second analysis result. The fourth analysis result between text b and text c is obtained in the result. If the third analysis result and the fourth analysis result are consistent, if the third analysis result and the fourth analysis result both indicate that the sentiment is consistent, then text a, text b and text c are directly merged to obtain the first merged result. If the third analysis result and the fourth analysis result are consistent, but the third analysis result and the fourth analysis result both indicate that the sentiment is inconsistent, then text a and text b are merged, but not merged with text c to obtain the first merged result. If the third analysis result and the fourth analysis result are inconsistent, then the first consistent text (text b) and the first evaluation text (text a) are merged to obtain the first merged text (text a text b), and the sentiment consistency between the first merged text and the fourth evaluation text (text c) is calculated according to the above sentiment analysis method to obtain the fifth analysis result corresponding to the first merged text. If the fifth analysis result indicates that the first merged text and the fourth evaluation text are emotionally consistent, then text a, text b and text c are directly merged to obtain the first merged result. If the fifth analysis result indicates that the first merged text and the fourth evaluation text are emotionally inconsistent, then text a and text b are merged, but not merged with text c to obtain the first merged result.

[0056] Specifically, by performing sentiment classification and analysis on different text subsets multiple times, the sentiment information in the text can be captured more meticulously. When dealing with the sentiment relationship between multiple texts, it is possible to effectively identify the contradictory sentiments that may exist in the text, and then through gradual text merging and analysis, it is possible to discover some sentiment information hidden in the text, so that by merging texts with consistent sentiments, accurate text merging results and sentiment analysis can help merchants better understand the real feelings of consumers, and thus provide good support for subsequent merchant value evaluation.

[0057] Step S103: determining a target data state corresponding to the initial evaluation text according to the target analysis result and the initial scoring information.

[0058] Exemplarily, when the target analysis result is consistent with the emotion corresponding to the initial evaluation text, the target emotion type corresponding to the initial evaluation text is obtained, and the scoring range corresponding to the target emotion type is obtained based on expert experience, so as to determine the target data state corresponding to the initial evaluation text according to whether the initial scoring information is within the scoring range. When the initial scoring information is within the scoring range, the target data state corresponding to the initial evaluation text is determined to be normal; when the initial scoring information is not within the scoring range, the target data state corresponding to the initial evaluation text is determined to be abnormal.

[0059] Exemplarily, when the target analysis result is that the emotions corresponding to the initial evaluation text are inconsistent, the text proportion corresponding to each emotion type in the initial evaluation text is obtained, and then the scoring range corresponding to each emotion type is obtained based on expert experience, and then the scoring range is weighted and summed according to the text proportion to obtain the target scoring range, and then the target data state corresponding to the initial evaluation text is determined according to whether the initial scoring information is in the target scoring range. When the initial scoring information is in the target scoring range, the target data state corresponding to the initial evaluation text is determined to be normal; when the initial scoring information is not in the target scoring range, the target data state corresponding to the initial evaluation text is determined to be abnormal.

[0060] In some embodiments, determining the target data state corresponding to the initial evaluation text based on the target analysis result in combination with the initial scoring information includes: when the target analysis result is the preset result, determining the first position distribution information corresponding to the first merged result based on the initial evaluation text and determining the second position distribution information corresponding to the second merged result based on the initial evaluation text; determining the core evaluation text corresponding to the initial evaluation text and the target intention corresponding to the core evaluation text of the target user based on the first position distribution information and the second position distribution information; determining the scoring interval corresponding to the initial evaluation text based on the target intention and the mapping table, and determining the target data state corresponding to the initial evaluation text based on the scoring interval and the initial scoring information.

[0061] Exemplarily, it is determined whether the target analysis result is a preset result. That is, it is determined whether the target analysis result is sentiment inconsistent. When the target analysis result is sentiment inconsistent, the first position distribution information corresponding to the first merged result in the initial evaluation text is determined according to the initial evaluation text, and the second position distribution information corresponding to the second merged result in the initial evaluation text is determined according to the initial evaluation text. The first position distribution information or the second position distribution information can be represented by a sentence number or a character position.

[0062] Exemplarily, the first position distribution information and the second position distribution information are combined to determine which of the first merged result and the second merged result in the initial evaluation text is more likely to be in the beginning or the end position, and then the emotional tendency and semantic expression of the initial evaluation text are determined based on the first position distribution information and the second position distribution information of the first merged result and the second merged result to obtain the target intention corresponding to the target user.

[0063] For example, if the text of the first merged result is concentrated in the beginning and expresses high praise for the product, while the text of the second merged result raises some minor questions at the end, then the text in the beginning may be the core evaluation text, and the target intention may be to emphasize the advantages of the product.

[0064] For example, a mapping table is established, which records the correspondence between different target intentions and scoring intervals. For example, the target intention of "highly praise" may correspond to a scoring interval of 8-10 points; "moderately satisfied with minor problems" corresponds to 6-8 points; "unsatisfied" corresponds to 1-3 points, etc., so that according to the determined target intention, the corresponding scoring interval is found in the mapping table.

[0065] Exemplarily, the determined scoring interval is compared with the initial scoring information. The initial scoring information may be a specific score given by the user. If the initial score is within the scoring interval, it means that the evaluation is reasonable, and the target data state can be marked as "normal"; if the initial score is higher than the upper limit of the scoring interval or lower than the lower limit of the scoring interval, the target data state can be marked as "abnormal".

[0066] Specifically, by analyzing the position distribution information of the text to determine the core evaluation text and target intent, we can understand the user's real evaluation more comprehensively and accurately. We avoid judging the rationality of the evaluation based on a single initial score, consider the richness and complexity of the text content, make the analysis of the evaluation more in-depth, and use a mapping table to match the target intent with the score range, providing an objective standard for the evaluation. This can reduce the impact of inconsistent rating scales of different users and improve the comparability and reliability of the evaluation.

[0067] Step S104: Calculate the information correlation between the historical consumption information and the initial evaluation text.

[0068] For example, based on the historical consumption information of the target user, which may be order records of online shopping platforms and offline consumption order records, the type of consumer products corresponding to the target user is determined. The type of consumer product may be a product of the target merchant.

[0069] Exemplarily, natural language processing techniques such as word segmentation and part-of-speech tagging are used to extract words related to consumer evaluation from the initial evaluation text, such as "good quality", "affordable price", "considerate service", etc. These words are consumer evaluation words, so that the corresponding consumer evaluation words are associated with each type of consumer product. For example, if the consumer product type is an electronic product, the consumer evaluation words may be "strong performance" and "clear screen", so that the association between different consumer product types and consumer evaluation words can be found through statistical analysis, such as which evaluation words appear more frequently in the evaluation text of a specific product type. Thus, the extracted consumer evaluation words are matched with the determined consumer product type, and each evaluation word is determined to be associated with which product type, and then the corresponding association index is determined by the frequency of occurrence of the evaluation word in the evaluation text of the specific product type, so as to calculate the score for the association between each group of consumer product types and consumer evaluation words, and then obtain the corresponding information association between the historical consumption information and the initial evaluation text.

[0070] Exemplarily, according to the above method, malicious evaluations of target merchants by peers can be reduced, thereby improving the accuracy and reliability of subsequent comprehensive merchant value assessments of the target merchants.

[0071] In some embodiments, the calculation of the information association degree corresponding to the historical consumption information and the initial evaluation text includes: performing part-of-speech analysis on the initial evaluation text to obtain relevant nouns and relevant evaluation words corresponding to relevant products in the initial evaluation text; determining preset evaluation words corresponding to the relevant nouns, and calculating the association degree between the relevant nouns and the relevant evaluation words based on the preset evaluation words; determining the association relationship between the relevant nouns and the relevant evaluation words based on the association degree; screening the relevant evaluation words based on the association relationship to obtain remaining evaluation words, and determining, based on the remaining evaluation words, related products that can be used to evaluate the target merchant; obtaining the target consumption product corresponding to the target user from the historical consumption information; performing overlap calculation based on the target consumption product, the relevant nouns and the related products to obtain a target overlap; and determining the information association degree corresponding to the historical consumption information and the initial evaluation text based on the target overlap.

[0072] Exemplarily, the initial evaluation text is processed using LTP of Harbin Institute of Technology, Jieba word segmentation, etc., so that each word in the initial evaluation text is marked with the corresponding part of speech, such as noun, verb, adjective, etc., and then nouns related to the product are screened out from the initial evaluation text with marked parts of speech, such as "mobile phone", "clothes", "cosmetics", etc.; at the same time, relevant evaluation words used to describe product characteristics, quality, usage experience, etc. are extracted, such as "easy to use", "beautiful", "durable", etc.

[0073] Exemplarily, a preset evaluation word library is established in advance according to different types of products. For example, for mobile phones, the preset evaluation words may include "good performance", "clear photos", "long battery life", etc., and then the correlation between the relevant nouns and the relevant evaluation words is calculated through statistical methods. For example, the frequency of the relevant evaluation words in the text related to the relevant noun can be calculated. The higher the frequency, the higher the correlation. The semantic similarity method can also be used to determine the degree of semantic similarity between the relevant evaluation words and the preset evaluation words to determine the correlation.

[0074] Exemplarily, a correlation threshold is set. When the correlation between the relevant nouns and the relevant evaluation words exceeds this threshold, it is considered that there is a correlation relationship between them. Based on the set threshold, it is determined which relevant evaluation words have a correlation relationship with the relevant nouns, and these combinations with a correlation relationship are recorded.

[0075] For example, according to the determined association relationship, the related evaluation words are screened, and those evaluation words that have no association relationship with the related nouns are removed to obtain the remaining evaluation words, and then according to the characteristics and attributes described by the remaining evaluation words, combined with the related nouns, the related products under the target merchant are determined to be used for evaluation. For example, if the remaining evaluation words are "soft" and "breathable", and the related noun is "towel", then it can be determined that the related product is the towel product under the target merchant.

[0076] Exemplarily, the historical consumption information of the target user is sorted and analyzed, and the product information consumed by the target user is extracted from it to determine the target consumption product. The target consumption product, related nouns and related products are then compared, and the overlap between them is calculated. For example, the target overlap can be obtained by calculating the ratio of the number of identical products to the total number of products, and then the corresponding information correlation between the historical consumption information and the initial evaluation text is determined based on the calculated target overlap. The higher the target overlap, the higher the information correlation between the two.

[0077] Specifically, in order to accurately identify malicious or erroneous evaluations of the target merchant in the initial evaluation text, the authenticity and rationality of the evaluation can be discovered by correlating the target user's historical consumption information with the initial evaluation text. Through this correlation analysis, such unreasonable evaluations can be effectively excluded, reducing the impact of malicious evaluations on the target merchant. After reducing malicious or erroneous evaluations, the subsequent user merchant comprehensive value evaluation data can more truly reflect the actual operating conditions and service quality of the target merchant, thereby making the evaluation results more credible, thereby significantly improving the accuracy and reliability of the merchant comprehensive value evaluation.

[0078] Step S105: deleting the initial evaluation text according to the target data state and the information relevance to obtain a target evaluation text, and obtaining associated scoring information corresponding to the target evaluation text from the initial scoring information.

[0079] Exemplarily, when the target data state corresponding to the initial evaluation text is abnormal or the information correlation between the initial evaluation text and the historical consumption information is low, the initial evaluation text is deleted to obtain the target evaluation text.

[0080] Exemplarily, the scoring information corresponding to each target evaluation text is obtained from the initial scoring information to obtain the associated scoring information.

[0081] Step S106: performing data clustering on the target evaluation text to obtain a target clustering result, and performing data correction on the associated scoring information according to the target clustering result to obtain target scoring information corresponding to the target evaluation text.

[0082] Exemplarily, keyword extraction is performed on the target evaluation text to obtain multiple target keywords, and then the word2vec model is used to obtain text vectors corresponding to the target keywords, and then the text vectors are used to perform text clustering on the target evaluation text according to the k-means clustering algorithm to obtain target clustering results.

[0083] Exemplarily, relevant scoring information corresponding to each subclass cluster in the target clustering result is obtained from the associated scoring information, so as to perform anomaly identification on the relevant scoring information to obtain abnormal scoring data, and then correct the abnormal scoring data according to the normal scoring data in the subclass cluster so that the abnormal scoring data is consistent with the normal scoring data, and then obtain the target scoring information corresponding to each target evaluation text.

[0084] For example, the mean of the relevant scoring information in each subclass cluster is calculated. If the relevant scoring information of a text differs greatly from the mean of the subclass cluster, its score can be corrected to the mean of the cluster. For example, if the mean score in a cluster is 8 points, and the score of one of the texts is only 3 points, and the text is semantically similar to other texts in the cluster, then the score of the text can be corrected to 8 points. Thus, according to the determined correction strategy, the associated scoring information is corrected. The score of each target evaluation text is adjusted to the corrected value, so as to obtain the target scoring information corresponding to the target evaluation text.

[0085] Specifically, by clustering the target evaluation text, outliers and deviations in the rating can be found and corrected. This enables the rating to more accurately reflect the actual situation expressed by the text and provide a more reliable basis for subsequent decision-making.

[0086] In some embodiments, the data clustering of the target evaluation text to obtain the target clustering result includes: selecting an arbitrary text from the target evaluation text to be determined as the current evaluation text, and calculating the text similarity between any text in the target evaluation text and the current evaluation text; determining the target probability corresponding to the target evaluation text when it is selected as the cluster center according to the text similarity; determining the initial cluster center corresponding to the target evaluation text under a preset number according to the target probability, and clustering the target evaluation text according to the initial cluster center to obtain an initial clustering result; determining the first text representation vector corresponding to the initial cluster center according to a text representation model, and determining the second text representation vector corresponding to each sub-text of each first sub-class cluster in the initial clustering result according to the text representation model; performing vector decomposition on the first text representation vector to obtain multiple first sub-vectors and performing vectorization on the second text representation vector; Decomposition is performed to obtain a plurality of second sub-vectors; the first sub-vector and the second sub-vector are processed for mean and standard deviation to obtain a first mean and a first standard deviation; each of the first sub-class clusters in the initial clustering result is counted to obtain the number of texts corresponding to the first sub-class cluster; the cluster weight corresponding to the first sub-class cluster is determined according to the number of texts, and the first membership probability corresponding to any one of the target evaluation texts under the first sub-class cluster is determined according to the first mean and the first standard deviation combined with a Gaussian model; the first mean and the first standard deviation are adjusted according to the cluster weight combined with the first membership probability to obtain a second mean and a second standard deviation; the second membership probability corresponding to any one of the target evaluation texts under the first sub-class cluster is determined according to the second mean and the second standard deviation combined with the Gaussian model; data clustering is performed on the target evaluation text according to the second membership probability to obtain the target clustering result.

[0087] Exemplarily, a text is randomly selected from the target evaluation text as the current evaluation text, and then the similarity between each remaining text in the target evaluation text and the current evaluation text is calculated according to a text similarity calculation method such as cosine similarity, edit distance, etc., so as to obtain text similarity.

[0088] Exemplarily, the target probability of the target evaluation text being selected as the cluster center is determined based on the calculated text similarity. The lower the similarity between the text and the current evaluation text, the greater the probability of being selected as the cluster center. Thus, according to the preset number of cluster centers, a corresponding number of texts are selected from low to high according to the target probability as the initial cluster centers.

[0089] Exemplarily, based on the selected initial cluster center, each text in the target evaluation text is assigned to the cluster where the initial cluster center most similar to it is located, so as to obtain the initial clustering result, and each cluster is a first subclass cluster.

[0090] Exemplarily, a text representation model such as a word vector model (Word2Vec, GloVe) or a pre-trained language model (BERT, RoBERTa) is used to convert the initial cluster center into a first text representation vector, and each sub-text in each first sub-class cluster in the initial clustering result is converted into a second text representation vector.

[0091] Exemplarily, a vector decomposition operation is performed on the first text representation vector and the second text representation vector, such as using a principal component analysis method, and then the first text representation vector is decomposed to obtain first sub-vectors corresponding to multiple dimensions and the second text representation vector is decomposed to obtain multiple second sub-vectors corresponding to multiple dimensions, so as to calculate the means and standard deviations of the first sub-vectors and the second sub-vectors in different dimensions, so as to obtain the first means and first standard deviations in different dimensions. The first mean reflects the average level of the first sub-vector and the second sub-vector in the relevant dimension, and the first standard deviation reflects the degree of discreteness between the first sub-vector and the second sub-vector in the relevant dimension.

[0092] Exemplarily, the number of texts contained in each first sub-cluster in the initial clustering result is counted. Thus, the cluster weight of each first sub-cluster is determined based on the number of texts obtained by counting. Generally, the more texts a sub-cluster has, the greater its cluster weight. The cluster weight can be obtained by dividing the number of texts in each sub-cluster by the total number of texts in all sub-clusters.

[0093] Exemplarily, a corresponding target model is constructed according to the first mean and the first standard deviation in combination with a Gaussian model, so that the second sub-vector corresponding to the target evaluation text is substituted into the target model to obtain the first membership probability of any text in the target evaluation text under each first subclass cluster.

[0094] Exemplarily, after obtaining the first membership probability of any text in the target evaluation text under each first subclass cluster, weighted sum normalization is performed in combination with the cluster weight to obtain the target membership probability corresponding to any text in the target evaluation text under each first subclass cluster, so that the target membership probability is combined with the second subvector to adjust the first standard deviation or the first mean to obtain the second standard deviation or the second mean. For example, the second standard deviation or the second mean is obtained according to the following formula:

[0095] ;

[0096] ;

[0097] in, represents the second mean corresponding to the i-th subclass cluster under the w-th dimension, Indicates the number of texts corresponding to the i-th sub-cluster, represents the target membership probability corresponding to the kth text of the ith sub-cluster under the wth dimension, represents the second sub-vector corresponding to the k-th text of the ith sub-cluster under the w-th dimension, represents the second standard deviation of the ith subclass cluster under the wth dimension, Represents the average of the second means corresponding to the i-th subclass cluster under all w-th dimensions.

[0098] Exemplarily, the target model is reconstructed according to the second mean and the second standard deviation in combination with the Gaussian model, and then the text vector corresponding to any text in the target evaluation text under the target dimension is substituted into the target model corresponding to each subclass cluster, so as to obtain the second membership probability corresponding to any text in the target evaluation text under each first subclass cluster under the target dimension, and then the final membership probability between each text in the target evaluation text and the first subclass cluster under all target dimensions is calculated according to the calculated second membership probability, and then each text in the target evaluation text is reallocated to the first subclass cluster with the largest final membership probability, so as to complete the final data clustering and obtain the target clustering result.

[0099] Specifically, in the clustering process, by processing text vectors and calculating membership probabilities, similar texts can be clustered together more accurately, thereby improving the quality and reliability of the data.

[0100] In some embodiments, the calculation of the text similarity between any one of the target evaluation texts and the current evaluation text includes: determining that any one of the target evaluation texts is a text to be processed, and performing keyword recognition on the text to be processed to obtain a first keyword; performing keyword recognition on the current evaluation text to obtain a second keyword; converting the first keyword into a target space to obtain a first feature representation corresponding to the target dimension, and converting the second keyword into the target space to obtain a second feature representation corresponding to the target dimension; determining first distribution information corresponding to the first keyword in the target space according to the first feature representation, and determining a first entropy value corresponding to the first keyword in the target space according to the first feature representation; determining second distribution information corresponding to the second keyword in the target space according to the second feature representation, and determining a second entropy value corresponding to the second keyword in the target space according to the second feature representation; determining a first degree of separation corresponding to the first keyword in the target dimension according to the first distribution information and the first entropy value, the first degree of separation being used to characterize the first The expression strength of the keyword under the text to be processed; determining the second separation degree corresponding to the second keyword under the target dimension according to the second distribution information and the second entropy value, the second separation degree is used to characterize the expression strength of the second keyword under the current evaluation text; determining the first dimension weight corresponding to the first keyword under the target dimension, and determining the first frequency of the first keyword appearing under the first similar text cluster corresponding to the current evaluation text; determining the first similarity between the first keyword and the current evaluation text according to the first dimension weight, the first frequency and the first separation degree; determining the second dimension weight corresponding to the second keyword under the target dimension, and determining the second frequency of the second keyword appearing under the second similar text cluster corresponding to the text to be processed; determining the second similarity between the second keyword and the text to be processed according to the second dimension weight, the second frequency and the second separation degree; fusing the first similarity and the second similarity to determine the text similarity between the text to be processed and the current evaluation text; wherein the first similarity is obtained according to the following formula:

[0101] ;

[0102] in, represents the kth first keyword of the i-th text to be processed The jth current evaluation text The first similarity between them, count represents the number corresponding to the target dimension, represents the first frequency at which the kth first keyword of the i-th text to be processed appears in the first similar text cluster corresponding to the j-th current evaluation text, and represents a constant, Indicates the i-th text to be processed The corresponding text length, Represents the average text length corresponding to the first similar text cluster corresponding to the jth current evaluation text.

[0103] For example, a text is selected from the target evaluation text as the text to be processed, and a keyword extraction algorithm (such as TF-IDF, TextRank, etc.) is used to process it, and important words therein are identified to obtain the first keyword. The current evaluation text is also processed using the above keyword extraction algorithm to obtain the second keyword.

[0104] Exemplarily, a target space is selected, which may be a pre-trained word vector space, and then the first keyword is mapped to the target space to obtain the first feature representation corresponding to the target dimension of the target space, which describes the position of the first keyword in the target space in the form of a vector. In the same way, the second keyword is converted to the target space to obtain the second feature representation corresponding to the target dimension.

[0105] Exemplarily, according to the first feature representation, the distribution of the first keyword in the target space is analyzed to obtain the first distribution information, which may be the probability distribution of the keyword in each dimension, and then the first entropy value of the first keyword in the target space is calculated based on the first feature representation. The entropy value can measure the uncertainty of the keyword distribution. The larger the entropy value, the more dispersed the keyword distribution. In the same way, the second distribution information and the second entropy value of the second keyword in the target space are determined according to the second feature representation.

[0106] Exemplarily, the first distribution information and the first entropy value are combined to calculate the first separation degree of the first keyword under the target dimension. The first separation degree can reflect the uniqueness and expression strength of the first keyword in the text to be processed. The higher the separation degree, the more important the keyword is in the text to be processed. And based on the second distribution information and the second entropy value, the second separation degree of the second keyword under the target dimension is calculated to characterize the expression strength of the second keyword under the current evaluation text.

[0107] Exemplarily, the first dimension weight under the target dimension is determined for the first keyword. The weight can be determined based on factors such as the importance of the keyword in the target space and its contribution to text classification, and then the first frequency of the first keyword appearing in the first similar text cluster corresponding to the current evaluation text is counted. The first similar text cluster can be a collection of texts similar to the current evaluation text obtained from the database through a clustering algorithm. Similarly, the second dimension weight under the target dimension is determined for the second keyword, and its second frequency of appearing in the second similar text cluster corresponding to the text to be processed is counted.

[0108] Exemplarily, the first dimension weight, the first frequency and the first separation degree are combined according to the following formula to calculate the first similarity between the first keyword and the current evaluation text:

[0109] ;

[0110] in, Indicates the kth first keyword of the i-th text to be processed With the jth current evaluation text The first similarity between them, count indicates the number of corresponding target dimensions, represents the first frequency at which the kth first keyword of the i-th to-be-processed text appears in the first similar text cluster corresponding to the j-th current evaluation text, and represents a constant, Indicates the i-th text to be processed The corresponding text length, Indicates the average text length of the first similar text cluster corresponding to the jth current evaluation text.

[0111] Exemplarily, the second similarity between the second keyword and the text to be processed is calculated in the same manner according to the second dimension weight, the second frequency and the second separation degree.

[0112] Exemplarily, after obtaining the first similarity between the first keyword and the current evaluation text according to the above method, the first similarity between each first keyword and the current evaluation text is obtained according to the above method, and then all the first similarities are summed and averaged to obtain the first total similarity between the text to be processed and the current evaluation text, and then the second similarity between the second keyword and the text to be processed is obtained according to the above method, and then the second similarity between each second keyword and the text to be processed is obtained according to the above method, and then all the second similarities are summed and averaged to obtain the second total similarity between the text to be processed and the current evaluation text, and then the first total similarity and the second total similarity are averaged to obtain the text similarity between the text to be processed and the current evaluation text.

[0113] Specifically, the above method comprehensively considers multiple factors such as keyword distribution information, entropy, separation, dimension weight and frequency. This enables the calculated text similarity to more accurately reflect the semantic relevance between texts, reduce misjudgments caused by words that are superficially similar but semantically different, and thus improve the reliability and accuracy of text similarity.

[0114] In some embodiments, the data correction of the associated scoring information according to the target clustering result to obtain the target scoring information corresponding to the target evaluation text includes: obtaining the corresponding relevant scoring information under each second subclass cluster in the target clustering result from the associated scoring information; performing abnormal data identification on the relevant scoring information to obtain abnormal scoring information and the related evaluation text corresponding to the abnormal scoring information; removing the abnormal scoring information from the relevant scoring information to obtain the remaining scoring information; and correcting the score of the relevant evaluation text according to the remaining scoring information to obtain the target scoring information.

[0115] Exemplarily, according to the target clustering result obtained above, each second subclass cluster in the target clustering result is associated with the associated scoring information. The associated scoring information corresponding to each second subclass cluster is found out from the associated scoring information. This means that the scoring data associated with each specific cluster is extracted separately so as to be processed for each subclass cluster later.

[0116] Exemplarily, the extracted relevant rating information is analyzed using an abnormal data identification method, for example, a machine learning-based method such as the isolation forest algorithm is used, which can identify isolated abnormal points in the data set that are different from most data patterns, thereby obtaining relevant evaluation text corresponding to the abnormal rating information.

[0117] Exemplarily, the identified abnormal rating information is removed from the relevant rating information to obtain the remaining rating information. This step is to ensure that the data used to correct the rating is relatively reliable and conforms to the overall pattern, and to avoid the adverse impact of abnormal data on the final rating result. Based on the mean, median and other statistics of the remaining rating information, it is used as the corresponding rating result after the relevant evaluation text in the sub-class cluster is modified, thereby obtaining the corrected target rating information.

[0118] Specifically, the reliability and stability of the scoring data are improved by eliminating abnormal scoring information, so that the corrected target scoring information is more in line with the actual situation.

[0119] Step S107: Determine the merchant value assessment result corresponding to the target merchant using an artificial intelligence model based on the historical consumption information and the target evaluation text combined with the target rating information.

[0120] Exemplarily, the consumption information, evaluation text and rating information corresponding to the historical merchants are obtained from the database, so that the consumption information features, evaluation text vectors and rating information are fused to combine the features from different sources into a complete feature vector. It is also possible to perform weighted fusion according to the importance of the features, highlight the role of certain key features, and then use artificial intelligence models, such as machine learning models (such as decision trees, random forests, support vector machines, etc.) or deep learning models (such as multi-layer perceptrons, long short-term memory networks (LSTM), etc.) to use the fused feature vector as input and the known merchant value assessment results as output to train the selected model. During the training process, the parameters of the model are adjusted to minimize the error between the predicted results and the actual results to optimize the performance of the model.

[0121] Exemplarily, the historical consumption information and target evaluation text corresponding to the target merchant are combined with the target rating information to perform feature fusion to obtain a fused feature vector, and then the fused feature vector of the target merchant is input into the trained artificial intelligence model. The model calculates according to the learned patterns and rules, and outputs the merchant value evaluation result corresponding to the target merchant. This result can be a specific numerical value or a classification label (such as high value, medium value, low value, etc.). Therefore, the target merchant can clearly understand its own evaluation in the market through the merchant value evaluation results, and then optimize and improve itself according to the merchant value evaluation results.

[0122] See also Figure 2 , Figure 2An artificial intelligence-based merchant comprehensive value assessment system 200 is provided for the embodiment of the present application. The artificial intelligence-based merchant comprehensive value assessment system 200 includes a data acquisition module 201, a sentiment analysis module 202, a state determination module 203, an association calculation module 204, a data association module 205, a data correction module 206, and a value assessment module 207, wherein the data acquisition module 201 is used to obtain the historical consumption information corresponding to the target user after consumption at the target merchant, and obtain the initial evaluation text corresponding to the target merchant under the historical consumption information and the initial scoring information corresponding to the initial evaluation text; the sentiment analysis module 202 is used to perform text sentiment consistency analysis on the initial evaluation text to obtain the target analysis result corresponding to the initial evaluation text; the state determination module 203 is used to calculate the target analysis result according to the target analysis result combined with the initial scoring information The target data state corresponding to the initial evaluation text is determined according to the information; an association calculation module 204 is used to calculate the information association degree corresponding to the historical consumption information and the initial evaluation text; a data association module 205 is used to delete the initial evaluation text according to the target data state and the information association degree to obtain a target evaluation text, and obtain the associated scoring information corresponding to the target evaluation text from the initial scoring information; a data correction module 206 is used to perform data clustering on the target evaluation text to obtain a target clustering result, and perform data correction on the associated scoring information according to the target clustering result to obtain the target scoring information corresponding to the target evaluation text; a value assessment module 207 is used to determine the merchant value assessment result corresponding to the target merchant by using an artificial intelligence model according to the historical consumption information and the target evaluation text in combination with the target scoring information.

[0123] In some implementations, the artificial intelligence-based merchant comprehensive value assessment system 200 may be applied to a terminal device.

[0124] It should be noted that technical personnel in the relevant field can clearly understand that, for the convenience and conciseness of description, the specific working process of the artificial intelligence-based merchant comprehensive value assessment system 200 described above can refer to the corresponding process in the aforementioned artificial intelligence-based merchant comprehensive value assessment method embodiment, and will not be repeated here.

[0125] See also Figure 3 , Figure 3 A schematic block diagram of the structure of a terminal device provided in an embodiment of the present invention.

[0126] like Figure 3 As shown, the terminal device 300 includes a processor 301 and a memory 302 , and the processor 301 and the memory 302 are connected via a bus 303 , such as an I2C (Inter-integrated Circuit) bus.

[0127] Specifically, the processor 301 is used to provide computing and control capabilities to support the operation of the entire terminal device. The processor 301 can be a central processing unit (CPU), and the processor 301 can also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0128] Specifically, the memory 302 may be a Flash chip, a read-only memory (ROM) disk, an optical disk, a USB flash drive, or a mobile hard disk.

[0129] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a partial structure related to the embodiment of the present invention, and does not constitute a limitation on the terminal device to which the embodiment of the present invention is applied. The specific server may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0130] The processor is used to run a computer program stored in the memory, and implement any one of the artificial intelligence-based merchant comprehensive value assessment methods provided by the embodiments of the present invention when executing the computer program.

[0131] In one embodiment, the processor is used to run a computer program stored in the memory, and implements the following steps when executing the computer program:

[0132] Obtaining historical consumption information corresponding to the target user's consumption at the target merchant, and obtaining an initial evaluation text corresponding to the target merchant under the historical consumption information and initial rating information corresponding to the initial evaluation text;

[0133] Performing text sentiment consistency analysis on the initial evaluation text to obtain a target analysis result corresponding to the initial evaluation text;

[0134] Determine the target data state corresponding to the initial evaluation text according to the target analysis result combined with the initial scoring information;

[0135] Calculating the information correlation between the historical consumption information and the initial evaluation text;

[0136] Deleting the initial evaluation text according to the target data state and the information relevance to obtain a target evaluation text, and obtaining the associated scoring information corresponding to the target evaluation text from the initial scoring information;

[0137] Performing data clustering on the target evaluation text to obtain a target clustering result, and performing data correction on the associated scoring information according to the target clustering result to obtain target scoring information corresponding to the target evaluation text;

[0138] The merchant value assessment result corresponding to the target merchant is determined using an artificial intelligence model based on the historical consumption information and the target evaluation text combined with the target rating information.

[0139] It should be noted that technical personnel in the relevant field can clearly understand that, for the convenience and brevity of description, the specific working process of the terminal device described above can refer to the corresponding process in the aforementioned embodiment of the merchant comprehensive value assessment method based on artificial intelligence, and will not be repeated here.

[0140] An embodiment of the present invention also provides a storage medium for computer-readable storage, wherein the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of any one of the artificial intelligence-based merchant comprehensive value assessment methods provided in the description of the embodiment of the present invention.

[0141] The storage medium may be an internal storage unit of the terminal device in the aforementioned embodiment, such as a hard disk or memory of the terminal device. The storage medium may also be an external storage device of the terminal device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the terminal device.

[0142] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware embodiment, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a microprocessor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0143] It should be understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations. It should be noted that, in this article, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "including a..." does not exclude the presence of other identical elements in the process, method, article or system including the element.

[0144] The serial numbers of the embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A merchant comprehensive value assessment method based on artificial intelligence, characterized in that: The method comprises: Obtaining historical consumption information corresponding to the target user's consumption at the target merchant, and obtaining an initial evaluation text corresponding to the target merchant under the historical consumption information and initial rating information corresponding to the initial evaluation text; Performing text sentiment consistency analysis on the initial evaluation text to obtain a target analysis result corresponding to the initial evaluation text; Determine the target data state corresponding to the initial evaluation text according to the target analysis result combined with the initial scoring information; Calculating the information correlation between the historical consumption information and the initial evaluation text; Deleting the initial evaluation text according to the target data state and the information relevance to obtain a target evaluation text, and obtaining the associated scoring information corresponding to the target evaluation text from the initial scoring information; Performing data clustering on the target evaluation text to obtain a target clustering result, and performing data correction on the associated scoring information according to the target clustering result to obtain target scoring information corresponding to the target evaluation text; The merchant value assessment result corresponding to the target merchant is determined using an artificial intelligence model based on the historical consumption information and the target evaluation text combined with the target rating information.

2. The method according to claim 1, characterized in that The performing text sentiment consistency analysis on the initial evaluation text to obtain a target analysis result corresponding to the initial evaluation text includes: Performing text segmentation on the initial evaluation text to obtain a plurality of segmented evaluation texts, and determining a first evaluation text from the plurality of segmented evaluation texts and a first remaining evaluation text obtained by removing the first evaluation text from the plurality of segmented evaluation texts; Performing sentiment classification on the first evaluation text to obtain a first sentiment type corresponding to the first evaluation text and first probability information corresponding to the first sentiment type; Merging any second evaluation text in the first remaining evaluation text with the first evaluation text and then performing sentiment analysis to obtain a second sentiment type corresponding to the second evaluation text and second probability information corresponding to the second sentiment type; Comparing the first emotion type and the second emotion type to obtain a type comparison result, and when the type comparison result is that the first emotion type and the second emotion type are the same, determining a first analysis result corresponding to the first evaluation text and the second evaluation text according to the first probability information and the second probability information, wherein the first analysis result is used to characterize an analysis result of emotion consistency between the first evaluation text and the second evaluation text; When the type comparison result is that the first emotion type and the second emotion type are different, the first analysis result corresponding to the first evaluation text and the second evaluation text is determined as a preset result; Merging the plurality of segmented evaluation texts according to the first analysis result to obtain a target merging result; The target analysis result corresponding to the initial evaluation text is obtained by performing text sentiment consistency analysis on the initial evaluation text according to the target merging result.

3. The method according to claim 2, characterized in that The step of performing text merging on the plurality of segmented evaluation texts according to the first analysis result to obtain a target merging result includes: Obtaining a first consistent text corresponding to the first evaluation text from the second evaluation text according to the first analysis result; Eliminate the first consistent text from the first remaining evaluation text to obtain a second remaining evaluation text; Performing sentiment classification on the first consistent text to obtain a third sentiment type corresponding to the first consistent text and third probability information corresponding to the third sentiment type; Performing sentiment analysis on any third evaluation text in the second remaining evaluation text and the first consistent text after merging to obtain a fourth sentiment type corresponding to the third evaluation text and fourth probability information corresponding to the fourth sentiment type; Determine a second analysis result corresponding to the first consistency text and the third evaluation text according to the third emotion type, the third probability information, the fourth emotion type and the fourth probability information; Determine a fourth evaluation text from the second remaining evaluation text, and obtain a third analysis result corresponding to the first evaluation text and the fourth evaluation text from the first analysis result, and obtain a fourth analysis result corresponding to the first consistent text and the fourth evaluation text from the second analysis result; When the third analysis result is consistent with the fourth analysis result, the first consistency text, the first evaluation text and the fourth evaluation text are merged according to the third analysis result to obtain a first merged result; When the third analysis result and the fourth analysis result are inconsistent, the first consistency text and the first evaluation text are merged to obtain a first merged text, and the sentiment consistency between the first merged text and the fourth evaluation text is calculated to obtain a fifth analysis result corresponding to the first merged text; Merging the first consistency text, the first evaluation text, and the fourth evaluation text according to the fifth analysis result to obtain the first merged result; Eliminating the first merging result from the initial evaluation text to obtain other evaluation texts, and performing text merging based on the other evaluation texts to obtain a second merging result; The target merging result corresponding to the plurality of segmented evaluation texts is determined according to the first merging result and the second merging result.

4. The method according to claim 3, characterized in that The step of determining the target data state corresponding to the initial evaluation text according to the target analysis result and the initial scoring information includes: When the target analysis result is the preset result, first position distribution information corresponding to the first merged result is determined according to the initial evaluation text, and second position distribution information corresponding to the second merged result is determined according to the initial evaluation text; Determine, according to the first position distribution information and the second position distribution information, a core evaluation text corresponding to the initial evaluation text and a target intention corresponding to the core evaluation text of the target user; The scoring interval corresponding to the initial evaluation text is determined according to the target intent and the mapping table, and the target data state corresponding to the initial evaluation text is determined according to the scoring interval and the initial scoring information.

5. The method according to claim 1, characterized in that The calculating the information correlation between the historical consumption information and the initial evaluation text includes: Performing part-of-speech analysis on the initial evaluation text to obtain relevant nouns and relevant evaluation words corresponding to relevant products in the initial evaluation text; Determining a preset evaluation word corresponding to the relevant noun, and calculating a correlation between the relevant noun and the relevant evaluation word according to the preset evaluation word; Determining the association relationship between the related noun and the related evaluation word according to the association degree; Screening the relevant evaluation words according to the association relationship to obtain remaining evaluation words, and determining, according to the remaining evaluation words, associated products that can be used to evaluate the target merchant; Obtaining a target consumption product corresponding to the target user from the historical consumption information; Calculating the degree of overlap according to the target consumer product, the related nouns and the associated products to obtain a target degree of overlap; The information association degree corresponding to the historical consumption information and the initial evaluation text is determined according to the target coincidence degree.

6. The method according to claim 1, characterized in that The step of performing data clustering on the target evaluation text to obtain a target clustering result includes: Selecting an arbitrary text from the target evaluation text to be determined as the current evaluation text, and calculating the text similarity between an arbitrary text in the target evaluation text and the current evaluation text; Determining the target probability corresponding to when the target evaluation text is selected as the cluster center according to the text similarity; Determining the initial clustering centers corresponding to the target evaluation texts under a preset quantity according to the target probability, and clustering the target evaluation texts according to the initial clustering centers to obtain initial clustering results; Determine a first text representation vector corresponding to the initial cluster center according to the text representation model, and determine a second text representation vector corresponding to each subtext of each first subclass cluster in the initial clustering result according to the text representation model; Performing vector decomposition on the first text representation vector to obtain a plurality of first sub-vectors and performing vector decomposition on the second text representation vector to obtain a plurality of second sub-vectors; Performing mean and standard deviation processing on the first sub-vector and the second sub-vector to obtain a first mean and a first standard deviation; Performing quantity statistics on each of the first sub-class clusters in the initial clustering result to obtain the quantity of texts corresponding to the first sub-class cluster; Determine a cluster weight corresponding to the first sub-cluster according to the number of texts, and determine a first membership probability corresponding to any one of the target evaluation texts under the first sub-cluster according to the first mean and the first standard deviation combined with a Gaussian model; Adjusting the first mean and the first standard deviation according to the cluster weight combined with the first membership probability to obtain a second mean and a second standard deviation; Determine a second membership probability corresponding to any one of the target evaluation texts under the first subclass cluster according to the second mean and the second standard deviation in combination with the Gaussian model; Data clustering is performed on the target evaluation text according to the second membership probability to obtain the target clustering result.

7. The method according to claim 6, characterized in that The calculating the text similarity between any one of the target evaluation texts and the current evaluation text comprises: Determine any one of the target evaluation texts as a text to be processed, and perform keyword recognition on the text to be processed to obtain a first keyword; Performing keyword recognition on the current evaluation text to obtain a second keyword; Converting the first keyword into a target space to obtain a first feature representation corresponding to a target dimension, and converting the second keyword into the target space to obtain a second feature representation corresponding to the target dimension; Determine first distribution information corresponding to the first keyword in the target space according to the first feature representation, and determine a first entropy value corresponding to the first keyword in the target space according to the first feature representation; Determine second distribution information corresponding to the second keyword in the target space according to the second feature representation, and determine a second entropy value corresponding to the second keyword in the target space according to the second feature representation; Determine a first degree of separation corresponding to the first keyword in the target dimension according to the first distribution information and the first entropy value, where the first degree of separation is used to characterize the expression strength of the first keyword in the text to be processed; Determine a second degree of separation corresponding to the second keyword in the target dimension according to the second distribution information and the second entropy value, where the second degree of separation is used to characterize the expression strength of the second keyword in the current evaluation text; Determine a first dimension weight corresponding to the first keyword under the target dimension, and determine a first frequency of occurrence of the first keyword under a first similar text cluster corresponding to the current evaluation text; Determine a first similarity between the first keyword and the current evaluation text according to the first dimension weight, the first frequency and the first separation degree; Determine a second dimension weight corresponding to the second keyword under the target dimension, and determine a second frequency at which the second keyword appears under a second similar text cluster corresponding to the text to be processed; Determine a second similarity between the second keyword and the text to be processed according to the second dimension weight, the second frequency and the second separation degree; Fusing the first similarity and the second similarity to determine the text similarity between the to-be-processed text and the current evaluation text; The first similarity is obtained according to the following formula: ； in, represents the kth first keyword of the i-th text to be processed The jth current evaluation text The first similarity between them, count represents the number corresponding to the target dimension, represents the first frequency at which the kth first keyword of the i-th text to be processed appears in the first similar text cluster corresponding to the j-th current evaluation text, and represents a constant, Indicates the i-th text to be processed The corresponding text length, Represents the average text length corresponding to the first similar text cluster corresponding to the jth current evaluation text.

8. The method according to claim 1, characterized in that The step of performing data correction on the associated scoring information according to the target clustering result to obtain the target scoring information corresponding to the target evaluation text includes: Obtaining relevant score information corresponding to each second subclass cluster in the target clustering result from the associated score information; Performing abnormal data identification on the relevant scoring information to obtain abnormal scoring information and relevant evaluation text corresponding to the abnormal scoring information; Eliminating the abnormal scoring information from the relevant scoring information to obtain remaining scoring information; The score of the relevant evaluation text is modified according to the remaining score information to obtain the target score information.

9. A merchant comprehensive value assessment system based on artificial intelligence, characterized in that: include: A data acquisition module is used to obtain the historical consumption information corresponding to the target user's consumption at the target merchant, and obtain the initial evaluation text corresponding to the target merchant under the historical consumption information and the initial rating information corresponding to the initial evaluation text; A sentiment analysis module, used to perform text sentiment consistency analysis on the initial evaluation text to obtain a target analysis result corresponding to the initial evaluation text; A state determination module, used to determine the target data state corresponding to the initial evaluation text according to the target analysis result combined with the initial scoring information; A correlation calculation module, used to calculate the information correlation degree between the historical consumption information and the initial evaluation text; A data association module, configured to delete the initial evaluation text according to the target data state and the information association degree to obtain a target evaluation text, and obtain associated scoring information corresponding to the target evaluation text from the initial scoring information; A data correction module, used for performing data clustering on the target evaluation text to obtain a target clustering result, and performing data correction on the associated scoring information according to the target clustering result to obtain target scoring information corresponding to the target evaluation text; The value assessment module is used to determine the merchant value assessment result corresponding to the target merchant by using an artificial intelligence model based on the historical consumption information and the target evaluation text combined with the target scoring information.

10. A terminal device, characterized in that: The terminal device includes a processor and a memory; The memory is used to store computer programs; The processor is used to execute the computer program and implement the merchant comprehensive value assessment method based on artificial intelligence as described in any one of claims 1 to 8 when executing the computer program.

Images