E-commerce platform user feedback analysis method and system based on artificial intelligence

Abstract

The invention relates to the technical field of e-commerce data analysis, and discloses an e-commerce platform user feedback analysis method and system based on artificial intelligence. The method comprises the following steps: acquiring text data and user behavior data of a user; using a text vectorization model to convert the text data into vectorized data; calculating a user behavior tag by using a semantic analysis model according to the vectorized data and the user behavior data; calculating a user behavior vector according to the user behavior tag; calculating an association point set according to the user behavior vector; calculating a user keyword according to the user behavior vector and the user behavior tag; and according to the association point set and the user keyword, obtaining user demand data and outputting the user demand data. The method has the following effects: the related data of the user is analyzed and processed through the artificial intelligence model, and the user data is converted into mathematical data for calculation, so that the demand of each user can be accurately and efficiently obtained, and misjudgment on the user demand is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of e-commerce data analysis, and in particular to an e-commerce platform user feedback analysis method and system based on artificial intelligence. Background Art

[0002] With the rapid development of the Internet and big data information, in daily life and work, many users will place their first choice for shopping on e-commerce platforms. E-commerce platforms will also make certain adjustments to products and services based on user feedback to better meet user consumption needs in the hope of increasing the number of platform users. Therefore, in order to accurately analyze user needs, the use of artificial intelligence technology to analyze user consumption needs has gradually become the mainstream.

[0003] However, there are still some defects when using artificial intelligence technology for analysis, which makes it impossible to accurately analyze user needs. Some methods only locate user problems based on vectorized data, which may lead to missing information that is not reflected in the user's text data, making it impossible to comprehensively locate problems and mine needs based on user feedback.

[0004] In one prior art, an artificial intelligence-based e-commerce platform user feedback analysis method and system are disclosed, which mainly include obtaining product feedback information corresponding to a target user; extracting product feedback text and platform feedback text in the product feedback information; calculating a significant eccentricity coefficient based on product feedback description knowledge corresponding to the product feedback text to obtain a significant eccentricity coefficient of product feedback, and adjusting the product feedback description knowledge based on the significant eccentricity coefficient of product feedback to obtain significant product feedback description knowledge; calculating a significant eccentricity coefficient based on platform feedback description knowledge corresponding to the platform feedback text to obtain a significant eccentricity coefficient of platform feedback, and adjusting the platform feedback description knowledge based on the significant eccentricity coefficient of platform feedback to obtain significant platform feedback description knowledge; performing knowledge interaction between the significant platform feedback description knowledge and the significant product feedback description knowledge to obtain fused description knowledge, performing user feedback type analysis based on the fused description knowledge to obtain a user feedback type analysis result corresponding to the product feedback information.

[0005] However, in the above-mentioned prior art, when performing calculations, the parts of speech of some words in the feedback text data may not be applicable to the algorithm, which will result in large errors in the vectorized data, thereby affecting the accuracy of subsequent analysis results. Therefore, it is not possible to comprehensively locate problems and explore needs based on user feedback. Summary of the invention

[0006] The present invention provides an e-commerce platform user feedback analysis method, system, electronic device and storage medium based on artificial intelligence to solve the problem of inaccurate analysis of user data.

[0007] In a first aspect, in order to solve the above technical problems, the present invention provides an e-commerce platform user feedback analysis method based on artificial intelligence, comprising:

[0008] Acquire text data of user feedback, user behavior data and an analysis model; wherein the analysis model includes a text vectorization model and a semantic analysis model;

[0009] Using the text vectorization model, converting the text data into vectorized data;

[0010] According to the vectorized data and the user behavior data, using the semantic analysis model, calculate the user behavior label;

[0011] Calculate a user behavior vector according to the user behavior label;

[0012] Calculate a set of associated points according to the user behavior vector;

[0013] Calculating a user keyword according to the user behavior vector and the user behavior label;

[0014] According to the association point set and the user keyword, user demand data is obtained and output.

[0015] In an optional implementation, the using the text vectorization model to convert the text data into vectorized data includes:

[0016] Performing data cleaning on the text data to obtain complete text data;

[0017] Substitute the complete text data into the text vectorization model to obtain the vectorized data.

[0018] In an optional implementation, the calculating the user behavior label based on the vectorized data and the user behavior data using the semantic analysis model includes:

[0019] Inputting the vectorized data and the user behavior data into the semantic analysis model to obtain an initial behavior label;

[0020] Modifying the initial behavior label to obtain the user behavior label by calculation;

[0021] The user behavior label is calculated according to the following formula:

[0022]

[0023] Among them, L′ i represents the user behavior label of the i-th user, L i represents the initial behavior label of the i-th user, N represents the total number of user behaviors, and w ij represents the influence weight of the jth user behavior on the ith user label, a ij represents the first coefficient of the i-th user label under the j-th user behavior, b ij represents the second coefficient of the i-th user label under the j-th user behavior, B j Indicates the behavior label correction value corresponding to the j-th user behavior.

[0024] In an optional implementation, calculating the user behavior vector according to the user behavior label includes:

[0025] The user behavior vector is calculated according to the following formula:

[0026]

[0027] Among them, u i represents the i-th user behavior vector, x ij represents the j-th user behavior label of the ith user, and D represents the number of behavior types.

[0028] In an optional implementation, the step of calculating a set of associated points according to the user behavior vector includes:

[0029] Perform association calculation according to the user behavior vector to obtain an initial association point set;

[0030] Modifying the initial set of associated points to obtain the set of associated points by calculation;

[0031] The initial set of associated points is calculated according to the following formula:

[0032] G={(v i , n ij )|i=1,2,…,K; j=1,2,…,M}

[0033] Where G represents the initial associated point set, v i represents the initial correlation coefficient, n ij represents the number of the jth behavior corresponding to the ith user behavior vector, K represents the total number of the user behavior vectors, and M represents the total number of current behaviors;

[0034] The set of associated points is calculated according to the following formula:

[0035]

[0036] G′={(v i ′,n ij )|i=1,2,…,K; j=1,2,…,M}

[0037] Where G′ represents the set of associated points, v i ′ represents the modified correlation coefficient, α ij represents the first correction coefficient of the i-th user label under the j-th user behavior, β ij It represents the second correction coefficient of the i-th user label under the j-th user behavior.

[0038] In an optional implementation, calculating the user keyword according to the user behavior vector and the user behavior label includes:

[0039] Performing secondary correlation calculation on the user behavior vector to obtain a correlation matrix;

[0040] Mapping the user behavior label and the user behavior vector to obtain a behavior mapping matrix;

[0041] Calculating behavior frequency values ​​and weights corresponding to each behavior frequency value according to the association matrix and the behavior mapping matrix;

[0042] The weights are sorted from large to small, and the user behavior vectors corresponding to the first several proportions of weights in the sorting are selected and output as the user keywords.

[0043] In an optional implementation, performing secondary correlation calculation on the user behavior vector to obtain a correlation matrix includes:

[0044] The association matrix is ​​calculated according to the following formula:

[0045]

[0046] Where A represents the association matrix, u i represents the i-th user behavior vector, u N represents the Nth user behavior vector, where N represents the number of users.

[0047] In an optional implementation, mapping the user behavior label and the user behavior vector to obtain a behavior mapping matrix includes:

[0048] The behavior mapping matrix is ​​calculated according to the following formula:

[0049] R=[u1·L1 … u i ·Li … u N ·L N ]

[0050] Where R represents the behavior mapping matrix, u i represents the i-th user behavior vector, u N represents the Nth user behavior vector, L i represents the i-th user behavior label, L N Represents the Nth user behavior label.

[0051] In an optional implementation, the calculating, according to the association matrix and the behavior mapping matrix, the behavior frequency value and the weight corresponding to each behavior frequency value includes:

[0052] The behavior frequency value is calculated according to the following formula:

[0053]

[0054] Among them, F j represents the jth frequency value of the behavior, A i，j represents the element in the i-th row and j-th column of the association matrix, B j represents the j-th column element of the behavior mapping matrix, and D represents the number of behavior types;

[0055] The weight corresponding to each behavior frequency value is calculated according to the following formula:

[0056]

[0057] Among them, W j represents the weight corresponding to the j-th behavior frequency value, and N represents the number of users.

[0058] In a second aspect, the present invention provides an e-commerce platform user feedback analysis system based on artificial intelligence, comprising:

[0059] A data acquisition module, used to acquire text data of user feedback and user behavior data and an analysis model; wherein the analysis model includes a text vectorization model and a semantic analysis model;

[0060] A vectorized data conversion module, used to convert the text data into vectorized data using the text vectorization model;

[0061] A behavior label calculation module, used to calculate the user behavior label based on the vectorized data and the user behavior data using the semantic analysis model;

[0062] A behavior vector calculation module, used to calculate a user behavior vector according to the user behavior label;

[0063] A correlation point calculation module, used to calculate a correlation point set according to the user behavior vector;

[0064] A keyword calculation module, used to calculate a user keyword according to the user behavior vector and the user behavior label;

[0065] The demand data acquisition module is used to obtain and output user demand data according to the association point set and the user keyword.

[0066] In a third aspect, the present invention also provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements any one of the above-mentioned artificial intelligence-based e-commerce platform user feedback analysis methods.

[0067] In a fourth aspect, the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned artificial intelligence-based e-commerce platform user feedback analysis methods.

[0068] Compared with the prior art, the present invention has the following beneficial effects:

[0069] The present invention provides an e-commerce platform user feedback analysis method based on artificial intelligence, comprising obtaining text data of user feedback, user behavior data and an analysis model; wherein the analysis model comprises a text vectorization model and a semantic analysis model; using the text vectorization model, the text data is converted into vectorized data; according to the vectorized data and the user behavior data, the semantic analysis model is used to calculate a user behavior label; according to the user behavior label, a user behavior vector is calculated; according to the user behavior vector, a set of association points is calculated; according to the user behavior vector and the user behavior label, a user keyword is calculated; according to the association point set and the user keyword, user demand data is obtained and output.

[0070] The method analyzes and processes relevant user data through an artificial intelligence model, converts the user data into mathematical data for calculation, thereby accurately and efficiently obtaining the personalized needs of each user, reducing misjudgment of user needs, and enabling timely feedback and adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1It is a flowchart of an e-commerce platform user feedback analysis method based on artificial intelligence provided by the first embodiment of the present invention;

[0072] Figure 2 It is a schematic diagram of the structure of an e-commerce platform user feedback analysis system based on artificial intelligence provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0073] 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 only 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.

[0074] Reference Figure 1 The first embodiment of the present invention provides an e-commerce platform user feedback analysis method based on artificial intelligence, comprising the following steps:

[0075] S1, obtaining text data of user feedback, user behavior data and an analysis model; wherein the analysis model includes a text vectorization model and a semantic analysis model;

[0076] S2, using the text vectorization model to convert the text data into vectorized data;

[0077] S3, calculating a user behavior label using the semantic analysis model according to the vectorized data and the user behavior data;

[0078] S4, calculating a user behavior vector according to the user behavior label;

[0079] S5, calculating and obtaining a set of associated points according to the user behavior vector;

[0080] S6, calculating a user keyword according to the user behavior vector and the user behavior label;

[0081] S7, obtaining user demand data according to the association point set and the user keyword and outputting the data.

[0082] It should be noted that some existing AI-based user analysis methods rely heavily on vectorized data to locate user problems. However, their limitation is that they only locate user problems based on vectorized data, which may not be able to fully identify information not reflected in the user's text data, and thus cannot locate problems and mine needs based on user feedback in a more comprehensive and accurate manner; at the same time, when establishing platform labels, the final result is to establish labels based on user behavior, but it is not possible to make timely judgments on the different situations of labels that may be formed by each user behavior, and when locating user problems and analyzing user needs based on the established labels, it is not possible to locate users' problems more accurately, and it is also impossible to mine corresponding needs based on user feedback.

[0083] In this method, by selecting and using multiple artificial intelligence models, a large amount of relevant data generated by users in shopping activities is comprehensively analyzed and processed. In this process, first, the user data is accurately converted into a mathematical form so that it can be calculated in the artificial intelligence model. Then the obtained user data in mathematical form is input into a pre-set model for judgment, thereby obtaining user demand data. In this way, the method can parse out the unique personalized needs of each user with very high accuracy and efficiency. This method not only significantly reduces the misjudgment rate of user needs, but also ensures the accuracy of demand identification, thereby improving the pertinence and satisfaction of the service. At the same time, this method can monitor and capture small changes in user needs in real time, provide feedback and adjustments in a timely manner, and ensure that the services provided are always highly synchronized with the latest needs of users.

[0084] In step S1, it is necessary to obtain text data of user feedback, user behavior data and an analysis model established in advance according to the current project requirements; wherein the analysis model includes a text vectorization model and a semantic analysis model.

[0085] Specifically, firstly, the text data of user feedback and the recorded user behavior data are obtained from the relevant database, then the required analysis model is established according to the current project requirements, and finally these pre-established analysis models are obtained.

[0086] It should be noted that the text data of user feedback directly reflects the user's experience, opinions and suggestions on the product or service; the user behavior data records the user's browsing, purchasing, following, collecting, replying to comments, liking, sharing, specific operation paths, click frequency, length of stay and other information on the platform. In addition, when establishing an analysis model, it is necessary to select a suitable analysis model and adjust the parameters according to the needs of the current project. For example, common text vectorization models include word2vec algorithm, Doc2Vec algorithm, fastText algorithm, BERT algorithm, etc., each of which has different advantages and disadvantages, and the most suitable algorithm needs to be selected according to the specific application scenario.

[0087] In step S2, the text data is converted into vectorized data using the preset text vectorization model and saved.

[0088] In one embodiment, the text data is converted into vectorized data using the text vectorization model, including performing data cleaning on the text data to obtain complete text data; and substituting the complete text data into the text vectorization model to obtain the vectorized data.

[0089] Specifically, the text data is first cleaned, including removing irrelevant characters (such as punctuation marks, special symbols, etc.) in the text, correcting spelling errors, processing non-standard character sets, deleting duplicate or redundant information, etc., so as to ensure the integrity and correctness of the obtained text data. Then the processed text data is substituted into the text vectorization model for calculation to obtain vectorized data, which is finally stored.

[0090] It should be noted that these vectorized data usually have a higher dimension, so that more information can be stored and the final result can be more accurate. Therefore, the preset text vectorization model needs to be able to process high-dimensional vectors to ensure that there are no errors in the data format.

[0091] In step S3, the user behavior label is calculated based on the previously obtained vectorized data and the user behavior data using the semantic analysis model.

[0092] In one embodiment, the step of calculating the user behavior label based on the vectorized data and the user behavior data using the semantic analysis model comprises the following steps:

[0093] S31, inputting the vectorized data and the user behavior data into the semantic analysis model to obtain an initial behavior label;

[0094] S32, modifying the initial behavior label and calculating to obtain the user behavior label.

[0095] In step S31, the pre-processed vectorized data and the user behavior data are substituted into the semantic analysis model for calculation to obtain an initial behavior label.

[0096] It should be noted that the semantic analysis model can deeply understand and interpret the text to extract the semantic information contained therein, so as to accurately analyze the user's text input and extract the user's interests and needs from it. For different practical scenarios, a more suitable semantic analysis model can be selected according to actual needs. For example, for scenarios with high processing speed requirements, a bag of words model can be selected to quickly process large amounts of text; a graph neural network model can more accurately extract the user's interests and needs.

[0097] In step S32, the obtained initial behavior label is modified, and the user behavior label is obtained according to the formula.

[0098] In one embodiment, the user behavior tag is calculated according to the following formula:

[0099]

[0100] Where L′i represents the user behavior label of the i-th user, L i represents the initial behavior label of the i-th user, N represents the total number of user behaviors, and w ij represents the influence weight of the jth user behavior on the ith user label, a ij represents the first coefficient of the i-th user label under the j-th user behavior, b ij represents the second coefficient of the i-th user label under the j-th user behavior, B j Indicates the behavior label correction value corresponding to the j-th user behavior.

[0101] It should be noted that since the initial behavior label is generated by the vectorized data and the user behavior data, the amount of data is too much and the semantic analysis model is imperfect, the final result is easily affected by noise data, outliers or data deviations, resulting in inaccurate labels. Therefore, it is necessary to correct the initial behavior label. At the same time, in the correction process, it is necessary to consider the specific needs in different application scenarios, which means that it is necessary to consider how to select the specific values ​​of the impact weights and the two coefficients in the correction formula.

[0102] In step S4, a user behavior vector is obtained by using a formula based on the calculated and corrected user behavior label.

[0103] In one implementation, calculating the user behavior vector according to the user behavior tag includes calculating the user behavior vector according to the following formula:

[0104]

[0105] Among them, u i represents the i-th user behavior vector, x ij represents the j-th user behavior label of the ith user, and D represents the number of behavior types.

[0106] It should be noted that the user behavior vector is a method of expressing the quantified user behavior labels, which can convert multiple user behavior labels into a point in a high-dimensional space through the above formula. By calculating only one point, the memory space consumed in the calculation process can be reduced, the computing resources can be reduced, and the calculation speed can be appropriately accelerated. In addition, it is also possible to consider standardizing and normalizing the obtained user behavior vector to ensure the dimensional consistency and comparability between different features.

[0107] It should be noted that when selecting the dimension for creating the user behavior vector, an appropriate dimension size needs to be selected. Too high a dimension may increase the computational complexity and storage cost, while too low a dimension may lose key information. Therefore, it is necessary to select an appropriate dimension size based on the user's data and the capabilities of the device.

[0108] In step S5, the association points between the user behavior vectors are calculated based on the obtained user behavior vectors to form an association point set.

[0109] In one implementation, calculating a set of associated points according to the user behavior vector comprises the following steps:

[0110] S51, performing association calculation according to the user behavior vector to obtain an initial association point set;

[0111] S52, modifying the initial set of associated points, and obtaining the set of associated points by calculation.

[0112] In step S51, association calculation is performed according to the user behavior vector to obtain an initial association point set.

[0113] In one implementation, the initial set of associated points is calculated according to the following formula:

[0114] G={(v i , n ij )|i=1,2,…,K; j=1,2,…,M}

[0115] Where G represents the initial associated point set, v i represents the initial correlation coefficient, n ij represents the number of the j-th behavior corresponding to the ith user behavior vector, K represents the total number of the user behavior vectors, and M represents the total number of current behaviors.

[0116] It should be noted that the initial set of association points is composed of a number of association points, wherein the association point refers to the connection between two user behavior vectors and is quantitatively displayed in the form of numerical values. This connection specifically refers to the similarity of the data generated by two users when shopping, such as visiting the same website in the same time period, purchasing similar products, etc. By calculating this degree of association, the user's behavior patterns, interests and needs can be better analyzed.

[0117] In another embodiment, in order to reduce the storage space occupied, it is possible to first determine whether there is a correlation point between the two user behavior vectors. If the calculated result is greater than the set threshold, it can be determined that there is a correlation point, and then the correlation point is calculated; if the calculated result is less than the set threshold, it is determined that there is no correlation point, and then there is no need to calculate the current two user behavior vectors.

[0118] In step S52, the initial set of associated points obtained in the above steps is modified to obtain the set of associated points by calculation.

[0119] In one implementation, the set of associated points is calculated according to the following formula:

[0120]

[0121] G′={(v i ′,n ij )|i=1,2,…,K; j=1,2,…,M}

[0122] Where G′ represents the set of associated points, v i represents the initial correlation coefficient, v i ′ represents the modified correlation coefficient, α ij represents the first correction coefficient of the i-th user label under the j-th user behavior, β ij It represents the second correction coefficient of the i-th user label under the j-th user behavior.

[0123] It should be noted that the initial set of association points may contain some noise data or erroneous associations, which may be caused by errors in the data collection process, the randomness of user behavior or other factors. Through correction, these inaccurate data can be eliminated to improve the overall accuracy of the set of association points. In addition, the correction process can strengthen the robustness of the set of association points, making it more resistant to data fluctuations or outliers, so as to draw more accurate conclusions in subsequent analysis.

[0124] It should be noted that when performing the correction operation, it is necessary to select an appropriate correction coefficient according to the actual environment. At the same time, in order to reduce the amount of correction calculation, the initial association point set can be screened before the correction, and only the more useful initial association points are further corrected.

[0125] In step S6, the user keyword is calculated based on the user behavior vector and the user behavior label.

[0126] In one implementation, calculating the user keyword according to the user behavior vector and the user behavior label includes the following steps:

[0127] S61, performing secondary correlation calculation on the user behavior vector to obtain a correlation matrix;

[0128] S62, mapping the user behavior label and the user behavior vector to obtain a behavior mapping matrix;

[0129] S63, calculating the behavior frequency value and the weight corresponding to each behavior frequency value according to the association matrix and the behavior mapping matrix;

[0130] S64, sorting the weights from large to small, and selecting the user behavior vectors corresponding to the first several proportions of weights in the sorting, and outputting them as the user keywords.

[0131] In step S61, a secondary correlation calculation is performed on the previous user behavior vector to obtain a correlation matrix.

[0132] In one implementation, performing secondary correlation calculation on the user behavior vector to obtain a correlation matrix includes:

[0133] The association matrix is ​​calculated according to the following formula:

[0134]

[0135] Where A represents the association matrix, u i represents the i-th user behavior vector, u Nrepresents the Nth user behavior vector, where N represents the number of users.

[0136] It should be noted that the association matrix associates the user behavior vectors and clearly represents the relationship between user behaviors in the form of a matrix. This is similar to the association point set in step S5, both of which quantify the association relationship, but using a matrix for representation can make matrix calculation more convenient, thereby improving the calculation speed.

[0137] In step S62, a mapping operation is performed on the user behavior label and the user behavior vector to obtain a behavior mapping matrix.

[0138] In one implementation, mapping the user behavior label and the user behavior vector to obtain a behavior mapping matrix includes:

[0139] The behavior mapping matrix is ​​calculated according to the following formula:

[0140] R=[u1·L1 … u i ·L i … u N ·L N ]

[0141] Where R represents the behavior mapping matrix, u i represents the i-th user behavior vector, u N represents the Nth user behavior vector, L i represents the i-th user behavior label, L N Represents the Nth user behavior label.

[0142] It should be noted that mapping user behavior labels and user behavior vectors to obtain a behavior mapping matrix is ​​a process that combines abstract user behavior descriptions (labels) with specific behavior feature data (vectors). This mapping result can more intuitively understand the inherent characteristics of user behavior and provide a computational basis for subsequent user behavior analysis and prediction.

[0143] It should be noted that when constructing the behavior mapping matrix, missing values ​​or outliers need to be processed. In addition, in order to ensure the comparability of values ​​between different features, some data need to be standardized when necessary.

[0144] In step S63, the behavior frequency value and the weight corresponding to each behavior frequency value are calculated according to the association matrix and the behavior mapping matrix according to a formula.

[0145] In one embodiment, the behavior frequency value and the weight corresponding to each behavior frequency value are calculated according to the association matrix and the behavior mapping matrix, including calculating the behavior frequency value according to the following formula:

[0146]

[0147] Among them, F j represents the jth frequency value of the behavior, A i，j represents the element in the i-th row and j-th column of the association matrix, B j represents the jth column element of the behavior mapping matrix, and D represents the number of behavior types; the weight corresponding to each behavior frequency value is calculated according to the following formula:

[0148]

[0149] Among them, W j represents the weight corresponding to the j-th behavior frequency value, and N represents the number of users.

[0150] It should be noted that the behavior frequency value refers to the number of times a certain behavior is performed within a certain period of time. By counting the behavior frequency values, the approximate needs of users can be obtained. At the same time, calculating the weights of the behavior frequency values ​​can analyze the impact of different behaviors on user interests and needs, and provide data support for inferring the user's potential needs.

[0151] It should be noted that the formula for calculating weights above uses normalized frequency values ​​as weights, that is, the weight of each behavior is its frequency value divided by the sum of all behavior frequency values. This method assumes that all behaviors are equivalent and only assigns weights based on their frequency of occurrence. In a real environment, there are many other factors to consider, such as the importance of the behavior (some behaviors may be more important than others, so they should be given higher weights), the prevalence of the behavior (if a behavior is very common, it may not be as informative as a rare behavior, so the weight may be lower), and business goals (user engagement, sales), etc. Therefore, additional weight allocation coefficients can be added when performing actual calculations.

[0152] In step S64, the weights are sorted from large to small, and the user behavior vectors corresponding to the first several proportions of weights in the sorting are selected and output as the user keywords.

[0153] It should be noted that the calculated weights are arranged in descending order, so that those weights that have the greatest influence on the description of user behavior can be identified. After the sorting is completed, a certain proportion of weights in the top column of the sorting is selected. The selection of the proportion can be determined according to actual needs and analysis objectives. It reflects how much information is selected from the numerous user behaviors as the focus of attention. The user behavior vector corresponding to the selected weight is the key information that can represent the main behavior characteristics of the user.

[0154] In step S7, user demand data is obtained from the association point set according to the user keyword and output.

[0155] Specifically, based on the user's search history, browsing history, purchase behavior and other data reflected in the user keyword, data information corresponding to the current user keyword is found in the highly summarized association point set, integrated into user demand data and output.

[0156] It should be noted that the user keywords not only reflect the user's search history and detailed browsing history, but also reveal multi-dimensional data such as the user's purchase behavior. The association point set is formed by integrating and refining the screened data information to form a complete, clear and easy-to-understand user demand data set. This data set accurately reflects the user's real needs and provides support for personalized recommendations. By searching for the required data from the association point set based on the user keywords, the current user's needs and interests can be accurately obtained.

[0157] To facilitate understanding of the present invention, the working process of the present invention is described below using a relatively common scenario as an example.

[0158] In this embodiment, it is planned to introduce an e-commerce platform user feedback analysis system based on artificial intelligence. The system is designed to be able to analyze the user's text data and behavior data and use the analysis model to analyze them, so as to obtain accurate user demand data.

[0159] Step 1: The system obtains text data of user feedback, user behavior data, and preset text vectorization model and semantic analysis model.

[0160] Step 2: The system uses a text vectorization model to convert text data into vectorized data.

[0161] Step 3: The system uses a semantic analysis model to calculate the vectorized data and user behavior data to obtain user behavior labels.

[0162] Step 4: The system calculates the user behavior vector based on the user behavior label.

[0163] Step 5: The system calculates the set of associated points based on the user behavior vector.

[0164] Step 6: The system calculates the user keyword based on the user behavior vector and user behavior label.

[0165] Step 7: The system obtains user demand data based on the set of association points and user keywords and outputs it.

[0166] Through the above steps, the artificial intelligence-based e-commerce platform user feedback analysis system can use a preset model to analyze user-related data, thereby quickly and accurately obtaining user demand data and providing data support for enterprises.

[0167] In summary, the present invention provides an e-commerce platform user feedback analysis method based on artificial intelligence, including obtaining text data of user feedback, user behavior data and an analysis model; wherein the analysis model includes a text vectorization model and a semantic analysis model; using the text vectorization model, the text data is converted into vectorized data; according to the vectorized data and the user behavior data, the semantic analysis model is used to calculate the user behavior label; according to the user behavior label, the user behavior vector is calculated; according to the user behavior vector, a set of association points is calculated; according to the user behavior vector and the user behavior label, the user keyword is calculated; according to the association point set and the user keyword, the user demand data is obtained and output. The method analyzes and processes the relevant data of the user through the artificial intelligence model, converts the user data into mathematical data for calculation, so that the personalized needs of each user can be accurately and efficiently obtained, the misjudgment of user needs is reduced, and feedback and adjustment can be made in a timely manner.

[0168] Reference Figure 2 The second embodiment of the present invention provides an e-commerce platform user feedback analysis system based on artificial intelligence, including:

[0169] A data acquisition module, used to acquire text data of user feedback and user behavior data and an analysis model; wherein the analysis model includes a text vectorization model and a semantic analysis model;

[0170] A vectorized data conversion module, used to convert the text data into vectorized data using the text vectorization model;

[0171] A behavior label calculation module, used to calculate the user behavior label based on the vectorized data and the user behavior data using the semantic analysis model;

[0172] A behavior vector calculation module, used to calculate a user behavior vector according to the user behavior label;

[0173] Among them, the behavior vector calculation module includes a data cleaning module, which is used to clean the text data to obtain complete text data; and a vector data calculation module, which is used to substitute the complete text data into the text vectorization model to obtain the vectorized data.

[0174] A correlation point calculation module, used to calculate a correlation point set according to the user behavior vector;

[0175] A keyword calculation module, used to calculate a user keyword according to the user behavior vector and the user behavior label;

[0176] Among them, the keyword calculation module includes an association matrix calculation module, which is used to perform secondary association calculation on the user behavior vector to obtain an association matrix; a behavior mapping matrix calculation module, which is used to map the user behavior label and the user behavior vector to obtain a behavior mapping matrix; a weight calculation module, which is used to calculate the behavior frequency value and the weight corresponding to each behavior frequency value according to the association matrix and the behavior mapping matrix; a keyword acquisition module, which is used to sort the weights from large to small, and select the user behavior vectors corresponding to the first several proportions of weights in the sorting, and output them as the user keywords.

[0177] The demand data acquisition module is used to obtain and output user demand data according to the association point set and the user keyword.

[0178] It should be noted that the artificial intelligence-based e-commerce platform user feedback analysis system provided in an embodiment of the present invention is used to execute all the process steps of the artificial intelligence-based e-commerce platform user feedback analysis method in the above embodiment. The working principles and beneficial effects of the two correspond one to one, and therefore will not be repeated here.

[0179] An embodiment of the present invention further provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as an e-commerce platform user feedback analysis program based on artificial intelligence. When the processor executes the computer program, the steps in the above-mentioned embodiments of the e-commerce platform user feedback analysis method based on artificial intelligence are implemented, such as Figure 1 Alternatively, when the processor executes the computer program, the functions of the modules / units in the above-mentioned system embodiments are realized, such as the vectorized data conversion module.

[0180] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the electronic device.

[0181] The electronic device may be a computing device such as a desktop computer, a notebook, a PDA, and a smart tablet. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. The electronic device may include more or fewer components than the above components, or may combine certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0182] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the electronic device, and uses various interfaces and lines to connect various parts of the entire electronic device.

[0183] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the electronic device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, 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), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0184] Wherein, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-0nly Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media does not include electrical carrier signals and telecommunication signals.

[0185] It should be noted that the system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the system embodiment provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0186] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. An e-commerce platform user feedback analysis method based on artificial intelligence, characterized in that: include: Acquire text data of user feedback, user behavior data and an analysis model; wherein the analysis model includes a text vectorization model and a semantic analysis model; Using the text vectorization model, converting the text data into vectorized data; According to the vectorized data and the user behavior data, using the semantic analysis model, calculate the user behavior label; Calculate a user behavior vector according to the user behavior label; Calculate a set of associated points according to the user behavior vector; Calculating a user keyword according to the user behavior vector and the user behavior label; According to the association point set and the user keyword, user demand data is obtained and output.

2. The method for analyzing user feedback on an e-commerce platform based on artificial intelligence according to claim 1, characterized in that: The step of using the text vectorization model to convert the text data into vectorized data includes: Performing data cleaning on the text data to obtain complete text data; Substitute the complete text data into the text vectorization model to obtain the vectorized data.

3. The method for analyzing user feedback on an e-commerce platform based on artificial intelligence according to claim 1, characterized in that: The step of calculating the user behavior label based on the vectorized data and the user behavior data and using the semantic analysis model includes: Inputting the vectorized data and the user behavior data into the semantic analysis model to obtain an initial behavior label; Modifying the initial behavior label to obtain the user behavior label by calculation; The user behavior label is calculated according to the following formula: Among them, L′ i represents the user behavior label of the i-th user, L i represents the initial behavior label of the i-th user, N represents the total number of user behaviors, and w ij represents the influence weight of the jth user behavior on the ith user label, a ij represents the first coefficient of the i-th user label under the j-th user behavior, b ij represents the second coefficient of the i-th user label under the j-th user behavior, B j Indicates the behavior label correction value corresponding to the j-th user behavior.

4. The method for analyzing user feedback on an e-commerce platform based on artificial intelligence according to claim 1, characterized in that: The calculating the user behavior vector according to the user behavior label includes: The user behavior vector is calculated according to the following formula: Among them, u i represents the i-th user behavior vector, x ij represents the j-th user behavior label of the ith user, and D represents the number of behavior types.

5. The method for analyzing user feedback on an e-commerce platform based on artificial intelligence according to claim 1, characterized in that: The step of calculating a set of associated points according to the user behavior vector includes: Perform association calculation according to the user behavior vector to obtain an initial association point set; Modifying the initial set of associated points to obtain the set of associated points by calculation; The initial set of associated points is calculated according to the following formula: G={(v i ,n i j)|i=1,2,…,K;j=1,2,…,M} Where G represents the initial associated point set, v i represents the initial correlation coefficient, n ij represents the number of the jth behavior corresponding to the ith user behavior vector, K represents the total number of the user behavior vectors, and M represents the total number of current behaviors; The set of associated points is calculated according to the following formula: G′={(v i ′,n ij )|i=1,2,…,K;j=1,2,…,M} Where G′ represents the set of associated points, v i ′ represents the modified correlation coefficient, α ij represents the first correction coefficient of the i-th user label under the j-th user behavior, β ij It represents the second correction coefficient of the i-th user label under the j-th user behavior.

6. The method for analyzing user feedback on an e-commerce platform based on artificial intelligence according to claim 1, characterized in that: The calculating the user keyword according to the user behavior vector and the user behavior label includes: Performing secondary correlation calculation on the user behavior vector to obtain a correlation matrix; Mapping the user behavior label and the user behavior vector to obtain a behavior mapping matrix; Calculating behavior frequency values ​​and weights corresponding to each behavior frequency value according to the association matrix and the behavior mapping matrix; The weights are sorted from large to small, and the user behavior vectors corresponding to the first several proportions of weights in the sorting are selected and output as the user keywords.

7. The method for analyzing user feedback on an e-commerce platform based on artificial intelligence according to claim 6, characterized in that: The performing secondary correlation calculation on the user behavior vector to obtain a correlation matrix includes: The association matrix is ​​calculated according to the following formula: Where A represents the association matrix, u i represents the i-th user behavior vector, u N represents the Nth user behavior vector, where N represents the number of users.

8. The method for analyzing user feedback on an e-commerce platform based on artificial intelligence according to claim 6, characterized in that: The mapping of the user behavior label and the user behavior vector to obtain a behavior mapping matrix includes: The behavior mapping matrix is ​​calculated according to the following formula: R=[u1·L1 … u i ·L i … the N ·L N ] Where R represents the behavior mapping matrix, u i represents the i-th user behavior vector, u N represents the Nth user behavior vector, L i represents the i-th user behavior label, L N Represents the Nth user behavior label.

9. The method for analyzing user feedback on an e-commerce platform based on artificial intelligence according to claim 6, characterized in that: The step of calculating the behavior frequency value and the weight corresponding to each behavior frequency value according to the association matrix and the behavior mapping matrix includes: The behavior frequency value is calculated according to the following formula: Among them, F j represents the jth frequency value of the behavior, A i，j represents the element in the i-th row and j-th column of the incidence matrix, B j represents the j-th column element of the behavior mapping matrix, and D represents the number of behavior types; The weight corresponding to each behavior frequency value is calculated according to the following formula: Among them, W j represents the weight corresponding to the j-th behavior frequency value, and N represents the number of users.

10. An e-commerce platform user feedback analysis system based on artificial intelligence, characterized in that: include: A data acquisition module, used to acquire text data of user feedback and user behavior data and an analysis model; wherein the analysis model includes a text vectorization model and a semantic analysis model; A vectorized data conversion module, used to convert the text data into vectorized data using the text vectorization model; A behavior label calculation module, used to calculate the user behavior label based on the vectorized data and the user behavior data using the semantic analysis model; A behavior vector calculation module, used to calculate a user behavior vector according to the user behavior label; A correlation point calculation module, used to calculate a correlation point set according to the user behavior vector; A keyword calculation module, used to calculate a user keyword according to the user behavior vector and the user behavior label; The demand data acquisition module is used to obtain and output user demand data according to the association point set and the user keyword.

Images