Product Strategy Improvement Method Based on LSGDM-SHAP

Abstract

This invention provides a product strategy improvement method, system, storage medium, and electronic device based on LSGDM-SHAP, relating to the field of natural language processing technology. In this invention, firstly, online comparative reviews published by several consumers after purchasing a target product or its competitors are collected; secondly, attribute information is extracted from the online comparative reviews and consumer sentiment analysis is performed; thirdly, the LSGDM method is used to seek consensus among different consumer viewpoints to obtain the product attribute performance that best represents the overall viewpoint; next, the SHAP interpretability method is used to obtain the importance of product attributes; finally, under the IPA model framework, product attributes are mapped to a two-dimensional plane to obtain product attribute improvement strategies for the target product. From the perspective of market competition, utilizing online comparative reviews allows for a quick and accurate understanding of consumer needs, helping companies grasp key areas for product improvement, providing product improvement strategies, and enabling the effective and rational use of company resources.

Description

Technical Field

[0001] This invention relates to the field of natural language processing technology, and more specifically to a product strategy improvement method, system, storage medium, and electronic device based on LSGDM-SHAP. Background Technology

[0002] With the diversification and personalization of consumer demands, product iteration frequency is accelerating and development cycles are shortening. The ability to quickly understand user needs and preferences and accurately grasp the direction of product improvement has become a key aspect of the product development process for enterprises.

[0003] In existing technologies, the Individual Product Analysis (IPA) model is a commonly used business research technique for understanding customer needs and developing product improvement strategies. Traditional IPA models, through market research, use two dimensions—importance and performance—to categorize product attributes from a consumer perspective into four quadrants, helping companies identify areas for product improvement. However, traditional IPA models typically collect consumer data through questionnaires or in-person interviews, a time-consuming and labor-intensive approach that is ill-suited to today's fiercely competitive market environment.

[0004] For example, existing researchers use online hotel reviews and ratings as data to conduct IPA model analysis on hotel services, helping hotels identify service improvement strategies. This study first extracts hotel service attributes mentioned by consumers from online reviews and performs sentiment analysis, using consumer sentiment outcomes as the performance indicators of the hotel services. Then, the study uses consumer sentiment outcomes and ratings to build an ensemble neural network model to measure the importance of hotel services. Finally, based on the measured performance and importance, an IPA model is constructed, categorizing the service attributes mentioned by consumers into four quadrants and providing strategies for hotel service improvement.

[0005] However, the original product improvement strategy model only analyzes a single product or service and provides suggestions for product improvement. Summary of the Invention

[0006] (a) Technical problems to be solved

[0007] To address the shortcomings of existing technologies, this invention provides a product strategy improvement method, system, storage medium, and electronic device based on LSGDM-SHAP, which solves the technical problem that the original product improvement strategy model only analyzes a single product or service.

[0008] (II) Technical Solution

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A product strategy improvement method based on LSGDM-SHAP includes:

[0011] Collect online comparative reviews published by several consumers after purchasing the target product or its competitors; each of the online comparative reviews includes attribute comparison information between the target product and the competing products;

[0012] Extract attribute information from the online comparison reviews, and perform consumer sentiment analysis on the attribute information to obtain sentiment scores for different product attributes in each online comparison review;

[0013] Based on the sentiment scores, the LSGDM method is used to seek consensus among different consumer viewpoints in order to obtain the product attribute performance that best represents the overall viewpoint.

[0014] Based on the sentiment score and purchase results, a neural network model corresponding to each product attribute is pre-trained. The Shapley value of each product attribute is calculated using the SHAP interpretability method and multiplied by the sentiment score to obtain the contribution value of different product attributes to the purchase results predicted by the model, which is used as the importance of product attributes.

[0015] Based on the performance and importance of the product attributes, the product attributes are mapped to a two-dimensional plane within the IPA model framework to obtain product attribute improvement strategies for the target product.

[0016] Preferably, LDA is used to extract attribute information from the online comparative reviews.

[0017] Preferably, the consumer sentiment analysis based on the attribute information includes:

[0018] Product name identification is performed using the Conditional Random Field method;

[0019] Sentiment analysis was performed using the Snownlp tool, with scores ranging from 0 to 1. A score greater than 0.5 indicated positive sentiment, while a score less than 0.5 indicated negative sentiment.

[0020] Preferably, the step of seeking consensus among different consumer viewpoints based on the sentiment score, in order to obtain the product attribute performance that best represents the overall viewpoint, includes:

[0021] The K-means algorithm was used to cluster all online comparative reviews into multiple subgroups of opinions, and the comprehensive expression of each subgroup of opinions was calculated.

[0022] By constructing a variance-minimizing objective function, the opinion weights of each subgroup's viewpoints are continuously optimized;

[0023] Based on the optimal opinion weight of each subgroup's viewpoint and the adjusted comprehensive expression, the weighted evaluation reaches an overall decision-making consensus and serves as the product attribute performance measurement result that best represents the overall viewpoint.

[0024] Preferably, the Shapley value for each product attribute is calculated using the SHAP interpretability method, and the calculation formula is as follows:

[0025]

[0026] in, It is attribute A i The Shapley value, F is the set of sentiment scores for all attributes; S does not include A. i Let f(S) be the subset of attributes S, and f(S) be the predicted value of the model on the subset of attributes S. i})-f(S) is attribute A i Marginal contribution to subset S; It is a weight used to ensure that the contribution of all product attribute combinations is distributed fairly.

[0027] Preferably, the product attribute enhancement strategy includes:

[0028] Set thresholds for distinguishing between the importance of product attributes and the performance of product attributes;

[0029] If both the importance and performance of a product attribute are higher than the corresponding threshold, the product attribute enhancement strategy is to maintain them.

[0030] If the importance of a product attribute is higher than the corresponding threshold, but the performance of the product attribute is lower than the corresponding threshold, then the product attribute improvement strategy is to focus on it.

[0031] If the importance of a product attribute is below the corresponding threshold, but the performance of the product attribute is above the corresponding threshold, then the product attribute improvement strategy is an over-effort.

[0032] If both the importance and performance of a product attribute are below the corresponding threshold, then the product attribute enhancement strategy will be of low priority.

[0033] A product strategy improvement system based on LSGDM-SHAP includes:

[0034] The review collection module is used to collect online comparative reviews published by several consumers after purchasing the target product or its single competitor; each of the online comparative reviews includes attribute comparison information between the target product and the single competitor;

[0035] The sentiment analysis module is used to extract attribute information from the online comparative reviews and to perform consumer sentiment analysis on the attribute information to obtain sentiment scores for different product attributes under each purchase result.

[0036] The performance acquisition module is used to seek consensus among different consumer viewpoints based on the sentiment score, using the LSGDM method to obtain the product attribute performance that best represents the overall viewpoint.

[0037] The attribute importance acquisition module is used to pre-train a neural network model corresponding to each product attribute based on the sentiment score and the purchase result, calculate the Shapley value of each product attribute using the SHAP interpretability method, and multiply it by the sentiment score to obtain the contribution value of different product attributes to the purchase result predicted by the model, and use it as the product attribute importance.

[0038] The strategy acquisition module is used to map product attributes to a two-dimensional plane within the IPA model framework, based on the performance and importance of the product attributes, in order to obtain product attribute acquisition strategies for the target product.

[0039] A storage medium storing a computer program for product strategy enhancement based on LSGDM-SHAP, wherein the computer program causes a computer to execute the product strategy enhancement method as described above.

[0040] An electronic device, comprising:

[0041] One or more processors; memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for performing product strategy enhancement as described above.

[0042] (III) Beneficial Effects

[0043] This invention provides a product strategy improvement method, system, storage medium, and electronic device based on LSGDM-SHAP. Compared with the prior art, it has the following advantages:

[0044] This invention first collects online comparative reviews from consumers who purchased a target product or its competitors. Second, it extracts attribute information from these reviews and performs consumer sentiment analysis. Third, it employs the LSGDM method to seek consensus among different consumer viewpoints, obtaining the product attribute performance that best represents the overall perspective. Next, it uses the SHAP interpretability method to determine the importance of product attributes. Finally, within the IPA model framework, it maps product attributes to a two-dimensional plane to obtain product attribute improvement strategies for the target product. From a market competition perspective, utilizing online comparative reviews allows for a quick and accurate understanding of consumer needs, helping companies grasp key areas for product improvement, providing product improvement strategies, and enabling the effective and rational use of company resources. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 A block diagram illustrating a product strategy improvement method based on LSGDM-SHAP provided in an embodiment of the present invention;

[0047] Figure 2 A flowchart of a product strategy improvement method based on LSGDM-SHAP provided for embodiments of the present invention;

[0048] Figure 3 This is an example diagram of an IPA model provided in an embodiment of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] This application provides a product strategy improvement method, system, storage medium, and electronic device based on LSGDM-SHAP, which solves the technical problem that the original product improvement strategy model only analyzes a single product or service.

[0051] The technical solution in this application embodiment is to solve the above-mentioned technical problems, and the key points are as follows:

[0052] 1. Product improvement strategies in a competitive environment are considered. This invention categorizes reviews into two types: general reviews and comparative reviews. Using comparative reviews as research data, the competitive advantages and disadvantages of the products are analyzed through consumer comparisons of two products and their final purchase decisions, and product improvement strategies are provided.

[0053] 2. Use large-scale group decision-making methods to obtain consumer consensus. By employing the concept of large-scale group decision-making, consensus is sought among different consumer perspectives to obtain product attribute performance measurement results that best represent the overall viewpoint.

[0054] 3. Use interpretable neural network models to measure the importance of product attributes. By using SHAP interpretability technology, the contribution value of each product attribute to the final purchase result is calculated, thereby measuring the importance of product attributes and solving the problem of poor interpretability of traditional neural networks.

[0055] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0056] Example 1:

[0057] like Figure 1 As shown, this embodiment of the invention provides a product strategy improvement method based on LSGDM-SHAP, including:

[0058] S1. Collect online comparative reviews published by several consumers after purchasing the target product or its competitors; each of the online comparative reviews includes attribute comparison information between the target product and the competing products;

[0059] S2. Extract attribute information from the online comparison reviews, and perform consumer sentiment analysis on the attribute information to obtain sentiment scores for different product attributes in each online comparison review;

[0060] S3. Based on the sentiment score, the LSGDM method is used to seek consensus among different consumer viewpoints in order to obtain the product attribute performance that best represents the overall viewpoint.

[0061] S4. Based on the sentiment score and purchase results, pre-train a neural network model corresponding to each product attribute, use the SHAP interpretability method to calculate the Shapley value of each product attribute, and multiply it with the sentiment score to obtain the contribution value of different product attributes to the purchase results predicted by the model, and use it as the importance of product attributes.

[0062] S5. Based on the performance and importance of the product attributes, the product attributes are mapped to a two-dimensional plane within the IPA model framework to obtain product attribute improvement strategies for the target product.

[0063] This invention, from the perspective of market competition, utilizes online comparative reviews to quickly and accurately understand consumer needs, helping companies grasp key areas for product improvement, providing product improvement strategies, and enabling companies to make effective and rational use of their resources.

[0064] like Figure 2 As shown, Figure 2 A flowchart for improving product strategy based on LSGDM-SHAP is disclosed.

[0065] Next, we will combine Figure 2 The steps of the above scheme are described in detail:

[0066] With the rise of online platforms, consumers are using social media software to post online reviews expressing their experiences using products. Due to the casual nature of consumer language, online reviews often contain a lot of irrelevant information, such as interjections, emoticons, and expressions unrelated to the product.

[0067] Furthermore, unstructured text is difficult to apply directly to subsequent analysis, so it is necessary to extract key information for analysis from online comments and transform it into structured data.

[0068] For details, please refer to steps S1 to S3:

[0069] In step S1, online comparative reviews published by several consumers after purchasing the target product or its competitors are collected; each of the online comparative reviews includes attribute comparison information between the target product and the competing products.

[0070] It should be noted that this embodiment of the invention categorizes online reviews into two types: general reviews and comparative reviews. This step uses online comparative reviews as research data. Online comparative reviews refer to consumers' frequent comparisons of the product with competing products in various aspects when posting online reviews, in addition to expressing their experience with the product. Such comparative reviews can reflect the degree to which the different attributes of the two products influence consumers' purchasing decisions.

[0071] For example, given the highly competitive and diverse functionalities of the automotive market, this embodiment of the invention selects two best-selling compact cars and collects user reviews of these two cars from three websites: Autohome, Autohome, and Pacific Auto.

[0072] In step S2, attribute information is extracted from the online comparison reviews, and consumer sentiment analysis is performed on the attribute information to obtain sentiment scores for different product attributes in each of the online comparison reviews.

[0073] In an optional implementation, LDA is used to extract attribute information from the online comparative reviews. LDA is an unsupervised machine learning method based on Bayesian probability, used to represent multiple topics in a document as a probability distribution. Studies have shown that LDA can effectively extract product attributes from online reviews.

[0074] To improve the extraction of product attributes, this step involves some preprocessing of online reviews. Typically, product attributes appear in user reviews as nouns or noun phrases; therefore, part-of-speech tagging is used to retain nouns and noun phrases from online reviews while removing adjectives, adverbs, punctuation marks, and other irrelevant content.

[0075] In particular, vehicle attributes exhibit hierarchical relationships. For example, the "space" attribute includes fine-grained attributes such as "front and rear seats," "cabin space," and "legroom." Therefore, considering the specialized nature of the automotive field, in an optional implementation, this step also leverages domain expert knowledge to manually construct an attribute hierarchy dictionary to further refine the LDA extraction results.

[0076] Furthermore, in an optional implementation, the consumer sentiment analysis based on the attribute information includes:

[0077] Product name identification is performed using the Conditional Random Field method;

[0078] Sentiment analysis was performed using the Snownlp tool, with scores ranging from 0 to 1. A score greater than 0.5 indicated positive sentiment, while a score less than 0.5 indicated negative sentiment.

[0079] Specifically, in product comparison reviews, consumers express their opinions on the performance of a product in certain attributes and compare it with other products. Therefore, comparison reviews will mention at least two or more product names. Furthermore, considering the variability of consumer language, there may also be reviews that only mention the names of the products being compared. Therefore, the comparison reviews studied in this embodiment refer to online consumer reviews that mention other products besides the purchased product itself. First, we use a conditional random field method to identify product names. Considering that consumers often use a large number of product nicknames and internet slang to refer to products, in an optional implementation, a product name dictionary is also manually constructed to more accurately identify whether and which product names are mentioned in product reviews.

[0080] In comparative reviews, consumers express their emotional response to product attributes by comparing them with other products. For example, considering products A and B, if a consumer mentions product B in their review of product A, it's assumed the consumer is comparing the two products. Based on this, if the consumer has a positive feeling about a certain attribute of product A, it's assumed the consumer feels product A is superior to product B in that attribute, recorded as 1; if the consumer expresses a negative feeling about an attribute, it's assumed the consumer feels product A is inferior to product B in that attribute, recorded as -1; if no attribute is mentioned, it's recorded as 0.

[0081] Building upon this foundation, this step employs the Snownlp tool for sentiment analysis. Snownlp calculates the sentiment score of a sentence, representing it with a number between 0 and 1. A score greater than 0.5 indicates positive sentiment, while a score less than 0.5 indicates negative sentiment. Ultimately, Table 1 shows the product attribute performance that best represents the overall viewpoint.

[0082] Table 1

[0083] Comment ID Purchase Results Attribute 1 Attribute 2 … attribute n 1 A 1 -1 … 0 2 B 0 -1 … -1 … … … … … … n A 1 -1 … 0

[0084] In step S3, based on the sentiment score, the LSGDM method is used to seek consensus among different consumer viewpoints in order to obtain the product attribute performance that best represents the overall viewpoint.

[0085] The emotional responses to product attributes expressed in online reviews reflect consumers' perceptions of product performance. Online reviews also reflect consumers' decision-making processes; the emotional expressions of different consumers regarding product attributes and their purchasing outcomes can be understood as the results of decisions made by different decision-makers after considering their respective preferred options. Large-scale group decision-making is a consensus-building process, capable of reaching a consensus viewpoint from online reviews with differing perspectives.

[0086] Therefore, this step employs Large-scale Group Decision Making (LSGDM) technology to measure product attribute performance from online reviews that accurately reflects the opinions of a group of consumers. In summary:

[0087] First, the K-means algorithm is used to cluster all online comparative reviews into multiple subgroups of opinions, and the comprehensive expression of each subgroup of opinions is calculated.

[0088] Secondly, by constructing a variance minimization objective function, the opinion weights of each subgroup's viewpoints are continuously optimized.

[0089] Finally, based on the optimal opinion weight of each subgroup's viewpoint and the adjusted comprehensive expression, a weighted evaluation is used to reach an overall decision-making consensus and serves as the product attribute performance measurement result that best represents the overall viewpoint.

[0090] Specifically:

[0091] 1) Decision-making subgroup division

[0092] For ease of explanation, assume there are n online comparison reviews r. k (k = 1, ..., n), m product attributes A were extracted from online comparative reviews. i (i = 1, ..., m), based on the sentiment scores perf for each attribute in online comments. i Online comments can be expressed in vector form.

[0093] Calculate the consensus (CD) between two comments. xy The calculation formula is as follows:

[0094]

[0095] in, Online comparison reviews r x r y Product Attribute A i The emotional score.

[0096] The K-means algorithm is used to cluster online comments, and consensus degree is used as the distance between cluster objects. The entire group LG can be divided into w decision subgroups SG. k (k=1,...,w) represents the views of w subgroups.

[0097] After clustering, it is necessary to calculate the comprehensive representation of the viewpoints of each subgroup. This comprehensive viewpoint can also be expressed as a vector form. The calculation formula is as follows:

[0098]

[0099] Where, n k For SG k The number of comments in r a Belongs to SG k One of the comments.

[0100] 2) Reach a consensus

[0101] After obtaining a comprehensive expression of the viewpoints of each subgroup, the internal cohesion of the subgroup is first calculated using the following formula:

[0102]

[0103] By weighting the cohesion within a subgroup with its overall expression, a preliminary overall decision-making scheme is obtained. in,

[0104]

[0105] At this point, a consensus has not yet been reached on the overall decision-making plan. In order to reach a consensus, it is necessary to identify one or more subgroups among the subgroups whose opinions differ the most from those of other subgroups. These subgroups are hindering the entire group from reaching a consensus and need to be adjusted.

[0106] Therefore, it is necessary to calculate the consensus among the subgroups, which needs to be divided into attribute consensus. and overall consistency The calculation formula is as follows:

[0107]

[0108] like The closer it is to 1, the more likely it is that the subgroup is SG. k The opinions of the subgroups are closer to those of the main group. In this case, a consistency threshold α∈[0.5,1) needs to be set to find the subgroups that need adjustment and whose opinions differ greatly from the main group.

[0109] Next, adjustments are made to subgroups with consensus below α, similar to the Delphi method where differing opinions in subgroups gradually converge through repeated communication. Assume SG... (k) For subgroups with consistency below α, the adjustment method is as follows:

[0110]

[0111] in, For SG (k) In attribute A m Emotional expression below α, δ r These are adjustment parameters used to control inconsistent subgroups SG. (k) The level of adjustment should be similar to that of the group opinion, and this should be provided in advance by the decision-maker.

[0112] To achieve group consensus, subgroups need to be weighted according to their internal cohesion, and the final weighted average yields the group consensus. As the subgroups SG... (k) The adjustment of the subgroup's cohesion W(SG) (k) The weights will also change. Therefore, it is necessary to construct a minimum variance objective function to obtain the optimal weights W = [W(SG1), W(SG2), ..., W(SG...]. w The objective function is calculated as follows:

[0113]

[0114]

[0115] By optimizing the model, the optimal weights for each subgroup can be obtained. The final consensus expression is then obtained by weighting these weights with the adjusted combined expression of each subgroup. This consensus can be seen as the shared sentiment scores of all online reviews regarding the product's various attributes, which is the consumer's perception of the product's attribute performance that needs to be measured.

[0116] In step S4, based on the sentiment score and the purchase result, a neural network model corresponding to each product attribute is pre-trained. The Shapley value of each product attribute is calculated using the SHAP interpretability method and multiplied by the sentiment score to obtain the contribution value of different product attributes to the purchase result predicted by the model, which is used as the importance of the product attribute.

[0117] The importance of product attributes reflects the degree to which consumers' emotional comparisons of various product attributes influence their final purchase decision, and the relationship between the two is non-linear. While traditional artificial neural networks can achieve excellent results in fitting this non-linear relationship, the black-box nature of neural networks makes it impossible to clearly identify which factors influence the final model output. To clarify the parameter relationships within the neural network, in an optional implementation, this invention uses the SHAP (SHapley Additive exPlanations) interpretability technique to obtain the degree of importance of product attributes in influencing the final purchase decision.

[0118] Specifically, this step includes:

[0119] 1) Constructing a neural network model

[0120] The neural network model is trained by using consumers' sentiment scores on different attributes as input and users' purchase results as output. Since consumers don't mention all attributes in their reviews, the product comparison data is relatively sparse. Sparse data can affect the network's training results; therefore, K-fold cross-validation is used to divide the dataset into K parts and train K neural network models to avoid overfitting caused by sparse data.

[0121] 2) Calculate the importance of product attributes

[0122] SHAP is an interpretation method based on Shapley values. It interprets the model's predictions by calculating the Shapley value for each feature. The core idea of ​​Shapley values ​​is to calculate the marginal contribution of each feature across all possible feature combinations and then take a weighted average. For attribute A... i The formula for calculating its Shapley value is:

[0123]

[0124] in, It is attribute A i The Shapley value, F is the set of sentiment scores for all attributes; S does not include A. i Let f(S) be the subset of attributes S, and f(S) be the predicted value of the model on the subset of attributes S. i})-f(S) is attribute A i Marginal contribution to subset S; It is a weight used to ensure that the contribution of all product attribute combinations is distributed fairly.

[0125] To estimate the Shapley value in a neural network model, the SHAP method infers the weights between the input and output layers by combining the effects of features of small neural network components on the entire neural network. In each trained neural network, the SHAP value of the product attributes mentioned in each review can be calculated, as shown in Table 2.

[0126] Table 2

[0127]

[0128]

[0129] Then, in the k-th neural network model, the importance of product attributes (Imp) can be obtained. ik The calculation formula is as follows:

[0130]

[0131] Where, n k The number of online comments used to train the k-th neural network, where j is the index.

[0132] Finally, using the F1 score of each neural network as a weight, the importance of product attributes measured from the k neural networks is weighted and summed, as shown in the following formula:

[0133]

[0134] Among them, w k Let f be the F1 score of the k-th neural network.

[0135] In step S5, based on the performance and importance of the product attributes, the product attributes are mapped to a two-dimensional plane within the IPA model framework to obtain product attribute improvement strategies for the target product.

[0136] Based on measurements and Imp m ,like Figure 3 As shown, this step maps the product attributes of the target product to a two-dimensional plane. Based on the IPA model framework, product attribute improvement strategies are typically divided into four categories, requiring the setting of thresholds for importance and performance. For these thresholds, a common average value is used. The four product attribute improvement strategies are as follows:

[0137] (1) Continue to maintain: The importance of product attributes is higher than the set threshold, indicating that product attributes are very important in consumers' purchasing decisions; the performance of product attributes is also higher than the set threshold, indicating that the performance of product attributes is in a comparative advantage position in the competitive market. Such product attributes belong to the core product competitiveness of enterprises and should maintain the competitive advantage of product attributes.

[0138] (2) Focus: If the importance of a product attribute is higher than the set threshold, it indicates that the product attribute is very important in the consumer's purchase decision; if the performance of a product attribute is lower than the set threshold, it indicates that the performance of the product attribute is at a relatively disadvantage in the competitive market. This type of product attribute is a significant product defect for the company, and its performance should be prioritized for improvement.

[0139] (3) Over-effort: If the importance of a product attribute is below a set threshold, it indicates that consumers do not value this type of product attribute during the purchase decision-making process; if the performance of a product attribute is above a set threshold, it indicates that the performance of the product attribute is above the average level of the competitive market. The superior performance of this type of product attribute has little impact on improving product competitiveness, and companies should avoid excessive resource investment in this type of product attribute.

[0140] (4) Low priority: The importance of product attributes is lower than the set threshold, indicating that consumers do not pay enough attention to these product attributes in the purchasing decision process; the performance of product attributes is also lower than the set threshold, indicating that the performance of product attributes is worse than the average level of the competitive market. Since improving the performance of these product attributes may generate very small market economic benefits, enterprises should give these product attributes low priority.

[0141] Thus, this embodiment of the invention completes the entire process of the product strategy improvement method based on LSGDM-SHAP.

[0142] Example 2:

[0143] This invention provides a product strategy improvement system based on LSGDM-SHAP, comprising:

[0144] The review collection module is used to collect online comparative reviews published by several consumers after purchasing the target product or its single competitor; each of the online comparative reviews includes attribute comparison information between the target product and the single competitor;

[0145] The sentiment analysis module is used to extract attribute information from the online comparative reviews and to perform consumer sentiment analysis on the attribute information to obtain sentiment scores for different product attributes under each purchase result.

[0146] The performance acquisition module is used to seek consensus among different consumer viewpoints based on the sentiment score, using the LSGDM method to obtain the product attribute performance that best represents the overall viewpoint.

[0147] The attribute importance acquisition module is used to pre-train a neural network model corresponding to each product attribute based on the sentiment score and the purchase result, calculate the Shapley value of each product attribute using the SHAP interpretability method, and multiply it by the sentiment score to obtain the contribution value of different product attributes to the purchase result predicted by the model, and use it as the product attribute importance.

[0148] The strategy acquisition module is used to map product attributes to a two-dimensional plane within the IPA model framework, based on the performance and importance of the product attributes, in order to obtain product attribute acquisition strategies for the target product.

[0149] Example 3:

[0150] This invention provides a storage medium storing a computer program for product strategy improvement based on LSGDM-SHAP, wherein the computer program causes a computer to execute the product strategy improvement method as described in Embodiment 1.

[0151] Example 4:

[0152] This invention provides an electronic device, comprising:

[0153] One or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for performing product strategy enhancement methods as described in Example 1.

[0154] It is understood that the product policy improvement system, storage medium and electronic device based on LSGDM-SHAP provided in the embodiments of the present invention correspond to the product policy improvement method based on LSGDM-SHAP provided in the embodiments of the present invention. The explanation, examples and beneficial effects of the relevant contents can be referred to the corresponding parts of the product policy improvement method, and will not be repeated here.

[0155] In summary, compared with existing technologies, it has the following beneficial effects:

[0156] In this embodiment of the invention, firstly, online comparative reviews published by several consumers after purchasing the target product or its competitors are collected; secondly, attribute information is extracted from the online comparative reviews and consumer sentiment analysis is performed; thirdly, the LSGDM method is used to seek consensus among different consumer viewpoints to obtain the product attribute performance that best represents the overall viewpoint; next, the SHAP interpretability method is used to obtain the importance of product attributes; finally, under the IPA model framework, product attributes are mapped to a two-dimensional plane to obtain product attribute improvement strategies for the target product. From the perspective of market competition, online comparative reviews are used to quickly and accurately understand consumer needs, help companies grasp the key points of product improvement, provide companies with product improvement strategies, and enable companies to make effective and rational use of resources.

[0157] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0158] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A product strategy improvement method based on LSGDM-SHAP, characterized in that, include: Collect online comparative reviews from consumers after purchasing the target product or its competitors; Each of the aforementioned online comparative reviews includes attribute comparison information between the target product and competing products; Extract attribute information from the online comparison reviews, and perform consumer sentiment analysis on the attribute information to obtain sentiment scores for different product attributes in each online comparison review; Based on the sentiment scores, the LSGDM method is used to seek consensus among different consumer viewpoints in order to obtain the product attribute performance that best represents the overall viewpoint. Based on the sentiment score and purchase results, a neural network model corresponding to each product attribute is pre-trained. The Shapley value of each product attribute is calculated using the SHAP interpretability method and multiplied by the sentiment score to obtain the contribution value of different product attributes to the purchase results predicted by the model, which is used as the importance of product attributes. Based on the performance and importance of the product attributes, the product attributes are mapped to a two-dimensional plane within the IPA model framework to obtain product attribute improvement strategies for the target product. Based on the sentiment score, the LSGDM method is used to seek consensus among different consumer viewpoints in order to obtain the product attribute performance that best represents the overall viewpoint, including: The K-means algorithm was used to cluster all online comparative reviews into multiple subgroups of opinions, and the comprehensive expression of each subgroup of opinions was calculated. By constructing a variance-minimizing objective function, the opinion weights of each subgroup's viewpoints are continuously optimized; Based on the optimal opinion weight of each subgroup's viewpoint and the adjusted comprehensive expression, the weighted evaluation reaches an overall decision-making consensus and serves as the product attribute performance measurement result that best represents the overall viewpoint.

2. The product strategy improvement method as described in claim 1, characterized in that, LDA was used to extract attribute information from the online comparative reviews.

3. The product strategy improvement method as described in claim 1, characterized in that, The consumer sentiment analysis based on the attribute information includes: Product name identification is performed using the Conditional Random Field method; Sentiment analysis was performed using the Snownlp tool, with scores ranging from 0 to 1. A score greater than 0.5 indicated positive sentiment, while a score less than 0.5 indicated negative sentiment.

4. The product strategy improvement method as described in claim 1, characterized in that, The Shapley value for each product attribute is calculated using the SHAP interpretability method. The formula is as follows: in, It is an attribute Shapley value, It is the collection of all attribute sentiment scores; It does not include A subset of attributes, The model is in the attribute subset The predicted value on, It is an attribute Pair subset The marginal contribution; It is a weight used to ensure that the contribution of all product attribute combinations is distributed fairly.

5. The product strategy improvement method as described in claim 1, characterized in that, The product attribute enhancement strategies include: Set thresholds for distinguishing between the importance of product attributes and the performance of product attributes; If both the importance and performance of a product attribute are higher than the corresponding threshold, the product attribute enhancement strategy is to maintain them. If the importance of a product attribute is higher than the corresponding threshold, but the performance of the product attribute is lower than the corresponding threshold, then the product attribute improvement strategy is to focus on it. If the importance of a product attribute is below the corresponding threshold, but the performance of the product attribute is above the corresponding threshold, then the product attribute improvement strategy is an over-effort. If both the importance and performance of a product attribute are below the corresponding threshold, then the product attribute enhancement strategy will be of low priority.

6. A product strategy improvement system based on LSGDM-SHAP, characterized in that, For performing the product strategy improvement method as described in claim 1, comprising: The review collection module is used to collect online comparative reviews published by several consumers after purchasing the target product or its single competitor; each of the online comparative reviews includes attribute comparison information between the target product and the single competitor; The sentiment analysis module is used to extract attribute information from the online comparative reviews and to perform consumer sentiment analysis on the attribute information to obtain sentiment scores for different product attributes under each purchase result. The performance acquisition module is used to seek consensus among different consumer viewpoints based on the sentiment score, using the LSGDM method to obtain the product attribute performance that best represents the overall viewpoint. The attribute importance acquisition module is used to pre-train a neural network model corresponding to each product attribute based on the sentiment score and the purchase result, calculate the Shapley value of each product attribute using the SHAP interpretability method, and multiply it by the sentiment score to obtain the contribution value of different product attributes to the purchase result predicted by the model, and use it as the product attribute importance. The strategy acquisition module is used to map product attributes to a two-dimensional plane within the IPA model framework, based on the performance and importance of the product attributes, in order to obtain product attribute acquisition strategies for the target product.

7. A storage medium, characterized in that, It stores a computer program for product strategy improvement based on LSGDM-SHAP, wherein the computer program causes the computer to execute the product strategy improvement method as described in any one of claims 1 to 5.

8. An electronic device, characterized in that, include: One or more processors; Memory; And one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for performing the product strategy improvement method as described in any one of claims 1 to 5.

Images