A SaaS system based on commodity exchange

Abstract

The present invention relates to the field of computer application technology, and provides a SaaS system based on commodity exchange, comprising: a module for determining commodities to be exchanged, which selects commodities to be exchanged based on the highest single item amount, average commodity amount, and accumulated points of a user's current shopping; a commodity review library module, which updates the commodity review library with commodity reviews from external large-scale shopping websites; a judgment matrix generation module, which generates a judgment matrix reflecting commodity attributes based on the review information in the commodity review library; a user propensity matrix generation module, which performs decision tree classification on the current shopping commodities and user registration information to generate a comparison matrix; and a hierarchical analysis model, which calculates the judgment matrix and the comparison matrix to select recommended commodities from the commodities to be exchanged. The present invention realizes that a small online shopping SaaS system can derive suitable recommended points exchange commodities based solely on the current shopping commodities of the user, without relying on user characteristics generated by the user's multiple shopping trips or multiple platform usages.

Description

Technical Field

[0001] The present invention relates to the field of computer application technology, and in particular to a SaaS system based on commodity exchange. Background Art

[0002] As a cloud platform for shopping, online shopping SaaS systems usually use the method of exchanging user points for goods to provide users with unexpected surprises during the shopping process in order to increase user stickiness to the platform and the frequency of product purchases, thereby enhancing user loyalty, promoting repeat purchases and improving user activity.

[0003] However, for small online shopping SaaS systems, users may seldom use them or have just started using them, or even have only used them once or twice. As a result, small online shopping SaaS systems are unable to build a user information database with sufficient data to perfectly match the recommended points redemption products with user information, resulting in the points product redemption being unable to fully play its role in improving the user shopping experience.

[0004] Therefore, how to enable a small online shopping SaaS system to select recommended points and redeem products that match the user based on the review information of external large shopping websites and the user's current shopping information, without relying on the user characteristics generated by multiple shopping and multiple use of the platform, is a technical problem that needs to be solved. Summary of the Invention

[0005] To this end, the present invention provides a SaaS system based on commodity exchange, which adopts a hierarchical analysis model to make a comprehensive judgment on the characteristics of the user's shopping commodities and the comments on the commodities on external large shopping websites to determine the recommended points exchange commodities that match the user and meet the user's preferences, thereby realizing a small online shopping SaaS system that can obtain suitable recommended points exchange commodities based only on the user's shopping commodities, without relying on the user characteristics generated by multiple shopping and multiple use of the platform.

[0006] To achieve the above objectives, the present invention proposes a SaaS system based on commodity exchange, comprising:

[0007] The module for determining the products to be exchanged is used to select multiple products to be exchanged from the products purchased by the user and the exchange product library based on the highest single item price, average product price and accumulated points of the user's current shopping;

[0008] The product review database module is connected to external large-scale shopping websites to match and filter product reviews on external large-scale shopping websites and generate / update the product review database;

[0009] A judgment matrix generation module, connected to the to-be-selected exchange commodity determination module and the commodity review library module, for generating a judgment matrix reflecting commodity attributes based on the review information of the to-be-selected exchange commodity in the commodity review library;

[0010] A user preference matrix generation module, connected to the module for determining the items to be redeemed, is used to perform a decision tree classification on the items purchased this time and the user registration information to generate a comparison matrix reflecting the user's preference for the attributes of the items;

[0011] The hierarchical analysis model is connected to the judgment matrix generation module and the user tendency matrix generation module, and uses the hierarchical analysis method to calculate the judgment matrix and the comparison matrix, and selects the recommended exchange product from the multiple exchange products to be selected.

[0012] Furthermore, the judgment matrix generation module includes a satisfaction judgment matrix unit, and the commodity attributes include commodity satisfaction;

[0013] The satisfaction judgment matrix unit uses a sentiment intensity analysis method to calculate the product satisfaction reflected in the review information, and generates a judgment matrix according to the product satisfaction.

[0014] Furthermore, the satisfaction judgment matrix unit includes a word set generation subunit and a satisfaction calculation subunit;

[0015] The word set generating subunit is used to generate a positive and satisfactory review set, a negative and satisfactory review set, a positive and untrustworthy review set, a negative and untrustworthy review set and a general review set through a sentiment dictionary;

[0016] The satisfaction calculation subunit is used to classify the review information based on the positive and good review set, the negative and good review set, the positive and credible review set, the negative and credible review set and the ordinary review set through the sentiment intensity analysis method to generate a judgment matrix.

[0017] Furthermore, the satisfaction judgment matrix unit further includes an emotional word set updating subunit;

[0018] The sentiment word set updating subunit is used to determine whether high-frequency words that are not classified into the positive and kind comment set, the negative and kind comment set, the positive and credible comment set, the negative and credible comment set, and the ordinary comment set are emphasis words. If not, they are classified into the positive and kind comment set, the negative and kind comment set, the positive and credible comment set, the negative and credible comment set, and the ordinary comment set through cluster analysis.

[0019] Furthermore, the judgment matrix generation module further includes an attention judgment matrix unit, and the commodity attributes further include commodity attention;

[0020] The attention judgment matrix unit calculates the product attention reflected in the comment information by using a word frequency inverse text frequency method, and generates a judgment matrix according to the product attention.

[0021] Furthermore, the judgment matrix generation module further includes a practicality judgment matrix unit, and the commodity attributes further include commodity practicality;

[0022] The utility judgment matrix unit uses a neural network model to calculate the utility of the product reflected in the review information and generates a judgment matrix based on the utility of the product.

[0023] In the above scheme, by using multiple models to determine the satisfaction, attention and practicality of the selected redemption products based on the product review information of external large shopping websites, it is possible to determine the satisfaction, attention and practicality values ​​of the recommended points redemption products even if the description information about the products on this website is not sufficient.

[0024] Furthermore, the hierarchical analysis model, whose target layer is the product recommendation score, the criterion layer is the product attributes, and the solution layer is the selected exchange product, is used to calculate the comparison matrix weight and the judgment matrix weight respectively through the comparison matrix and the judgment matrix, and calculate the product recommendation score through the comparison matrix weight and the judgment matrix weight.

[0025] Furthermore, the user propensity matrix generation module is configured to perform decision tree classification based on the user registration information and the types of the goods purchased this time to generate the comparison matrix when the user purchases goods exceeding a quantity threshold;

[0026] Alternatively, when a user purchases commodities with a quantity less than or equal to the quantity threshold, a decision tree classification is performed according to the user registration information and the commodity type with the highest single item amount to generate the comparison matrix.

[0027] Furthermore, the user registration information includes gender and year of birth;

[0028] The user propensity matrix is ​​used to perform decision tree classification into subcategories according to the gender, the birth year and the product type, and to generate the comparison matrix according to the correspondence between the subcategories and the product attributes.

[0029] In the above solution, the number of items purchased by the user is used to classify the user's gender, birth year, and item type to generate a comparison matrix. This allows for accurate classification of the types of items purchased by the user through a mature decision tree, fully considering the user's preferences for satisfaction, attention, and practicality.

[0030] Furthermore, the module for determining the commodities to be exchanged is configured to select the commodities to be exchanged within a first set amount range based on the highest single item amount when the user purchases commodities exceeding a quantity threshold, and to remove the commodities to be exchanged with an amount less than the amount corresponding to the accumulated points;

[0031] Alternatively, when a user purchases goods whose quantity is less than or equal to a threshold, the selected exchange goods within a second set amount range are selected based on the average amount of the goods, and the selected exchange goods whose amount is less than the amount corresponding to the accumulated points are removed.

[0032] In the above scheme, based on the different amounts of users' single shopping, redemption products in different amount ranges are recommended to users, thereby enhancing the user's sense of surprise.

[0033] Compared with the prior art, the present invention has the following advantages:

[0034] 1. Using the hierarchical analysis model, based on the characteristics of the user's shopping items and the comments on the items on external large shopping websites, we can comprehensively judge and determine the recommended points redemption items that match the user and meet the user's preferences. This enables a small online shopping SaaS system to derive appropriate recommended points redemption items based solely on the user's shopping items, without relying on user characteristics generated by multiple shopping and multiple use of the platform.

[0035] 2. Based on the number of items purchased by the user, the user's gender, birth year, and item type are classified according to the situation to generate a comparison matrix. This allows for accurate classification of the types of items purchased by the user through a mature decision tree, fully considering the user's preferences for satisfaction, attention, and practicality.

[0036] 3. Based on the different amounts of users' single purchases, we can recommend redemption products in different amount ranges to users, enhancing their sense of surprise. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a schematic diagram of the general structure of a SaaS system based on commodity exchange according to an embodiment of the present invention;

[0038] Figure 2 This is a schematic diagram of the general process of a SaaS system based on commodity exchange according to an embodiment of the present invention;

[0039] Figure 3 This is a schematic diagram of a judgment matrix generation process of a SaaS system based on commodity exchange according to an embodiment of the present invention;

[0040] Figure 4This is a schematic diagram of the hierarchical analysis model structure of the SaaS system based on commodity exchange in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0042] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0043] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0044] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0045] like Figures 1 to 4 As shown, the present invention provides a SaaS system based on commodity exchange, which adopts a hierarchical analysis model to make a comprehensive judgment on the characteristics of the user's shopping commodities and the comments on the commodities on external large shopping websites to determine the recommended points exchange commodities that match the user and meet the user's preferences, thereby realizing a small online shopping SaaS system that can obtain suitable recommended points exchange commodities based only on the user's shopping commodities, without relying on the user characteristics generated by multiple shopping and multiple use of the platform.

[0046] in, Figure 1 This is a schematic diagram of the general structure of a SaaS system based on commodity exchange according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the general process of a SaaS system based on commodity exchange according to an embodiment of the present invention; Figure 3 This is a schematic diagram of a judgment matrix generation process of a SaaS system based on commodity exchange according to an embodiment of the present invention; Figure 4This is a schematic diagram of the hierarchical analysis model structure of the SaaS system based on commodity exchange in an embodiment of the present invention.

[0047] like Figures 1 to 4 As shown, this embodiment proposes a SaaS system based on commodity exchange, including: a module for determining commodities to be exchanged, for selecting multiple commodities to be exchanged from the commodities purchased by the user this time and the exchange commodity library according to the purchase amount and accumulated points of the user's current shopping; a commodity review library module, interconnected with an external large shopping website, for matching and screening commodity reviews on the external large shopping website, and generating / updating the commodity review library; a judgment matrix generation module, connected to the module for determining commodities to be exchanged and the commodity review library module, for generating a judgment matrix reflecting commodity attributes based on the review information of the commodities to be exchanged in the commodity review library; a user tendency matrix generation module, connected to the module for determining commodities to be exchanged, for performing decision tree classification on the commodities purchased this time and the user registration information, and generating a comparison matrix reflecting the user's tendency towards the commodity attributes; a hierarchical analysis model, connected to the judgment matrix generation module and the user tendency matrix generation module, for calculating the judgment matrix and the comparison matrix using the hierarchical analysis method, and selecting recommended exchange commodities from the multiple commodities to be exchanged.

[0048] Specifically, the product review database is generated by matching the brand names, category names, and key benefits of products in the SaaS system with products on large external shopping websites. Product review information from these websites is automatically captured through crawlers, third-party data collection tools, or API calls. Data is then cleaned and processed to extract valuable information, remove duplicate data, address missing values, and convert data formats. This database addresses the issue of insufficient review information in small shopping SaaS systems, which cannot fully reflect product attributes.

[0049] The large shopping websites may be JD.com, Taobao, etc.

[0050] The accumulated points include points given to customers, corresponding points obtained by customers from shopping, etc.

[0051] Furthermore, the judgment matrix generation module includes a satisfaction judgment matrix unit, and the product attributes include product satisfaction; the satisfaction judgment matrix unit uses a sentiment intensity analysis method to calculate the product satisfaction reflected in the review information, and generates a judgment matrix based on the product satisfaction.

[0052] Furthermore, the satisfaction judgment matrix unit includes a word set generation subunit, a satisfaction calculation subunit and a matrix generation subunit; the word set generation subunit is used to generate a positive satisfaction comment set, a negative satisfaction comment set, a positive untrustworthy comment set, a negative untrustworthy comment set and an ordinary comment set through a sentiment dictionary; the satisfaction calculation subunit is used to classify the comment information based on the positive goodwill comment set, the negative goodwill comment set, the positive credible comment set, the negative credible comment set and the ordinary comment set through a sentiment intensity analysis method to generate a judgment matrix.

[0053] Furthermore, the satisfaction judgment matrix unit also includes a sentiment word set updating subunit; the sentiment word set updating subunit is used to judge whether the high-frequency words that are not classified into the positive and kind comment set, the negative and kind comment set, the positive and credible comment set, the negative and credible comment set and the ordinary comment set are emphasis words, and if not, classify them into the positive and kind comment set, the negative and kind comment set, the positive and credible comment set, the negative and credible comment set and the ordinary comment set through cluster analysis.

[0054] Specifically, a judgment matrix reflecting satisfaction is generated by comparing all review word sets on a large shopping website with the positive satisfaction review set, negative satisfaction review set, positive untrustworthy review set, negative untrustworthy review set, and general review set from the HowNet sentiment dictionary to obtain the positive satisfaction review set, negative satisfaction review set, positive untrustworthy review set, negative untrustworthy review set, and general review set for a particular product attribute Cj. If a word in the high-frequency word set Aj does not belong to any of the above sets, it is determined whether it is an emphasis word. If so, it is deleted from the high-frequency word set Aj. Otherwise, it is classified into the positive good review set, the negative good review set, the positive trustworthy review set, the negative trustworthy review set, and the general review set through cluster analysis.

[0055] Determine whether the number of words in the positive and satisfied review set, the negative and satisfied review set, the positive and untrustworthy review set, the negative and untrustworthy review set, and the ordinary review set is within the set number range. If not, delete the words in the larger set until the number of words in each set is within the set number range. The set number is set according to the specificity level of the product, and the specificity level is determined according to the total amount of the category to which the product belongs. For example, toilet paper and paper towels are daily necessities. There are more than 80 items of daily necessities in the SaaS system. Since 80 is greater than 40, its specificity level is the lowest, and the set number corresponds to 20. Therefore, by adjusting the word difference of the positive and kind review set, the negative and kind review set, the positive and credible review set, the negative and credible review set, and the ordinary review set through the specificity level, it is ensured that when it meets the product specificity characteristics, the scoring will not be inaccurate due to the difference in the number of words.

[0056] The sentiment intensity analysis method is used to classify the review information of each candidate redemption product in the product review database into a set of positive and satisfied reviews, a set of negative and satisfied reviews, a set of positive and untrustworthy reviews, a set of negative and untrustworthy reviews, and a set of general reviews to obtain a satisfaction level Pi. The satisfaction level Pi∈{2,-2,1,-1,0}, where 2,-2,1,-1,0 corresponds to the set with the largest number of positive and satisfied reviews, negative and satisfied reviews, positive and untrustworthy reviews, negative and untrustworthy reviews, and general reviews, thereby determining the sentiment intensity of the candidate redemption product. For example, if the review information of a paper towel candidate redemption product in the product review database contains the most words classified in the negative and positive review set, then the satisfaction level of the paper towel product is -2.

[0057] The ratio of the satisfaction ratings of the two selected exchange commodities is used as an element of the judgment matrix unit. This step is a conventional judgment matrix construction method of the hierarchical analysis model and will not be described in detail here.

[0058] It should be noted that the sentiment intensity analysis method is a support vector machine model (SVM), which is trained by a large number of product reviews on large shopping websites, so that it can be used to classify the review information in the product review library.

[0059] Furthermore, the judgment matrix generation module also includes an attention judgment matrix unit, and the product attributes include product attention; the attention judgment matrix unit uses the word frequency inverse text frequency method to calculate the product attention reflected in the comment information, and generates a judgment matrix based on the product attention.

[0060] Specifically, the term frequency inverse text frequency method multiplies the term frequency (TF) and inverse document frequency (IDF) of the selected redemption product to obtain its satisfaction, so as to combine the importance of the word in the document with its prevalence in the entire product review library to obtain a more accurate measurement value.

[0061] Term frequency (TF) indicates how often a word appears in the text of a product review database. This is a normalized frequency, calculated by dividing the number of occurrences by the total number of words in the article to prevent frequency bias towards longer documents. The more times a word appears in a document, the more important it is within that document. However, using term frequency alone as a measure of importance has its drawbacks: it tends to favor overly common words while ignoring more meaningful ones.

[0062] Inverse Document Frequency (IDF) indicates the prevalence of a keyword and is used to measure its prevalence across the entire product review database. The IDF is calculated by dividing the total number of documents in the corpus by the number of documents containing the term and then taking the logarithm of the resulting quotient. As you can see, if a term appears in many documents, its IDF will be low; conversely, it will be high. The purpose of IDF is to de-emphasize common terms and increase the weight of rare terms, thereby highlighting their importance.

[0063] The satisfaction level of a certain redemption product is divided by the total satisfaction level of all redemption products to obtain a normalized satisfaction level.

[0064] The ratio of the normalized satisfaction scores of the two selected exchange commodities is used as an element of the judgment matrix unit. This step is a conventional judgment matrix construction method of the hierarchical analysis model and will not be described in detail here.

[0065] Furthermore, the judgment matrix generation module also includes a practicality judgment matrix unit, and the product attributes include product practicality; the practicality judgment matrix unit uses a neural network model to calculate the product practicality reflected in the review information, and generates a judgment matrix based on the product practicality.

[0066] Specifically, the neural network model adopts the neural network technology of the text evaluation knowledge base of the Venn Consulting Company, and by establishing rules, the neural network model has an accuracy rate of about 95% in classifying and recognizing practical text evaluation information.

[0067] More specifically, the evaluation classification data used by Wien Consulting for training neural network models was quantitatively processed as follows: points were given according to the number of practicality category words contained in the review information, with each practicality category word receiving 1 point, positive evaluations receiving positive points, and negative evaluations receiving negative points, and the practicality being the sum of the absolute values ​​of the positive and negative evaluation scores.

[0068] Divide the utility of a particular exchangeable item by the total utility of all exchangeable items to obtain a normalized utility value. The ratio of the normalized utility values ​​of two exchangeable items is used as an element of the judgment matrix. This step is a conventional judgment matrix construction method for the hierarchical analysis model and will not be repeated here.

[0069] In the above scheme, by using multiple models to determine the satisfaction, attention and practicality of the selected redemption products based on the product review information of external large shopping websites, it is possible to determine the satisfaction, attention and practicality values ​​of the recommended points redemption products even if the description information about the products on this website is not sufficient.

[0070] Further, see Figure 3 The hierarchical analysis model, whose target layer is the product recommendation score, the criterion layer is the product attributes, and the solution layer is the selected redemption product, is used to calculate the comparison matrix weight and the judgment matrix weight respectively through the comparison matrix and the judgment matrix, and calculate the product recommendation score through the comparison matrix weight and the judgment matrix weight.

[0071] Specifically, the hierarchical analysis model performs a normalization test on the comparison matrix weights and the judgment matrix weights, that is, calculates the maximum eigenvalue root, the consistency index CI and the test coefficient CR. If the test coefficient CR is less than 0.1, the comparison matrix weights and the judgment matrix weights pass the consistency test; otherwise, the product review library is updated, and the sentiment word set update subunit is updated.

[0072] The number of keywords corresponding to the second-level risk with the most keywords in the financial negative vocabulary level library is determined by deleting the keywords with the most repeated words in the second-level risk keywords.

[0073] Furthermore, the user propensity matrix is ​​used to perform decision tree classification based on the user registration information and the types of goods purchased this time to generate the comparison matrix when the user purchases goods exceeding the quantity threshold; the user propensity matrix is ​​also used to perform decision tree classification based on the user registration information and the type of goods with the highest amount in the current shopping when the user purchases goods less than or equal to the quantity threshold to generate the comparison matrix.

[0074] Furthermore, the user registration information includes gender and birth year; the user preference matrix is ​​used to perform decision tree classification into subcategories based on the gender, birth year and product type, and generate the comparison matrix based on the correspondence between the subcategories and the product attributes.

[0075] Specifically, the decision tree classification is XGBoost, which is a machine learning algorithm based on the gradient boosting decision tree GBDT (Gradient Boosting Decision Tree), which is trained based on residuals using the gradient descent method. After the first decision tree is trained based on the available data and target variables, the residual of the tree is calculated based on the difference between the predicted value and the actual value. The second tree is trained using the residual as the target, and subsequent decision trees are trained based on the residual of the previous round of decision trees. The purpose of data regression is achieved by continuously reducing the residuals generated during the training process. The final predicted value is the sum of the target value predicted by the first tree and the product of the learning rate (LearningRate) and the residual of all other trees.

[0076] In this embodiment, preference arrays of gender, birth year, product category, satisfaction, attention, and practicality are trained on XGBoost so that a small online shopping SaaS system can accurately classify and determine user preferences.

[0077] First, we use multiple categories of product types to determine whether the shopper's gender matches the gender in their registered information. If they match, we then classify them based on their registered birth year. Each birth year corresponds to a set of satisfaction, attention, and practicality scores, and the ratio of these two product attributes is used as an element in the comparison matrix. For example, a male user born in the 1980s would score satisfaction, attention, and practicality as 1, 1, and 2, so toilet paper products would likely not be recommended for redemption.

[0078] In the above solution, the number of items purchased by the user is used to classify the user's gender, birth year, and item type to generate a comparison matrix. This allows for accurate classification of the types of items purchased by the user through a mature decision tree, fully considering the user's preferences for satisfaction, attention, and practicality.

[0079] Furthermore, the purchase amount includes the highest single item amount and the average amount of goods; the module for determining goods to be exchanged is used to select the goods to be exchanged within the first set amount range according to the highest single item amount when the user purchases goods exceeding the quantity threshold; the module for determining goods to be exchanged is also used to select the goods to be exchanged within the second set amount range according to the average amount of goods when the user purchases goods less than or equal to the quantity threshold; the module for determining goods to be exchanged is also used to remove the goods to be exchanged whose amount is less than the amount corresponding to the accumulated points.

[0080] In the above scheme, based on the different amounts of users' single shopping, redemption products in different amount ranges are recommended to users, thereby enhancing the user's sense of surprise.

[0081] It is understandable that the use of a hierarchical analysis model allows the characteristics of the user's current purchase and the reviews of the products on external large shopping websites to be comprehensively judged to determine recommended points redemption products that match the user and meet the user's preferences. This enables a small online shopping SaaS system to derive appropriate recommended points redemption products based solely on the user's current purchase, without relying on user characteristics generated by multiple purchases and platform usage. Based on the number of items purchased by the user, the user's gender, birth year, and product type are categorized to generate a comparison matrix. This allows for accurate classification of the types of products purchased by the user through a mature decision tree, fully considering the user's preferences for satisfaction, attention, and practicality. Based on the different amounts of the user's single purchase, redemption products of different amount ranges are recommended to the user, enhancing the user's sense of surprise.

[0082] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

[0083] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A SaaS system based on commodity exchange, characterized in that: include: The module for determining the items to be exchanged is used to select multiple items to be exchanged from the user's shopping items and exchange item library based on the highest single item price, average item price, and accumulated points of the user's current purchase; The product review database module is connected to external large-scale shopping websites to match and filter product reviews on external large-scale shopping websites and generate / update the product review database; A judgment matrix generation module, connected to the to-be-selected exchange commodity determination module and the commodity review library module, for generating a judgment matrix reflecting commodity attributes based on the review information of the to-be-selected exchange commodity in the commodity review library; A user preference matrix generation module, connected to the module for determining the items to be redeemed, is used to perform a decision tree classification on the items purchased this time and the user registration information to generate a comparison matrix reflecting the user's preference for the attributes of the items; A hierarchical analysis model, connected to the judgment matrix generation module and the user propensity matrix generation module, calculates the judgment matrix and the comparison matrix using the hierarchical analysis method, and selects a recommended redemption product from the plurality of redemption products to be selected; The commodity attributes include commodity satisfaction, commodity attention and commodity practicality, and the judgment matrix generation module includes a satisfaction judgment matrix unit, an attention judgment matrix unit and a practicality judgment matrix unit; The satisfaction judgment matrix unit calculates the product satisfaction reflected in the review information using a sentiment intensity analysis method, and generates a judgment matrix based on the product satisfaction; The attention judgment matrix unit calculates the product attention reflected in the review information using a word frequency inverse text frequency method, and generates the judgment matrix according to the product attention; The utility judgment matrix unit uses a neural network model to calculate the utility of the product reflected in the review information and generates the judgment matrix based on the utility of the product; The user propensity matrix generation module is used to perform decision tree classification based on the user registration information and the product types of the shopping items when the user purchases goods exceeding the quantity threshold, and generate the comparison matrix; or, when the user purchases goods less than or equal to the quantity threshold, perform decision tree classification based on the user registration information and the product type with the highest single item price, and generate the comparison matrix.

2. The SaaS system based on commodity exchange according to claim 1 is characterized in that: The hierarchical analysis model, whose target layer is the product recommendation score, the criterion layer is the product attributes, and the solution layer is the selected redemption product, is used to calculate the comparison matrix weight and the judgment matrix weight respectively through the comparison matrix and the judgment matrix, and calculate the product recommendation score through the comparison matrix weight and the judgment matrix weight.

3. The SaaS system based on commodity exchange according to claim 1 is characterized in that: The user registration information includes gender and birth year; The user propensity matrix is ​​used to perform decision tree classification into subcategories according to the gender, the birth year and the product type, and to generate the comparison matrix according to the correspondence between the subcategories and the product attributes.

4. The SaaS system based on commodity exchange according to claim 3 is characterized in that: The module for determining the commodities to be exchanged is configured to select the commodities to be exchanged within a first set amount range based on the highest single item amount when the user purchases commodities exceeding a quantity threshold, and to remove commodities to be exchanged with an amount less than the amount corresponding to the accumulated points; Alternatively, when a user purchases goods whose quantity is less than or equal to a threshold, the selected exchange goods within a second set amount range are selected based on the average amount of the goods, and the selected exchange goods whose amount is less than the amount corresponding to the accumulated points are removed.

Images