E-commerce platform purchase order evaluation management method and system based on intelligent evaluation
Through intelligent evaluation technology, intelligent identification and automatic classification are carried out for unevaluated orders on e-commerce platforms, and intelligent correction is carried out in combination with historical evaluation data and supplier praise, which solves the problem of untimely order evaluation and improves evaluation management efficiency and transaction promotion effect.
Patent Information
- Application Number
- CN202411642242.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-18
- Publication Date
- 2025-02-28
AI Technical Summary
In the prior art, order evaluation cannot be carried out in time after the purchase of e-commerce platforms is completed, resulting in low evaluation rates and affecting transactions and supply chain improvements.
The purchase order evaluation management method of e-commerce platform based on intelligent evaluation is adopted. By obtaining the associated feature content of unevaluated orders, the intelligent evaluation training model is established and optimized, business personnel are automatically reminded to conduct evaluation, and intelligent correction is made based on historical evaluation data and supplier praise.
It has achieved timely and intelligent evaluation of purchase orders on e-commerce platforms, improved evaluation management efficiency, and promoted transaction conclusions and continuous improvement of supply chains.
Smart Images
Figure CN119539599A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of e-commerce platform order evaluation management, and in particular to an e-commerce platform purchase order evaluation management method and system based on intelligent evaluation. Background Art
[0002] With the rapid development of e-commerce, product evaluation on e-commerce platforms has become an important reference indicator for people when shopping online. After the purchase is completed on the e-commerce platform, order evaluation is required. Order evaluation has two major functions: 1. Sales value. Almost all customers will browse product evaluations before shopping on the e-commerce platform. The good reviews of historical customers will greatly promote the completion of transactions; 2. Suppliers can obtain user evaluation data from it, clearly understand the advantages and disadvantages of their own products and services, and guide the continuous improvement of the supplier industry chain. In the existing technology, there is a problem that the order cannot be evaluated in time or is not evaluated after the user completes the purchase. Summary of the invention
[0003] In view of the problems existing in the above-mentioned prior art, the present invention provides an e-commerce platform purchase order evaluation management method and system based on intelligent evaluation, which realizes timely reminder and intelligent evaluation management of e-commerce platform purchase order evaluation. The technical solution of the present invention is as follows:
[0004] In a first aspect, a method for managing purchase order evaluation on an e-commerce platform based on intelligent evaluation is provided, comprising the following steps:
[0005] For the orders that have been shipped by the supplier, obtain the unevaluated orders;
[0006] For the unevaluated orders, a message is sent to the business personnel to remind them to evaluate the orders;
[0007] Before or after sending the information to the business personnel, an intelligent evaluation is performed on the unevaluated order.
[0008] In some implementations, the intelligent evaluation of the unevaluated order includes:
[0009] Obtain historically evaluated orders and corresponding evaluation contents as training data;
[0010] Obtaining the associated feature content of the order, wherein the associated feature content includes the order title, the content of the order product introduction page, and the product type of the order;
[0011] Dividing the associated feature content into a plurality of submodule contents, and determining a mapping relationship between the submodule content and the evaluation content based on each of the submodule contents;
[0012] Generate corresponding semantic labels for sub-module contents and evaluation contents with mapping relationships;
[0013] Based on the semantic labels, the submodule contents with mapping relationships and the evaluation contents as samples, training is performed to obtain a first model;
[0014] Based on the order to be evaluated, the associated feature content of the order is obtained, and the associated feature content is divided into a plurality of sub-module contents; recorded as the second sub-module content, input into the first model, and obtain the second sub-module encoding data;
[0015] Record the submodule content corresponding to the historical evaluated order as the first submodule content, input the first submodule content into the first model, and obtain the first submodule encoding data;
[0016] Based on the evaluation content corresponding to the first submodule encoding data similar to the second submodule encoding data as the candidate evaluation content of the second submodule content;
[0017] The key feature information appearing in the first submodule content is deleted from the candidate evaluation content, and the evaluation content of the order to be evaluated is obtained based on the deleted candidate evaluation content and the associated feature content of the order to be evaluated, which is recorded as the first evaluation content.
[0018] In some embodiments, the first model is used to encode sub-module contents. During training, the model parameters of the first model are optimized by taking the maximum similarity of the encoded data corresponding to the sub-module contents with the same semantic labels and similar evaluation contents and the minimum similarity of the encoded data corresponding to the sub-module contents with different labels and dissimilar evaluation contents as the objective function.
[0019] In some implementations, dividing the associated feature content into a plurality of submodule contents, and determining a mapping relationship between the submodule contents and the evaluation content based on each of the submodule contents, includes:
[0020] Preliminarily dividing the associated feature content into a plurality of initial sub-module contents, and splitting the evaluation content into a plurality of initial sub-comment fields according to semantic unity;
[0021] Continuously optimize the splitting and merging of initial sub-comment fields based on the initial sub-module content;
[0022] At the same time, the splitting and merging of the initial submodules are continuously optimized with the currently determined sub-comment fields;
[0023] Taking the minimum correlation between different mapping relationships as the objective function, the final multiple sub-module contents and multiple sub-comment fields, as well as the mapping relationship between the sub-module contents and the evaluation content are obtained.
[0024] In some implementations, the intelligent evaluation of the unevaluated order further includes:
[0025] Based on the supplier name of the unevaluated order, all purchase orders with the supplier name are obtained in the historical purchase orders;
[0026] Determine the degree of praise for the materials supplied by the supplier based on the evaluation contents of all purchase orders of the supplier;
[0027] The first evaluation content is modified based on the degree of praise, the positive evaluation angle, and the negative evaluation angle.
[0028] In some embodiments, determining the degree of praise for the materials supplied by the supplier based on the evaluation content of all purchase orders of the supplier includes:
[0029] Determine whether the evaluation is positive or negative based on the evaluation content, and determine the favorable rating based on the proportion of positive evaluations; when the evaluation content contains both positive and negative evaluation fields, the evaluation content is not included in the proportion analysis process;
[0030] Based on the evaluation content, keywords are extracted, and the evaluation angles and corresponding evaluation adjectives in all positive evaluation fields are summarized, and the evaluation angles and corresponding evaluation adjectives in all negative evaluation fields are summarized.
[0031] In some implementations, the intelligent evaluation further comprises the following steps:
[0032] Intelligently identify and automatically classify the unevaluated orders;
[0033] For similar orders, perform batch intelligent evaluation.
[0034] In some implementations, the intelligent identification and automatic classification of the unevaluated orders includes:
[0035] Get order content information, get the supplier name, purchase material name and material attribute information corresponding to the order;
[0036] Semantic recognition is performed based on the name of the purchased materials and the attribute information of the materials, and the unevaluated orders are classified based on preset classification rules or the unevaluated orders are automatically classified based on the similarity of the information related to the purchased materials.
[0037] In some embodiments, the unevaluated orders are classified based on preset classification rules, including: based on all review data of all materials on the mall, a product and all evaluation data corresponding to the product are used as a review set, and the products in at least one review set with similar review content in the review set are classified into one category; the unevaluated orders are automatically classified based on the similarity of information related to purchased materials, including: the unevaluated orders are automatically classified based on the category to which the materials belong.
[0038] In the second aspect, an e-commerce platform purchase order evaluation management system based on intelligent evaluation is provided, including:
[0039] The un-evaluated order unit is used to obtain the un-evaluated orders among the orders that have been shipped by the supplier;
[0040] An order reminder unit, for sending information to the business personnel for reminding the business personnel to evaluate the order with respect to the unevaluated order;
[0041] The order intelligent evaluation unit is used to perform intelligent evaluation on the unevaluated order before or after sending information to the business personnel.
[0042] The present invention provides an e-commerce platform purchase order evaluation management method and system based on intelligent evaluation, which has the following beneficial effects: the present invention implements intelligent evaluation of unevaluated orders that meet preset conditions before sending information to business personnel according to pre-settings, and then sends information to business personnel for the remaining unevaluated orders to remind them to evaluate the orders; or after sending information to business personnel, intelligent evaluation is performed on unevaluated orders that meet the conditions according to the feedback and selection of business personnel. The evaluation management of e-commerce platform purchase orders is realized, and the efficiency of the evaluation management of e-commerce platform purchase orders is improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] Figure 1 It is a flow chart of a method for evaluating and managing purchase orders on an e-commerce platform based on intelligent evaluation in an embodiment of the present application;
[0044] Figure 2 It is a flow chart of a method for intelligently evaluating unevaluated orders in an embodiment of the present application;
[0045] Figure 3 It is a flowchart of a method for obtaining a mapping relationship between associated feature content and evaluation content of an order in an embodiment of the present application;
[0046] Figure 4 It is a structural diagram of an e-commerce platform purchase order evaluation management system based on intelligent evaluation in an embodiment of the present application. DETAILED DESCRIPTION
[0047] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.
[0048] In order to improve service standards and quality, reduce costs, improve work efficiency, and promote the development of the mall business, we have developed an e-commerce platform purchase order evaluation management method and system based on intelligent evaluation by introducing a new generation of digital technology. It uses automation and intelligent technology to realize automatic reminders and intelligent evaluation of completed purchase orders that have not been evaluated. Specifically, see Figure 1 The embodiment of the present application provides an e-commerce platform purchase order evaluation management method based on intelligent evaluation, comprising the following steps:
[0049] Step 1: For the orders that have been shipped by the supplier, obtain the unevaluated orders;
[0050] Step 2: For the unevaluated order, a message is sent to the business personnel to remind them to evaluate the order;
[0051] Step 3, before or after sending the information to the business personnel, includes intelligently evaluating the unevaluated order.
[0052] In the embodiment of the present application, it is possible to perform intelligent evaluation on unevaluated orders that meet the preset conditions before sending information to the business personnel according to the pre-set settings, and then send information to the business personnel for the remaining unevaluated orders to remind the business personnel to evaluate the orders; it is also possible to perform intelligent evaluation on unevaluated orders that meet the conditions according to the feedback and selection of the business personnel after sending the information to the business personnel. The evaluation management of the purchase orders of the e-commerce platform is realized. It is understandable that for unevaluated orders, it is possible to set to send reminder information to be evaluated to the business personnel regularly or irregularly.
[0053] In one embodiment, the e-commerce platform purchase order evaluation management method based on intelligent evaluation includes the following steps:
[0054] For the orders that have been shipped by the supplier, obtain the unevaluated orders;
[0055] For the unevaluated orders, a message is sent to the business personnel to remind them to evaluate the orders;
[0056] According to user feedback and selection, the unevaluated orders are intelligently evaluated.
[0057] In one embodiment, see Figure 2 In the above step 3, the intelligent evaluation of the unevaluated order includes the following steps:
[0058] Step 301, obtaining historical evaluated orders and corresponding evaluation contents as training data;
[0059] Step 302, obtaining the associated characteristic content of the order, wherein the associated characteristic content includes the order title, the content of the order product introduction page, and the product type of the order;
[0060] Step 303, dividing the associated feature content into a plurality of sub-module contents, and determining a mapping relationship between the sub-module content and the evaluation content based on each of the sub-module contents;
[0061] Step 304, generating corresponding semantic tags for the submodule contents and evaluation contents having a mapping relationship;
[0062] Step 305, training is performed based on the semantic labels, the submodule contents with mapping relationships, and the evaluation contents as samples to obtain a first model;
[0063] Step 306, based on the order to be evaluated, obtain the associated feature content of the order, divide the associated feature content into multiple sub-module contents; record it as the second sub-module content, input the first model, and obtain the second sub-module encoding data;
[0064] Step 307, record the submodule content corresponding to the historical evaluated order as the first submodule content, input the first submodule content into the first model, and obtain the first submodule encoding data;
[0065] Step 308, using the evaluation content corresponding to the first submodule encoding data similar to the second submodule encoding data as candidate evaluation content of the second submodule content;
[0066] Step 309, delete the key feature information appearing in the first submodule content from the candidate evaluation content, and obtain the evaluation content of the order to be evaluated based on the deleted candidate evaluation content and the associated feature content of the order to be evaluated, which is recorded as the first evaluation content.
[0067] In the embodiment of the present application, for the existing order products and the corresponding evaluation contents as training data, it can be understood that the associated feature contents of the order products and the corresponding evaluation contents have a mapping relationship, the associated feature contents are detailed features, and the evaluation contents are general features, and a local field content of the associated feature contents or a combination of multiple local field contents corresponds to a local field content of an evaluation content. A first model is trained based on the training data, and the first model can encode the different associated feature contents of the order products differently. The evaluation contents corresponding to the associated feature contents with similar contents encoded by the first model are used as candidate evaluation contents. Considering that specific evaluation fields for specific products may appear in the candidate evaluation contents, the key feature information appearing in the content of the first submodule is deleted in the present application to avoid affecting the evaluation contents generated by the second submodule.
[0068] When evaluating the orders to be evaluated, the evaluation is not based on the combination of existing evaluation content of the order products, but is based on the combination of evaluation content of all order products. The evaluation content of the sub-module content is generated based on the evaluation content corresponding to the similar sub-module coding content, which increases the diversity of the evaluation content during the intelligent evaluation of the order products. When obtaining historical evaluated orders for similarity analysis, the selection of historical evaluated orders is randomly determined, so as to achieve non-repetitive evaluation content during the intelligent evaluation of the same order product (orders with the same associated feature content such as order title, order product introduction page content, and order product style).
[0069] In one embodiment, in the above step 305, the first model is used to encode the sub-module content. During training, the model parameters of the first model are optimized by taking the maximum similarity of the encoded data corresponding to the sub-module contents with the same semantic label and similar evaluation content and the minimum similarity of the encoded data corresponding to the sub-module contents with different labels and dissimilar evaluation content as the objective function.
[0070] In one embodiment, in step 309, the evaluation content of the order to be evaluated is obtained based on the deleted candidate evaluation content and the associated feature content of the order to be evaluated, which is recorded as the first evaluation content, wherein the deleted candidate evaluation content is used as the reference content of the evaluation content of the order to be evaluated, and the evaluation content reference content and the associated feature content of the order to be evaluated are input into the trained model together, the evaluation content reference content and the associated feature content of the order to be evaluated are fused, and the attention mechanism is used to select important key features to generate the evaluation content of the order to be evaluated. It can be understood that the trained model can be obtained through model training based on the training samples, and is used to output the evaluation content of the order to be evaluated based on the evaluation content reference content and the associated feature content of the order to be evaluated as input.
[0071] In the embodiment of the present application, the first model encodes the sub-module content and uses the objective function in the training process to achieve characteristics in which the encoded data significantly distinguishes the contents of different sub-modules.
[0072] In one embodiment, see Figure 3 In the above step 303, the associated feature content is divided into a plurality of sub-module contents, and based on each of the sub-module contents, a mapping relationship between the sub-module contents and the evaluation contents is determined, including the following steps:
[0073] Step 3031, preliminarily dividing the associated feature content into a plurality of initial sub-module contents, and splitting the evaluation content into a plurality of initial sub-comment fields according to semantic unity;
[0074] Step 3032, continuously optimizing the splitting and merging of the initial sub-comment fields based on the initial sub-module contents;
[0075] Step 3033, while continuously optimizing the splitting and merging of the initial submodules with the currently determined sub-comment fields;
[0076] Step 3034, taking the minimum correlation between different mapping relationships as the objective function, obtains the final plurality of sub-module contents and plurality of sub-comment fields, and the mapping relationship between the sub-module contents and the evaluation contents.
[0077] It can be understood that the associated feature content is divided into multiple sub-module contents, and there can be multiple division methods. The evaluation content is divided into multiple sub-comment fields, and there are also multiple division methods. In the embodiment of the present application, the mapping relationship after division is used, and the correlation between different mapping relationships is minimized as the objective function to obtain the optimal division method. For example, the multiple sub-module contents after the associated feature content is divided are respectively recorded as A1, A2, A3..., and the evaluation content is divided into multiple sub-comment fields and recorded as B1, B2, B3..., assuming that A1 and B1 are mapped, A2 and B2 are mapped, and A3 and B3 are mapped, then the correlation between the combined content of A1 and B1 and the combined content of A2 and B2 is the smallest. In other words, under the above assumptions, the semantic correlation between the divided A1 and the divided B1 should be the largest, and the A1 after the division should be the smallest with the remaining B2, B3..., and the semantic correlation between the divided B1 and the divided A1 should be the largest, and the B1 after the division should be the smallest with the remaining A2, A3. That is, "the minimum correlation between different mapping relationships" can be determined based on the minimum correlation between the combined content of sub-modules and sub-comment fields in different mapping relationships, and further can be determined based on the maximum correlation between the sub-module content and sub-comment fields belonging to the same mapping relationship and the minimum correlation between the sub-module content and sub-comment fields not belonging to the same mapping relationship.
[0078] In one embodiment, in the above step 3, performing intelligent evaluation on the unevaluated order further includes the following steps:
[0079] Step 311, based on the supplier name of the unevaluated order, obtaining all purchase orders of the supplier name in the historical purchase orders;
[0080] Step 312, determining the degree of praise for the materials supplied by the supplier based on the evaluation contents of all purchase orders of the supplier;
[0081] Step 313, modifying the first evaluation content based on the degree of praise, the positive evaluation angle, and the negative evaluation angle.
[0082] In an embodiment of the present application, the above-mentioned first evaluation content is also corrected and optimized based on the evaluation content of the historical purchase orders of the order supplier, and the historical quality of the supplier's supply materials and user evaluation are considered as reference data for the evaluation content of the unevaluated order.
[0083] In one embodiment, in the above step 312, determining the favorable rating of the materials supplied by the supplier based on the evaluation content of all purchase orders of the supplier includes the following steps:
[0084] Step 3121, determining whether the evaluation content is a positive evaluation or a negative evaluation, and determining the favorable evaluation degree based on the proportion of positive evaluations; when the evaluation content contains both positive evaluation and negative evaluation fields, the evaluation content is not included in the proportion analysis process;
[0085] Step 3122, based on the evaluation content, extract keywords, summarize the evaluation angles and corresponding evaluation adjectives in all positive evaluation fields, and summarize the evaluation angles and corresponding evaluation adjectives in all negative evaluation fields.
[0086] In the embodiment of the present application, the quality of the supplier's historical supplies is analyzed based on the proportion of positive reviews. It can be understood that if the favorable rating is high, the fields, semantics, and emotions of the evaluation content for the order to be evaluated tend to represent high satisfaction. When the evaluation content contains both positive and negative evaluation fields, the evaluation content is not included in the proportion analysis process.
[0087] When aggregating and collecting positive evaluation fields and negative evaluation fields, evaluation content that contains both positive evaluation fields and negative evaluation fields can also be included in the collection process. The positive evaluation fields and negative evaluation fields in the evaluation content are extracted through semantic recognition models, sentiment recognition models, etc., and the advantages and disadvantages of the material products are analyzed through the summary results, thereby assisting in the correction and improvement of the first evaluation content.
[0088] In one embodiment, the above-mentioned step 313 modifies the first evaluation content based on the degree of praise, positive evaluation angle, and negative evaluation angle, including: based on the negative evaluation angle, deleting the field associated with the negative evaluation angle in the first evaluation content, and the associated field includes the negative evaluation angle and the corresponding evaluation adjective.
[0089] In one embodiment, the above-mentioned step 313, which corrects the first evaluation content based on the praise degree, positive evaluation angle, and negative evaluation angle, also includes: correcting the content in the first evaluation content that is semantically inconsistent with the positive evaluation content corresponding to the praise degree and positive evaluation angle based on the praise degree and positive evaluation angle.
[0090] In one implementation, the intelligent evaluation in step 3 above further includes the following steps:
[0091] Step 321, intelligently identifying and automatically classifying the unevaluated orders;
[0092] Step 322, perform batch intelligent evaluation for similar orders.
[0093] In an embodiment of the present application, orders that require automatic evaluation can be classified and evaluated in batches according to the user's pre-set settings or the user's selections, and orders can be manually checked and intelligently evaluated in batches, reducing the workload of manual evaluation.
[0094] In one embodiment, in the above step 321, intelligent identification and automatic classification of the unevaluated orders include:
[0095] Step 3211, obtain order content information, obtain the supplier name, purchase material name and material attribute information corresponding to the order;
[0096] Step 3212, perform semantic recognition based on the name of the purchased materials and the attribute information of the materials, classify the unevaluated orders based on preset classification rules, or automatically classify the unevaluated orders based on the similarity of the information related to the purchased materials.
[0097] In one embodiment, in the above step 3212, unevaluated orders are classified based on preset classification rules, including: based on all review data of all materials on the mall, a product and all evaluation data corresponding to the product are used as a review set, and the products in at least one review set with similar review content in the review set are classified into one category.
[0098] Specifically, in the embodiment of the present application, the similarity of material commodities is reversely analyzed from the similarity of commodity review contents, and then the evaluation orders are classified. For two or more commodities with similar review contents, it can be preliminarily considered that the two or more commodity orders can be classified into one category and evaluated in one batch.
[0099] In one embodiment, in the above step 3212, the unevaluated orders are automatically classified based on the similarity of the information related to the purchased materials, including: automatically classifying the unevaluated orders based on the categories to which the materials belong. Specifically, the categories to which the materials belong can be determined based on the relevant information of the purchased materials, for example, the categories include tools (files, hammers, wrenches, etc.), office supplies, primary equipment, secondary equipment, instruments, etc., and the unevaluated orders are classified based on the categories to which the materials belong.
[0100] See also Figure 4 The embodiment of the present application also provides an e-commerce platform purchase order evaluation management system based on intelligent evaluation, the system comprising:
[0101] The un-evaluated order unit is used to obtain the un-evaluated orders among the orders that have been shipped by the supplier;
[0102] An order reminder unit, for sending information to the business personnel for reminding the business personnel to evaluate the order with respect to the unevaluated order;
[0103] The order intelligent evaluation unit is used to perform intelligent evaluation on the unevaluated order before or after sending information to the business personnel.
[0104] For the specific definition of the e-commerce platform purchase order evaluation management system, please refer to the definition of the e-commerce platform purchase order evaluation management method above, which will not be repeated here. Each unit in the above-mentioned e-commerce platform purchase order evaluation management system can be implemented in whole or in part by software, hardware and their combination. The above-mentioned units can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned units.
[0105] In some embodiments, the e-commerce platform purchase order evaluation management system provided by the embodiments of the present invention can be implemented by a combination of software and hardware. As an example, the e-commerce platform purchase order evaluation management system provided by the embodiments of the present invention can be directly embodied as a combination of software modules executed by a processor. The software module can be located in a storage medium, and the storage medium is located in a memory. The processor reads the executable instructions included in the software module in the memory, and combines with the necessary hardware (for example, including a processor and other components connected to a bus) to complete the e-commerce platform purchase order evaluation management method provided by the embodiments of the present invention.
[0106] The present invention is not limited to the above-mentioned specific implementation modes. Various changes made by ordinary technicians in this field based on the above-mentioned concepts without creative work are all within the protection scope of the present invention.
Claims
1. A method for evaluating and managing purchase orders on an e-commerce platform based on intelligent evaluation, characterized in that: The steps include: For the orders that have been shipped by the supplier, obtain the unevaluated orders; For the unevaluated orders, a message is sent to the business personnel to remind them to evaluate the orders; Before or after sending the information to the business personnel, an intelligent evaluation is performed on the unevaluated order.
2. The method for managing the evaluation of purchase orders on an e-commerce platform based on intelligent evaluation according to claim 1, characterized in that: The intelligent evaluation of the unevaluated order comprises: Obtain historically evaluated orders and corresponding evaluation contents as training data; Obtaining the associated feature content of the order, wherein the associated feature content includes the order title, the content of the order product introduction page, and the product type of the order; Dividing the associated feature content into a plurality of submodule contents, and determining a mapping relationship between the submodule content and the evaluation content based on each of the submodule contents; Generate corresponding semantic labels for sub-module contents and evaluation contents with mapping relationships; Based on the semantic labels, the submodule contents with mapping relationships and the evaluation contents as samples, training is performed to obtain a first model; Based on the order to be evaluated, the associated feature content of the order is obtained, and the associated feature content is divided into a plurality of sub-module contents; recorded as the second sub-module content, input into the first model, and obtain the second sub-module encoding data; Record the submodule content corresponding to the historical evaluated order as the first submodule content, input the first submodule content into the first model, and obtain the first submodule encoding data; Based on the evaluation content corresponding to the first submodule encoding data similar to the second submodule encoding data as the candidate evaluation content of the second submodule content; The key feature information appearing in the first submodule content is deleted from the candidate evaluation content, and the evaluation content of the order to be evaluated is obtained based on the deleted candidate evaluation content and the associated feature content of the order to be evaluated, which is recorded as the first evaluation content.
3. The method for managing the evaluation of purchase orders on an e-commerce platform based on intelligent evaluation according to claim 2 is characterized in that: The first model is used to encode the sub-module contents. During training, the model parameters of the first model are optimized by taking the maximum similarity of the encoded data corresponding to the sub-module contents with the same semantic labels and similar evaluation contents and the minimum similarity of the encoded data corresponding to the sub-module contents with different labels and dissimilar evaluation contents as the objective function.
4. The method for managing the evaluation of purchase orders on an e-commerce platform based on intelligent evaluation according to claim 2, characterized in that: The step of dividing the associated feature content into a plurality of submodule contents and determining a mapping relationship between the submodule content and the evaluation content based on each of the submodule contents includes: Preliminarily dividing the associated feature content into a plurality of initial sub-module contents, and splitting the evaluation content into a plurality of initial sub-comment fields according to semantic unity; Continuously optimize the splitting and merging of initial sub-comment fields based on the initial sub-module content; At the same time, the splitting and merging of the initial submodules are continuously optimized with the currently determined sub-comment fields; Taking the minimum correlation between different mapping relationships as the objective function, the final multiple sub-module contents and multiple sub-comment fields, as well as the mapping relationship between the sub-module contents and the evaluation content are obtained.
5. The method for managing the evaluation of purchase orders on an e-commerce platform based on intelligent evaluation according to claim 2, characterized in that: The intelligent evaluation of the unevaluated order further includes: Based on the supplier name of the unevaluated order, all purchase orders with the supplier name are obtained in the historical purchase orders; Determine the degree of praise for the materials supplied by the supplier based on the evaluation contents of all purchase orders of the supplier; The first evaluation content is modified based on the degree of praise, the positive evaluation angle, and the negative evaluation angle.
6. The method for managing the evaluation of purchase orders on an e-commerce platform based on intelligent evaluation according to claim 5 is characterized in that: Determining the degree of praise for the materials supplied by the supplier based on the evaluation content of all purchase orders of the supplier includes: Determine whether the evaluation is positive or negative based on the evaluation content, and determine the favorable rating based on the proportion of positive evaluations; when the evaluation content contains both positive and negative evaluation fields, the evaluation content is not included in the proportion analysis process; Based on the evaluation content, keywords are extracted, and the evaluation angles and corresponding evaluation adjectives in all positive evaluation fields are summarized, and the evaluation angles and corresponding evaluation adjectives in all negative evaluation fields are summarized.
7. The method for managing the evaluation of purchase orders on an e-commerce platform based on intelligent evaluation according to claim 1, characterized in that: The intelligent evaluation further comprises the following steps: Intelligently identify and automatically classify the unevaluated orders; For similar orders, perform batch intelligent evaluation.
8. The method for managing the evaluation of purchase orders on an e-commerce platform based on intelligent evaluation according to claim 7 is characterized in that: The intelligent identification and automatic classification of the unevaluated orders includes: Get order content information, get the supplier name, purchase material name and material attribute information corresponding to the order; Semantic recognition is performed based on the name of the purchased materials and the attribute information of the materials, and the unevaluated orders are classified based on preset classification rules or the unevaluated orders are automatically classified based on the similarity of the information related to the purchased materials.
9. The method for managing the evaluation of purchase orders on an e-commerce platform based on intelligent evaluation according to claim 8, characterized in that: The unevaluated orders are classified based on the preset classification rules, including: according to all the review data of all the materials on the mall, a product and all the evaluation data corresponding to the product are used as a review set, and the products in at least one review set with similar review content in the review set are classified into one category; The automatically classifying the unevaluated orders based on the similarity of the related information of the purchased materials includes: automatically classifying the unevaluated orders based on the categories to which the materials belong.
10. An e-commerce platform purchase order evaluation management system based on intelligent evaluation, characterized in that: include: The un-evaluated order unit is used to obtain the un-evaluated orders among the orders that have been shipped by the supplier; An order reminder unit, for sending information to the business personnel for reminding the business personnel to evaluate the order with respect to the unevaluated order; The order intelligent evaluation unit is used to perform intelligent evaluation on the unevaluated order before or after sending information to the business personnel.