Product-based feedback information processing method and device, equipment and storage medium

Abstract

The invention provides a product-based feedback information processing method and device, equipment and a storage medium, and relates to the field of artificial intelligence, in particular to the field of big data and natural language processing. According to the specific implementation scheme, feedback information of a user is obtained; wherein the feedback information represents the use condition of the product by the user; performing validity analysis on the feedback information according to a preset vector database to obtain an analysis result; wherein the preset vector database comprises multiple pieces of vector data, the vector data are preset invalid information, the validity analysis represents that the validity of the semantics of the feedback information is analyzed, and the analysis result represents that the feedback information is valid or invalid; if the analysis result shows that the feedback information is valid, determining an information category to which the feedback information belongs; wherein the information category represents the effect of the feedback information on product improvement. The feedback information of the user is automatically processed, and the classification efficiency and precision of the feedback information are improved.

Description

Technical Field

[0001] The present disclosure relates to the field of big data and natural language processing in the field of artificial intelligence, and in particular to a method, device, equipment and storage medium for processing product feedback information. Background Art

[0002] With the rapid development of information technology and the continuous deepening of user experience, user feedback, as an important bridge connecting users and product development teams, plays a vital role in product optimization. By analyzing user feedback information, we can determine the aspects that need to be improved in the product, so as to better serve users.

[0003] Traditional feedback information processing methods often have problems such as low efficiency and inaccurate classification. Therefore, it is necessary to perform refined automatic processing of feedback information to meet the needs of rapid iteration of product development. Summary of the invention

[0004] The present disclosure provides a method, device, equipment and storage medium for processing product-based feedback information.

[0005] According to a first aspect of the present disclosure, a method for processing product-based feedback information is provided, comprising:

[0006] Obtaining user feedback information; wherein the feedback information represents the user's use of the product;

[0007] According to a preset vector database, the feedback information is analyzed for validity to obtain an analysis result; wherein the preset vector database includes a plurality of vector data, the vector data is preset invalid information, the validity analysis represents the analysis of the semantic validity of the feedback information, and the analysis result represents whether the feedback information is valid or invalid;

[0008] If the analysis result indicates that the feedback information is valid, then the information category to which the feedback information belongs is determined; wherein the information category indicates the role of the feedback information in improving the product.

[0009] According to a second aspect of the present disclosure, there is provided a device for processing feedback information based on a product, comprising:

[0010] An acquisition unit, used to acquire user feedback information; wherein the feedback information represents the user's use of the product;

[0011] An analysis unit, configured to perform validity analysis on the feedback information according to a preset vector database to obtain an analysis result; wherein the preset vector database includes a plurality of vector data, the vector data is preset invalid information, the validity analysis represents the analysis of the validity of the semantics of the feedback information, and the analysis result represents whether the feedback information is valid or invalid;

[0012] A determination unit is used to determine the information category to which the feedback information belongs if the analysis result indicates that the feedback information is valid; wherein the information category indicates the role of the feedback information in improving the product.

[0013] According to a third aspect of the present disclosure, there is provided an electronic device, including:

[0014] at least one processor; and

[0015] a memory communicatively coupled to the at least one processor;

[0016] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect of the present disclosure.

[0017] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute the method according to the first aspect of the present disclosure.

[0018] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising a computer program, which implements the steps of the method described in the first aspect of the present disclosure when executed by a processor.

[0019] According to the technology disclosed in the present invention, the processing efficiency and accuracy of feedback information are improved.

[0020] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0022] Figure 1 It is a flowchart of a method for processing product feedback information provided according to an embodiment of the present disclosure;

[0023] Figure 2 It is a flowchart of a method for processing product feedback information provided according to an embodiment of the present disclosure;

[0024] Figure 3 It is a flowchart of a method for processing product feedback information provided according to an embodiment of the present disclosure;

[0025] Figure 4 It is a flowchart of a method for processing product feedback information provided according to an embodiment of the present disclosure;

[0026] Figure 5 is a structural block diagram of a device for processing product feedback information provided according to an embodiment of the present disclosure;

[0027] Figure 6 is a structural block diagram of a device for processing product feedback information provided according to an embodiment of the present disclosure;

[0028] Figure 7 is a block diagram of an electronic device for implementing the method for processing product-based feedback information according to an embodiment of the present disclosure;

[0029] Figure 8 The present invention is a block diagram of an electronic device for implementing the method for processing product-based feedback information according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0030] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0031] With the rapid development of information technology and the continuous deepening of user experience, the status of application testing business in the software development life cycle is becoming increasingly prominent. In the process of application testing, user feedback information is an important bridge connecting users and product development teams. Its effectiveness and timeliness play a vital role in improving product quality and optimizing user experience. However, traditional feedback information processing methods often have problems such as low efficiency and inaccurate classification, which are difficult to meet the needs of rapid iteration of product development. For example, traditional user feedback processing methods often rely on manual analysis and classification, which are inefficient and easily affected by subjective factors, resulting in inaccurate classification results of feedback information. Alternatively, pre-set matching rules for information categories, based on a preset set of rules, match keywords or phrases in feedback information to classify it. Rules can be formulated based on common user feedback patterns or problem types. Although rule matching methods can improve processing efficiency, their accuracy often depends on the completeness and adaptability of rule formulation. For complex or novel feedback information, rule matching methods cannot accurately classify it.

[0032] By automatically processing user feedback information, we can reduce reliance on manual analysis and lower operating costs. In recent years, LLM (Large Language Model) has shown broad application prospects in many fields with its powerful natural language processing and deep learning capabilities. Especially in the quantification of user experience, the large language model can automatically extract key information and classify it by learning and understanding a large amount of user feedback information, thereby providing more accurate and valuable feedback to the product development team.

[0033] The present invention discloses a method, device, equipment and storage medium for processing product-based feedback information, which are applied to the fields of big data and natural language processing in the field of artificial intelligence to improve the processing efficiency and accuracy of feedback information.

[0034] It should be noted that the model in this embodiment is not a model for a specific user and cannot reflect the personal information of a specific user. It should be noted that the data in this embodiment comes from a public data set.

[0035] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0036] In order to enable readers to more deeply understand the implementation principle of the present disclosure, the following Figure 1-Figure 8 The embodiment is further refined.

[0037] Figure 1FIG. 0 is a schematic flowchart of a method for processing feedback information based on a product according to an embodiment of the present disclosure, which can be executed by a processing device for feedback information based on a product. As Figure 1 shown, the method includes the following steps:

[0038] S101. Obtain the feedback information of the user; wherein, the feedback information characterizes the user's usage of the product.

[0039] Exemplarily, the user uses a software product and can submit feedback information regarding their own usage of the product, that is, the feedback information can characterize the user's usage of the product. For example, the feedback information can include the user's usage experience, functional requirements, problem reports, and improvement suggestions of the product. The product can be a mobile application product, a web application product, a desktop application product, etc. The user can submit feedback information by posting comments, sending emails, etc.

[0040] The feedback information of the user can be collected through multiple terminals or multiple channels. For example, the feedback information can be collected through mobile phones, tablets, computer terminals, and various IoT (Internet of Things) devices, etc. The feedback information can include information such as the product name, feedback time, and feedback description content. The feedback information of the user can be obtained in real time or at regular intervals. For example, the feedback information of the user can be obtained once every 24 hours. The feedback information of each user regarding the product can be obtained and processed for that user, or the feedback information of all users regarding the product can be obtained and processed for all users.

[0041] After obtaining the feedback information, the feedback information can be preprocessed, and the preprocessed feedback information can be used to execute steps S102 and S103. The preprocessing steps can include one or more operations such as data cleaning, text tokenization, and stop word removal. These operations together ensure the data quality and accuracy in the subsequent steps. Data cleaning can effectively remove redundant, incorrect, or irrelevant data in the feedback information, ensuring the purity and consistency of the data. Text tokenization can split continuous text information into independent lexical units, providing a basis for subsequent natural language processing. The stop word removal operation can further filter out common words that have no actual contribution to semantic analysis, such as "de" (的), "le" (了), etc., thereby highlighting the core information in the text. Through these preprocessing operations, not only can the processing efficiency of the large language model for the feedback information be improved, but also the depth of the model's understanding of the content of the feedback information can be enhanced, providing a solid data foundation for the subsequent classification of the feedback information.

[0042] S102. Perform validity analysis on the feedback information according to a preset vector database to obtain analysis results; wherein the preset vector database includes multiple vector data, the vector data is preset invalid information, the validity analysis represents the analysis of the semantic validity of the feedback information, and the analysis result represents whether the feedback information is valid or invalid.

[0043] Exemplarily, a vector database is pre-constructed, and the vector database may include multiple vector data, and the vector data is preset invalid information. Invalid information may refer to invalid feedback information, that is, multiple invalid feedback information is pre-set, and each preset invalid feedback information may correspond to a vector data, and a vector database is composed of multiple vector data. Invalid information may be feedback information without valid semantics, for example, it may be non-bug (program vulnerability), unclear description, incoherent sentence or garbled feedback information. Various documents such as internal knowledge documents, professional field research materials, technical manuals, user guides, and industry reports can be collected in advance to construct a vector database. For example, the collected data can be processed by slicing cleaning and text vectorization to obtain a vector database.

[0044] The vector database includes a variety of invalid information. Through the invalid information in the preset vector database, the feedback information in S101 can be analyzed for validity, or the feedback information after preprocessing can be analyzed for validity, so as to determine whether the feedback information is valid or invalid. The validity analysis can represent the analysis of the semantic validity of the feedback information, and the analysis result obtained after the analysis can represent whether the feedback information is valid or invalid.

[0045] The vector data in the vector database can be compared with the feedback information. If the feedback information is consistent with the vector data in the vector database, the feedback information is considered invalid; if not, the feedback information is considered valid. Alternatively, the similarity between the feedback information and the vector data can be determined. If the similarity is large, the feedback information is considered invalid; if the similarity is small, the feedback information is considered valid.

[0046] In this embodiment, the feedback information may also be directly subjected to semantic analysis. If the semantics can be analyzed, the feedback information is considered valid; if the semantics cannot be analyzed, the feedback information is considered invalid. Table 1 is an example table of invalid feedback information in this embodiment.

[0047] Table 1 Example of invalid feedback information

[0048] Invalid feedback information Invalid reason Report as a known issue Known problem on the business side, no need to fix it yet Feedback meets product design Not a bug, in line with product design Network stability issues This is not a bug, but an occasional network problem. Unclear description The problem description is vague and difficult to reproduce

[0049] S103: If the analysis result indicates that the feedback information is valid, determine the information category to which the feedback information belongs; wherein the information category indicates the role of the feedback information in product improvement.

[0050] Exemplarily, if the feedback information is determined to be invalid, the feedback information is no longer processed in step S103, that is, the feedback information does not need to be classified and the feedback information is discarded; if the feedback information is determined to be valid, the information category to which the feedback information belongs can be further determined, and the information category can characterize the role of the feedback information in product improvement. For example, the information category can include functional requirements, strategy logic, information display, interaction design, etc., that is, through the feedback information, the functional requirements, strategy logic, information display, interaction design, etc. of the product can be improved.

[0051] Multiple information categories can be pre-set to determine which information category the feedback information conforms to. For example, keywords that can express semantics can be extracted from the feedback information, and if the keywords are consistent with a certain information category, it is determined that the feedback information belongs to that information category.

[0052] In this embodiment, it also includes: obtaining the information category to which the feedback information determined within a preset time period belongs; generating a category analysis document based on the information category to which the feedback information determined within the preset time period belongs; wherein the category analysis document represents the result of information category analysis of the feedback information of the product within the preset time period.

[0053] Specifically, after determining the information category of the feedback information, the information category corresponding to the feedback information is recorded and stored. A time period is preset, and the information category to which each piece of feedback information determined within the preset time period belongs is obtained at regular intervals. For example, a batch of feedback information may be obtained in batches every 24 hours, and the information category of the feedback information may be determined, and the determined information categories may be obtained once every 24 hours.

[0054] Generate a category analysis document based on the information category to which the feedback information determined within a preset time period belongs. The category analysis document is a report analyzing the information category, which can represent the results of the information category analysis of the feedback information of the product within a preset time period. For example, the category analysis document may include the classification results of each piece of feedback information, the number of feedback information of various information categories, typical cases in the feedback information, and product optimization suggestions.

[0055] In this embodiment, the information categories may include "usable category" and "easy-to-use category", etc. "usable category" means that the product meets the basic functional requirements, users can use the product to complete specific tasks or achieve goals, the basic functions of the product are clear, easy to understand and operate, and there are no obvious defects or obstacles; "easy-to-use category" means that the product also needs to optimize the user experience. On the basis of meeting the basic functional requirements, the product should have a better user experience. Easy-to-use products can allow users to complete tasks or achieve goals easily and efficiently, and have good operating procedures and interactive design. At the same time, the interface and visual design of the product should also be able to provide a pleasant experience. The category analysis document shows the number and proportion of feedback information in the two categories of available and easy-to-use through detailed data statistics and in-depth analysis. This classification and statistics not only helps application testers quickly understand the overall situation of user feedback, but also guides them to formulate optimization strategies for specific problems. For example, if the number of easy-to-use feedback is small, testers may need to focus on the user experience design of the product or service to improve user satisfaction and loyalty.

[0056] Category analysis documents can also provide detailed summary reports for each information category, including cluster analysis of user's specific feedback content, sentiment, usage scenarios and other problem descriptions. This information provides valuable insights for testers, allowing them to have a deeper understanding of user needs and expectations, so as to more accurately meet the requirements of user experience in the design and improvement of products or services.

[0057] The beneficial effect of this setting is that the category analysis document provides a detailed summary report for each information category, digging deep into key information such as users' specific feedback content, emotional tendencies, and usage scenarios. Through cluster analysis of these problem descriptions, similar or related feedback information can be clustered, allowing testers to more clearly identify the main problems and trends in each category. The category analysis document can also include product experience improvement suggestions for each information category. These suggestions are based on an in-depth understanding of user feedback and the results of cluster analysis, and are intended to guide testers to optimize and improve products or services from the perspective of user experience. These suggestions may involve adjustments to the interface design, enhancement of functions, simplification of operating procedures, etc., aiming to improve the product's usability, ease of use, and overall user satisfaction, and to improve the efficiency of product development and improvement.

[0058] In the disclosed embodiment, user feedback information on the product is obtained, and a vector database is pre-constructed. The vector database includes a plurality of vector data representing invalid information. The feedback information is analyzed for validity through the preset vector database to determine whether the feedback information is valid or invalid, thereby avoiding the invalid feedback information from affecting the information classification. If the feedback information is determined to be valid, the information category to which the feedback information belongs is further determined, so that the product is improved and optimized according to the information category. There is no need for manual processing and classification of the feedback information, which reduces the impact of human subjective factors and saves manpower and time. The automated processing of feedback information is achieved, which improves the efficiency and accuracy of the classification of feedback information, thereby improving the efficiency of product development.

[0059] Figure 2 A flowchart of a method for processing product-based feedback information provided by an embodiment of the present disclosure.

[0060] In this embodiment, the feedback information is analyzed for validity according to a preset vector database to obtain an analysis result, including: determining the similarity between the feedback information and the vector data in the preset vector database through a preset large language model; wherein the similarity represents the degree of association between the feedback information and the vector data; and performing a validity analysis on the feedback information based on a preset first prompt word according to the similarity between the feedback information and the vector data in the preset vector database to obtain an analysis result of the feedback information; wherein the preset first prompt word is the prompt word information in the preset large language model, which is used to determine the analysis result of the feedback information.

[0061] like Figure 2 As shown, the method comprises the following steps:

[0062] S201. Obtain user feedback information; wherein the feedback information represents the user's use of the product.

[0063] Exemplarily, this step may refer to the above-mentioned step S101 and will not be described in detail.

[0064] S202: Determine the similarity between the feedback information and the vector data in a preset vector database through a preset large language model; wherein the similarity represents the degree of association between the feedback information and the vector data.

[0065] Exemplarily, a large language model is pre-built and trained. The large language model is a natural language processing model of deep learning. In this embodiment, the model structure of the large language model is not specifically limited. When training the large language model, we introduced SFT (Supervised Fine-Tuning) technology to further improve the performance of the model. A large amount of pre-collected feedback information can be used to perform parameter fine-tuning training on the large language model. Through supervised fine-tuning, it is ensured that the model can capture more subtle differences and more accurate intentions when understanding feedback information, so that the model can more deeply understand the semantics and intentions of the feedback information, and obtain a trained large language model.

[0066] The large language model has demonstrated excellent generalization performance and accuracy in many fields such as natural language processing. By fine-tuning feedback information and adopting advanced reinforcement learning methods, the model's ability to understand user intent has been further improved, providing strong support for the management of feedback information, and helping companies to process and analyze user feedback information more efficiently.

[0067] In this embodiment, the feedback information can be input into the large language model, and the similarity between the feedback information and each piece of vector data can be calculated through the large language model. The similarity can characterize the degree of association or similarity between the feedback information and the vector data. For example, the large language model can perform semantic analysis on the feedback information to obtain a feature vector of the feedback information, compare the feature vector of the feedback information with the vector data, and obtain the similarity between the feedback information and the vector data. In this embodiment, the calculation process of the similarity is not specifically limited. For example, it can be implemented by using Euclidean distance, cosine distance or other vector similarity calculation methods.

[0068] S203. According to the similarity between the feedback information and the vector data in the preset vector database, based on the preset first prompt word, perform validity analysis on the feedback information to obtain the analysis result of the feedback information; wherein the preset first prompt word is the prompt word information in the preset large language model, which is used to determine the analysis result of the feedback information.

[0069] Exemplarily, during the application of the large language model, model reasoning can be performed in combination with a preset prompt. Prompt is a carefully selected and designed prompt word information used to guide the behavior of the large language model when processing a specific task. A prompt is pre-set as the first prompt word, which can be used to determine the analysis result of the feedback information. The first prompt word is input into the large language model so that the large language model can evaluate the effectiveness of the feedback information based on certain effectiveness standards and output formats. By making rational use of prompts, we can further improve the directionality and accuracy of the large language model when processing feedback information.

[0070] After obtaining the similarity between the feedback information and the vector data, the feedback information can be analyzed for validity based on the similarity and through the large language model and the first prompt word to obtain the analysis result of the feedback information. For example, the first prompt word is provided with a validity analysis rule, and the large language model can determine whether the feedback information is valid based on the similarity corresponding to the first prompt word and the feedback information, and output the judgment result of whether it is valid.

[0071] A complete prompt may include the role positioning of the large model, task description, examples, output format, and precautions. Among them, for the large model role, the prompt usually clarifies the role of the large language model, that is, the identity that the model should play in a specific task. For example, in this embodiment, the model is designated as an "experienced product manager" responsible for providing accurate and relevant information. This role definition helps the model better adapt to its task environment and produce output that meets the expectations of the role. For the task description, the task description is the core part of the prompt, which details the specific tasks that the model needs to perform. The task description needs to be clear and specific enough to ensure that the model can accurately understand its goals. For the output format, the output format is another important component of the prompt, which specifies the structure and style of the model output. For example, the output format may include parts such as the title, introduction, body, and conclusion. By clarifying the output format, the text generated by the model can be ensured to be consistent and readable. For other precautions, in addition to the above, the prompt may also include some other precautions to remind the model of additional factors that need to be considered when processing tasks. These factors can include privacy protection, data security, etc. By setting these considerations in the prompt, you can help the model avoid potential problems and risks when generating output. In order for the model to better understand the task requirements and output format, the prompt can also include some examples so that the large model can refer to the examples for reasoning.

[0072] In this embodiment, the feedback information is input into a preset large model, and the similarity of the input feedback information is determined based on the preset first prompt word, and then the effectiveness analysis is performed to automatically determine the valid feedback information and invalid feedback information in the feedback information, thereby reducing the manual processing process and improving the efficiency and accuracy of subsequent classification.

[0073] In this embodiment, according to the similarity between the feedback information and the vector data in a preset vector database, based on the preset first prompt word, the feedback information is analyzed for effectiveness, and the analysis result of the feedback information is obtained, including: according to the similarity between the feedback information and the vector data in the preset vector database, determining the target vector corresponding to the feedback information; wherein the target vector represents the vector data that is closely associated with the feedback information; according to the target vector corresponding to the feedback information and the preset first prompt word, the feedback information is analyzed for effectiveness, and the analysis result of the feedback information is obtained.

[0074] Specifically, the large language model can determine the similarity between the feedback information and each vector data, and determine one or more vector data from these vector data according to the similarity corresponding to each vector data as the target vector. The target vector refers to the vector data that is closely associated with the feedback information. For example, the vector data corresponding to the high similarity can be determined as the target vector. That is, the large language model will perform efficient vector retrieval operations in the constructed vector database, and based on the similarity or correlation between the feedback information and the vector data, it will quickly locate and extract the K retrieval results that are most relevant to the feedback information from the massive vector database as the target vector. K is a natural number greater than or equal to 1.

[0075] After determining the target vector corresponding to the feedback information, the large language model will refer to the effectiveness evaluation criteria clearly specified in the first prompt word, and conduct a strict effectiveness evaluation on the feedback information based on the retrieved target vector, and accurately divide the feedback information into two categories: effective and invalid. For example, the evaluation can be performed based on the number of target vectors, or based on the similarity corresponding to the target vectors.

[0076] The beneficial effect of this arrangement is that the feedback information is associated with the vector data to determine whether there is vector data highly associated with the feedback information, that is, the target vector. Combined with the first prompt word, the feedback information is preliminarily screened through the large language model to obtain effective feedback information, thereby improving the processing efficiency and accuracy of the feedback information.

[0077] In this embodiment, the target vector corresponding to the feedback information is determined based on the similarity between the feedback information and the vector data in the preset vector database, including: if the similarity between the feedback information and the vector data in the preset vector database is greater than a preset first threshold, then the vector data in the preset vector database is determined to be the target vector corresponding to the feedback information.

[0078] Specifically, a similarity threshold is pre-set as the first threshold. Each calculated similarity is compared with the first threshold. If the similarity corresponding to the determined vector data is greater than the preset first threshold, the vector data is determined to be the target vector; if the similarity corresponding to the determined vector data is less than or equal to the preset first threshold, the vector data is determined not to be the target vector. For example, if the preset first threshold is 80%, all vector data with a similarity greater than 80% are determined to be the target vector.

[0079] If each similarity is less than or equal to the preset first threshold, it is determined that the target vector does not exist, that is, the feedback information is not similar to each vector data in the vector database, and the feedback information is valid.

[0080] The beneficial effect of such a setting is that by comparing the similarities, the target vector can be quickly obtained, thereby improving the efficiency of effectiveness analysis and further improving the efficiency of subsequent information classification.

[0081] In this embodiment, the feedback information is analyzed for effectiveness based on the target vector corresponding to the feedback information and the preset first prompt word to obtain an analysis result of the feedback information, including: determining the number of target vectors corresponding to the feedback information based on the preset first prompt word; if it is determined that the number of target vectors corresponding to the feedback information is less than a preset quantity threshold, then determining that the analysis result of the feedback information is valid.

[0082] Specifically, the task description of the first prompt word can be set with evaluation criteria or rules for effectiveness analysis. For example, the first prompt word requires the large model to determine the number of target vectors. If the number of target vectors is large, the feedback information is considered invalid; if the number of target vectors is small, the feedback information is considered valid.

[0083] That is to say, the large model determines the task it needs to perform through the preset first prompt word. For example, the number of target vectors can be determined through the preset first prompt word. A quantity threshold is preset, and the number of target vectors is compared with the preset quantity threshold. If it is determined that the number of target vectors is less than the preset quantity threshold, the analysis result of the feedback information is determined to be valid; if it is determined that the number of target vectors is equal to or greater than the preset quantity threshold, the analysis result of the feedback information is determined to be invalid.

[0084] The beneficial effect of such a setting is that, by determining the number of target vectors, the effectiveness of the feedback information can be analyzed, thereby avoiding the subsequent classification of invalid feedback information and improving the efficiency and accuracy of information classification.

[0085] In this embodiment, a target vector corresponding to the feedback information is determined based on the similarity between the feedback information and the vector data in a preset vector database, including: sorting the vector data in the preset vector database based on the similarity between the feedback information and each vector data in the preset vector database; and determining the vector data located within a preset sorting position range as the target vector corresponding to the feedback information.

[0086] Specifically, the magnitude of the similarity corresponding to each piece of vector data is determined, and the vector data is sorted according to the magnitude of the similarity to obtain a sorting result. For example, the vector data may be sorted in descending order of similarity.

[0087] A preset sorting position range, for example, the preset sorting position range is a sorting range of the top K positions. According to the sorting result, the vector data within the preset sorting position range is determined as the target vector corresponding to the feedback information. For example, the vector data in the top K positions are all determined as the target vector.

[0088] The beneficial effect of this setting is that the large model performs efficient vector retrieval operations in the constructed vector database, quickly locates and extracts the top K retrieval results that are most relevant to the feedback information, thereby improving the efficiency of effectiveness evaluation.

[0089] In this embodiment, the feedback information is analyzed for validity based on the target vector corresponding to the feedback information and the preset first prompt word to obtain an analysis result of the feedback information, including: if it is determined based on the preset first prompt word that the similarities corresponding to the target vectors are greater than a preset second threshold, then the analysis result of the feedback information is determined to be invalid.

[0090] Specifically, a similarity threshold is pre-set as the second threshold, and the first threshold and the second threshold may be the same or different. The task description in the first prompt word may be to compare the similarity corresponding to the target vector with the second threshold, and determine whether the feedback information is valid based on the comparison result. For example, the similarity of the target vector ranked last may be compared with the second threshold. If the similarity is greater than the preset second threshold, that is, the similarity corresponding to each target vector is greater than the preset second threshold, then it can be determined that even the smallest similarity in the target vector is highly correlated with the feedback information, and the feedback information may be considered invalid; if there is a similarity in the target vector that is less than or equal to the preset second threshold, then the analysis result of the feedback information is determined to be valid.

[0091] The beneficial effect of such a setting is that only the K similarities are compared with the second threshold, thereby improving the efficiency of validity analysis and further improving the efficiency of information classification.

[0092] S204: If the analysis result indicates that the feedback information is valid, determine the information category to which the feedback information belongs; wherein the information category indicates the role of the feedback information in product improvement.

[0093] Exemplarily, this step may refer to the above-mentioned step S103 and will not be described in detail.

[0094] In the disclosed embodiment, user feedback information on the product is obtained, and a vector database is pre-constructed. The vector database includes a plurality of vector data representing invalid information. The feedback information is analyzed for validity through the preset vector database to determine whether the feedback information is valid or invalid, thereby avoiding the invalid feedback information from affecting the information classification. If the feedback information is determined to be valid, the information category to which the feedback information belongs is further determined, so that the product is improved and optimized according to the information category. There is no need for manual processing and classification of the feedback information, which reduces the impact of human subjective factors and saves manpower and time. The automated processing of feedback information is achieved, which improves the efficiency and accuracy of the classification of feedback information, thereby improving the efficiency of product development.

[0095] Figure 3 A flowchart of a method for processing product-based feedback information provided by an embodiment of the present disclosure.

[0096] In this embodiment, determining the information category to which the feedback information belongs includes: inputting the feedback information into a preset large language model, and determining the initial category to which the feedback information belongs based on a preset second prompt word; wherein the preset second prompt word is prompt word information in the preset large language model, and is used to determine the initial category of the feedback information, and the initial category represents a parent category of the information category to which the feedback information belongs; and determining the information category to which the feedback information belongs based on the initial category to which the feedback information belongs and the preset large language model.

[0097] like Figure 3 As shown, the method comprises the following steps:

[0098] S301. Obtain user feedback information; wherein the feedback information represents the user's use of the product.

[0099] Exemplarily, this step may refer to the above-mentioned step S101 and will not be described in detail.

[0100] S302. Perform validity analysis on the feedback information according to a preset vector database to obtain analysis results; wherein the preset vector database includes multiple vector data, the vector data is preset invalid information, the validity analysis represents the analysis of the semantic validity of the feedback information, and the analysis result represents whether the feedback information is valid or invalid.

[0101] Exemplarily, this step may refer to the above-mentioned step S102 and will not be described in detail.

[0102] S303. If the analysis result indicates that the feedback information is valid, the feedback information is input into a preset large language model, and based on a preset second prompt word, an initial category to which the feedback information belongs is determined; wherein the preset second prompt word is prompt word information in the preset large language model, and is used to determine the initial category of the feedback information, and the initial category represents a parent category of the information category to which the feedback information belongs.

[0103] Exemplarily, a prompt is pre-set as the second prompt word, which is also the prompt word information in the large language model. The content of the second prompt word is different from that of the first prompt word. The second prompt word can be used to determine the initial category of the feedback information. The initial category refers to the parent category of the information category to which the feedback information belongs, that is, the initial category is a coarse-grained category, and the information category is a fine-grained category.

[0104] After determining that the feedback information is valid, the valid feedback information is input into the large language model. The large language model performs coarse-grained classification of the feedback information according to the preset second prompt word to obtain the initial category to which the feedback information belongs. For example, the initial category may be "usable category" or "easy to use category".

[0105] That is to say, the effective feedback information screened in S302 is input into the pre-trained large language model, and the designed second prompt word is used to guide the large model to perform deep natural language processing. The large model will refer to the classification criteria of the usable / easy-to-use categories clearly specified in the second prompt word to evaluate each input feedback information. Through this step, the large model can accurately divide the feedback data into two categories: usable and easy-to-use. The classification criteria of usable and easy-to-use are based on user experience. User experience refers to all the feelings of users before, during and after using a product or system, including emotions, beliefs, preferences, cognitive impressions, physiological and psychological reactions, behaviors and achievements.

[0106] The second prompt may also include content such as the large model role, task description, examples, output format, and precautions. In order for the model to better understand the task requirements and output format, the prompt usually includes some examples. In this embodiment, cases of available classes and easy-to-use classes can be set in the second prompt to show the model the behavior patterns that should be expected in a given task. By learning these examples, the model can grasp the key points of the task more quickly and generate output data that meets expectations. The following is an example of the content in the second prompt. The second prompt can be:

[0107] Your role: You are an experienced product manager

[0108] Your task: You received a lot of feedback information in a crowd test, and now you need to judge this feedback information: You need to judge step by step whether the feedback category belongs to the [usable category] or the [easy-to-use category] according to the [judgment process] and [classification basis] I gave, mainly based on the content after the [feedback description] or [reproduction steps], and output the classification results strictly according to the [output format].

[0109] ###

Classification basis

[0110] Usability: The feedback is more objective, which means that the product cannot meet the basic functional requirements of users and there are obvious defects, obstacles or errors in the use process. It is intolerable to fail to complete the original main functions designed by the program.

[0111] Usability: Feedback is subjective, meaning that the user experience is not pleasant during use. The basic functions of the program can run, but the user experience is not good. Although it does not meet expectations, it is barely acceptable. It may be an incidental problem outside the main function. Experience problems and product suggestions are more likely to be usability issues. The overall impact is relatively small.

[0112] ###[Judgment process]:

[0113] 1. The number in front of the feedback information is the number of this feedback information, and the content after it is regarded as the content of one feedback information.

[0114] 2. Determine whether this piece of feedback meets the definition of usefulness

[0115] 3. Determine whether this feedback information meets the definition of available classes

[0116] 4. If both "easy to use" and "usable" are met, whichever one do you think accounts for a larger proportion, then you will judge this feedback information as the category with a larger proportion.

[0117] ###The example is as follows

[0118] User feedback:

[0119] 2. On the PPT editing page, click on the full screen AI input interface. There is a box in the dark mode, which should not exist.

[0120] 4. When generating PPT, pull down the page to view the generated page, and the page will flash.

[0121] The output is:

[0122] 2. Easy to use

[0123] 4. Easy to use

[0124] ###

Output format

[0125] Number, available / useful (line break)

[0126] Number, available / easy to use

[0127] ###Please note:

[0128] 1. The output does not need to present or repeat the user feedback

[0129] 2. The output result does not require answering the reason and basis for judgment

[0130] 3. Output results strictly follow the output format, do not include content other than numbering and usable / easy to use

[0131] 4. Please follow the output result number provided by the user.

[0132] S304: Determine the information category to which the feedback information belongs according to the initial category to which the feedback information belongs and a preset large language model.

[0133] Exemplarily, after obtaining the initial category to which the feedback information belongs, the feedback information and the initial category to which the feedback information belongs can be input into the large language model, and the large language model outputs the information category to which the feedback information belongs, and the final classification result is obtained. The subcategories existing under various parent categories can be preset, and after determining the initial category, the various subcategories preset under the initial category are determined. Determine which subcategory the feedback information belongs to, that is, obtain the information category to which the feedback information belongs. For example, the initial category is "easy-to-use category", and its subordinate subcategories include "functional requirements", "strategy logic", "information display" and "interaction design", etc. The feedback information can be semantically analyzed to determine which subcategory the analyzed semantic information is more in line with, thereby obtaining the final information category.

[0134] In this embodiment, the feedback information is first divided into major categories, and then divided into minor categories, so as to gradually determine the information categories and improve the classification accuracy of the feedback information. The division into minor categories can more comprehensively understand the user's specific evaluation of product characteristics, and then formulate corresponding optimization measures for different classification results to improve the efficiency of product optimization.

[0135] In this embodiment, determining the information category to which the feedback information belongs according to the initial category to which the feedback information belongs and a preset large language model includes: determining the third prompt word corresponding to the initial category to which the feedback information belongs according to the association relationship between the preset initial category and the third prompt word; wherein the preset third prompt word is prompt word information in the preset large language model, and is used to determine the information category to which the feedback information belongs; inputting the feedback information into the preset large language model, and determining the category information to which the feedback information belongs based on the third prompt word corresponding to the initial category to which the feedback information belongs.

[0136] Specifically, a prompt is pre-set as the third prompt word, and the content of the third prompt word is different from the content of the first prompt word and the content of the second prompt word. The third prompt word can be used to determine the information category to which the feedback information belongs.

[0137] Different parent categories may correspond to different methods of determining subcategories, that is, different initial categories may correspond to different third prompt words. The association relationship between the initial category and the third prompt word is stored in advance, and the third prompt word corresponding to the initial category to which the feedback information belongs is determined according to the preset association relationship. The feedback information and the corresponding third prompt word are input into the large language model, and the large language model performs semantic understanding of the feedback information through the third prompt word, and outputs the category information to which the feedback information belongs. For example, the initial category is "easy-to-use category", and the information category is "functional requirements" under the "easy-to-use category", indicating that the feedback information indicates whether the functional requirements of the product are reasonable, such as whether the functional requirements involved belong to the basic functions within the core value proposition of the page, or whether they belong to the functions that should be expanded under the competition trend.

[0138] The beneficial effect of this setting is that fine-grained category division is carried out according to the large model and the third prompt word, and the advantages of the large language model in natural language processing and deep learning are fully utilized to achieve efficient and accurate classification of user feedback, thereby improving the efficiency of product development and optimization.

[0139] In the disclosed embodiment, user feedback information on the product is obtained, and a vector database is pre-constructed. The vector database includes a plurality of vector data representing invalid information. The feedback information is analyzed for validity through the preset vector database to determine whether the feedback information is valid or invalid, thereby avoiding the invalid feedback information from affecting the information classification. If the feedback information is determined to be valid, the information category to which the feedback information belongs is further determined, so that the product is improved and optimized according to the information category. There is no need for manual processing and classification of the feedback information, which reduces the impact of human subjective factors and saves manpower and time. The automated processing of feedback information is achieved, which improves the efficiency and accuracy of the classification of feedback information, thereby improving the efficiency of product development.

[0140] Figure 4 The following is a flow chart of a method for processing product feedback information provided by an embodiment of the present disclosure. Figure 4 As shown, the method comprises the following steps:

[0141] S401: Obtain preprocessed feedback information.

[0142] S402: Retrieve vector data from a vector database through the large language model, and the vector database returns a target vector to the large language model.

[0143] S403, the large language model determines whether the feedback information is valid through the target vector and the preset first prompt word, if not, execute S404; if yes, execute S405.

[0144] S404: The feedback information is invalid, and the processing flow of the feedback information is terminated.

[0145] S405: The feedback information is valid. The initial category of the feedback information is determined through the large language model and the preset second prompt word.

[0146] Exemplarily, the initial category can be divided into "usable category" and "easy-to-use category". If the initial category is "usable category", S406 is executed; if the initial category is "easy-to-use category", S407 is executed.

[0147] S406: Determine, based on the large language model and the corresponding third prompt word, whether the feedback information belongs to an information category under the “usable category”.

[0148] S407: Determine, based on the large language model and the corresponding third prompt word, whether the feedback information belongs to the information category of “easy-to-use category”.

[0149] S408. Generate a category analysis document based on the classification results of S406 and S407.

[0150] Figure 5 The structure block diagram of a device for processing product feedback information provided by the embodiment of the present disclosure. For the convenience of explanation, only the part related to the embodiment of the present disclosure is shown. Figure 5 The processing device 500 based on product feedback information includes: an acquisition unit 501, an analysis unit 502 and a determination unit 503.

[0151] The acquisition unit 501 is used to acquire user feedback information; wherein the feedback information represents the user's use of the product;

[0152] The analysis unit 502 is used to perform validity analysis on the feedback information according to a preset vector database to obtain an analysis result; wherein the preset vector database includes a plurality of vector data, the vector data is preset invalid information, the validity analysis represents the analysis of the validity of the semantics of the feedback information, and the analysis result represents whether the feedback information is valid or invalid;

[0153] The determination unit 503 is configured to determine the information category to which the feedback information belongs if the analysis result indicates that the feedback information is valid; wherein the information category indicates the role of the feedback information in improving the product.

[0154] Figure 6A structural block diagram of a device for processing product feedback information provided by an embodiment of the present disclosure, such as Figure 6 As shown, the processing device 600 based on product feedback information includes an acquisition unit 601 , an analysis unit 602 , and a determination unit 603 , wherein the analysis unit 602 includes a similarity determination module 6021 and an analysis module 6022 .

[0155] A similarity determination module 6021 is used to determine the similarity between the feedback information and the vector data in the preset vector database through a preset large language model; wherein the similarity represents the degree of association between the feedback information and the vector data;

[0156] The analysis module 6022 is used to perform validity analysis on the feedback information based on the similarity between the feedback information and the vector data in the preset vector database and based on the preset first prompt word to obtain the analysis result of the feedback information; wherein the preset first prompt word is the prompt word information in the preset large language model, which is used to determine the analysis result of the feedback information.

[0157] In one example, the analysis module 6022 includes:

[0158] A target determination submodule, configured to determine a target vector corresponding to the feedback information according to a similarity between the feedback information and the vector data in the preset vector database; wherein the target vector represents vector data that is closely associated with the feedback information;

[0159] The effectiveness analysis submodule is used to perform effectiveness analysis on the feedback information according to the target vector corresponding to the feedback information and the preset first prompt word to obtain the analysis result of the feedback information.

[0160] In one example, the target determination submodule is specifically used to:

[0161] If the similarity between the feedback information and the vector data in the preset vector database is greater than a preset first threshold, the vector data in the preset vector database is determined to be the target vector corresponding to the feedback information.

[0162] In one example, the validity analysis submodule is specifically used to:

[0163] Determining the number of target vectors corresponding to the feedback information according to a preset first prompt word;

[0164] If it is determined that the number of target vectors corresponding to the feedback information is less than a preset number threshold, then the analysis result of the feedback information is determined to be valid.

[0165] In one example, the target determination submodule is specifically used to:

[0166] Sorting the vector data in the preset vector database according to the similarity between the feedback information and each vector data in the preset vector database;

[0167] The vector data within the preset sorting position range is determined as the target vector corresponding to the feedback information.

[0168] In one example, the validity analysis submodule is specifically used to:

[0169] If it is determined according to the preset first prompt word that the similarities corresponding to the target vectors are all greater than a preset second threshold, then the analysis result of the feedback information is determined to be invalid.

[0170] In one example, the determining unit 503 includes:

[0171] an initial determination module, used to input the feedback information into a preset large language model, and determine the initial category to which the feedback information belongs based on a preset second prompt word; wherein the preset second prompt word is prompt word information in a preset large language model, and is used to determine the initial category of the feedback information, and the initial category represents a parent category of the information category to which the feedback information belongs;

[0172] The category determination module is used to determine the information category to which the feedback information belongs according to the initial category to which the feedback information belongs and the preset large language model.

[0173] In one example, the category determination module includes:

[0174] a prompt word determination submodule, used to determine a third prompt word corresponding to the initial category to which the feedback information belongs according to the association relationship between the preset initial category and the third prompt word; wherein the preset third prompt word is prompt word information in a preset large language model, and is used to determine the information category to which the feedback information belongs;

[0175] The category determination submodule is used to input the feedback information into a preset large language model, and determine the category information to which the feedback information belongs based on a third prompt word corresponding to the initial category to which the feedback information belongs.

[0176] In one example, it also includes:

[0177] A category acquisition unit, used to acquire the information category to which the feedback information determined within a preset time period belongs;

[0178] The document generation unit is used to generate a category analysis document according to the information category to which the feedback information determined within the preset time period belongs; wherein the category analysis document represents the result of information category analysis of the feedback information of the product within the preset time period.

[0179] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device.

[0180] Figure 7 A structural block diagram of an electronic device provided in an embodiment of the present disclosure, such as Figure 7 As shown, the electronic device 700 includes: at least one processor 702; and a memory 701 communicatively connected to the at least one processor 702; wherein the memory stores instructions executable by the at least one processor 702, and the instructions are executed by the at least one processor 702 so that the at least one processor 702 can execute the product-based feedback information processing method disclosed in the present invention.

[0181] The electronic device 700 further includes a receiver 703 and a transmitter 704. The receiver 703 is used to receive instructions and data sent by other devices, and the transmitter 704 is used to send instructions and data to external devices.

[0182] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0183] According to an embodiment of the present disclosure, the present disclosure also provides a computer program product, which includes: a computer program, the computer program is stored in a readable storage medium, at least one processor of an electronic device can read the computer program from the readable storage medium, and at least one processor executes the computer program so that the electronic device executes the solution provided by any of the above embodiments.

[0184] Figure 8 A schematic block diagram of an example electronic device 800 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0185] like Figure 8As shown, the device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0186] A number of components in the device 800 are connected to the I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0187] The computing unit 801 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above, such as a processing method based on product feedback information. For example, in some embodiments, the processing method based on product feedback information may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 800 via ROM 802 and / or communication unit 809. When the computer program is loaded into RAM 803 and executed by the computing unit 801, one or more steps of the processing method based on product feedback information described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to perform a processing method based on product feedback information in any other appropriate manner (e.g., by means of firmware).

[0188] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0189] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0190] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0191] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0192] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0193] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server may also be a server of a distributed system, or a server combined with a blockchain.

[0194] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0195] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for processing product feedback information, comprising: Obtaining user feedback information; wherein the feedback information represents the user's use of the product; According to a preset vector database, the feedback information is analyzed for validity to obtain an analysis result; wherein the preset vector database includes a plurality of vector data, the vector data is preset invalid information, the validity analysis represents the analysis of the semantic validity of the feedback information, and the analysis result represents whether the feedback information is valid or invalid; If the analysis result indicates that the feedback information is valid, then the information category to which the feedback information belongs is determined; wherein the information category indicates the role of the feedback information in improving the product.

2. The method according to claim 1, wherein: The effectiveness analysis of the feedback information is performed according to the preset vector database to obtain the analysis result, including: Determining the similarity between the feedback information and the vector data in the preset vector database through a preset large language model; wherein the similarity represents the degree of association between the feedback information and the vector data; According to the similarity between the feedback information and the vector data in the preset vector database, based on the preset first prompt word, the feedback information is analyzed for validity to obtain the analysis result of the feedback information; wherein the preset first prompt word is the prompt word information in the preset large language model, which is used to determine the analysis result of the feedback information.

3. The method according to claim 2, wherein: The performing validity analysis on the feedback information based on the similarity between the feedback information and the vector data in the preset vector database and based on the preset first prompt word to obtain the analysis result of the feedback information includes: Determining a target vector corresponding to the feedback information according to the similarity between the feedback information and the vector data in the preset vector database; wherein the target vector represents vector data that is closely associated with the feedback information; The feedback information is analyzed for effectiveness according to the target vector corresponding to the feedback information and the preset first prompt word to obtain an analysis result of the feedback information.

4. The method according to claim 3, wherein: The determining, according to the similarity between the feedback information and the vector data in the preset vector database, a target vector corresponding to the feedback information includes: If the similarity between the feedback information and the vector data in the preset vector database is greater than a preset first threshold, the vector data in the preset vector database is determined to be the target vector corresponding to the feedback information.

5. The method according to claim 4, wherein: The performing validity analysis on the feedback information according to the target vector corresponding to the feedback information and the preset first prompt word to obtain the analysis result of the feedback information includes: Determining the number of target vectors corresponding to the feedback information according to a preset first prompt word; If it is determined that the number of target vectors corresponding to the feedback information is less than a preset number threshold, then the analysis result of the feedback information is determined to be valid.

6. The method according to any one of claims 3 to 5, wherein: The determining, according to the similarity between the feedback information and the vector data in the preset vector database, a target vector corresponding to the feedback information includes: Sorting the vector data in the preset vector database according to the similarity between the feedback information and each vector data in the preset vector database; The vector data within the preset sorting position range is determined as the target vector corresponding to the feedback information.

7. The method according to claim 6, wherein: The performing validity analysis on the feedback information according to the target vector corresponding to the feedback information and the preset first prompt word to obtain the analysis result of the feedback information includes: If it is determined according to the preset first prompt word that the similarities corresponding to the target vectors are all greater than a preset second threshold, then the analysis result of the feedback information is determined to be invalid.

8. The method according to any one of claims 1 to 7, wherein: The determining the information category to which the feedback information belongs includes: Inputting the feedback information into a preset large language model, and determining the initial category to which the feedback information belongs based on a preset second prompt word; wherein the preset second prompt word is prompt word information in a preset large language model, and is used to determine the initial category of the feedback information, and the initial category represents a parent category of the information category to which the feedback information belongs; The information category to which the feedback information belongs is determined according to the initial category to which the feedback information belongs and the preset large language model.

9. The method according to claim 8, wherein: The step of determining the information category to which the feedback information belongs according to the initial category to which the feedback information belongs and the preset large language model includes: Determining a third prompt word corresponding to the initial category to which the feedback information belongs according to the association relationship between the preset initial category and the third prompt word; wherein the preset third prompt word is prompt word information in a preset large language model, which is used to determine the information category to which the feedback information belongs; The feedback information is input into a preset large language model, and the category information to which the feedback information belongs is determined based on a third prompt word corresponding to the initial category to which the feedback information belongs.

10. The method according to any one of claims 1 to 9, further comprising: Obtaining the information category to which the feedback information determined within a preset time period belongs; A category analysis document is generated according to the information category to which the feedback information determined within the preset time period belongs; wherein the category analysis document represents the result of information category analysis of the feedback information of the product within the preset time period.

11. A device for processing product feedback information, comprising: An acquisition unit, used to acquire user feedback information; wherein the feedback information represents the user's use of the product; An analysis unit, configured to perform validity analysis on the feedback information according to a preset vector database to obtain an analysis result; wherein the preset vector database includes a plurality of vector data, the vector data is preset invalid information, the validity analysis represents the analysis of the validity of the semantics of the feedback information, and the analysis result represents whether the feedback information is valid or invalid; A determination unit is used to determine the information category to which the feedback information belongs if the analysis result indicates that the feedback information is valid; wherein the information category indicates the role of the feedback information in improving the product.

12. The device according to claim 11, wherein: The analysis unit comprises: A similarity determination module, used to determine the similarity between the feedback information and the vector data in the preset vector database through a preset large language model; wherein the similarity represents the degree of association between the feedback information and the vector data; The analysis module is used to perform validity analysis on the feedback information based on the similarity between the feedback information and the vector data in the preset vector database and based on the preset first prompt word to obtain the analysis result of the feedback information; wherein the preset first prompt word is the prompt word information in the preset large language model, which is used to determine the analysis result of the feedback information.

13. The device according to claim 12, wherein: The analysis module comprises: A target determination submodule, configured to determine a target vector corresponding to the feedback information according to a similarity between the feedback information and the vector data in the preset vector database; wherein the target vector represents vector data that is closely associated with the feedback information; The effectiveness analysis submodule is used to perform effectiveness analysis on the feedback information according to the target vector corresponding to the feedback information and the preset first prompt word to obtain the analysis result of the feedback information.

14. The device according to claim 13, wherein: The target determination submodule is specifically used for: If the similarity between the feedback information and the vector data in the preset vector database is greater than a preset first threshold, the vector data in the preset vector database is determined to be the target vector corresponding to the feedback information.

15. The device according to claim 14, wherein: The effectiveness analysis submodule is specifically used for: Determining the number of target vectors corresponding to the feedback information according to a preset first prompt word; If it is determined that the number of target vectors corresponding to the feedback information is less than a preset number threshold, then the analysis result of the feedback information is determined to be valid.

16. The device according to any one of claims 13 to 15, wherein: The target determination submodule is specifically used for: Sorting the vector data in the preset vector database according to the similarity between the feedback information and each vector data in the preset vector database; The vector data within the preset sorting position range is determined as the target vector corresponding to the feedback information.

17. The device according to claim 16, wherein: The effectiveness analysis submodule is specifically used for: If it is determined according to the preset first prompt word that the similarities corresponding to the target vectors are all greater than a preset second threshold, then the analysis result of the feedback information is determined to be invalid.

18. The device according to any one of claims 11 to 17, wherein: The determining unit comprises: an initial determination module, used to input the feedback information into a preset large language model, and determine the initial category to which the feedback information belongs based on a preset second prompt word; wherein the preset second prompt word is prompt word information in a preset large language model, and is used to determine the initial category of the feedback information, and the initial category represents a parent category of the information category to which the feedback information belongs; The category determination module is used to determine the information category to which the feedback information belongs according to the initial category to which the feedback information belongs and the preset large language model.

19. The device according to claim 18, wherein: The category determination module comprises: a prompt word determination submodule, used to determine a third prompt word corresponding to the initial category to which the feedback information belongs according to the association relationship between the preset initial category and the third prompt word; wherein the preset third prompt word is prompt word information in a preset large language model, and is used to determine the information category to which the feedback information belongs; The category determination submodule is used to input the feedback information into a preset large language model, and determine the category information to which the feedback information belongs based on a third prompt word corresponding to the initial category to which the feedback information belongs.

20. The apparatus according to any one of claims 11 to 19, further comprising: A category acquisition unit, used to acquire the information category to which the feedback information determined within a preset time period belongs; The document generation unit is used to generate a category analysis document according to the information category to which the feedback information determined within the preset time period belongs; wherein the category analysis document represents the result of information category analysis of the feedback information of the product within the preset time period.

21. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 10.

22. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-10.

23. A computer program product, wherein: The invention comprises a computer program, which implements the steps of the method according to any one of claims 1 to 10 when being executed by a processor.

Images