Data processing method applied to feedback platform

Abstract

The invention discloses a data processing method applied to a feedback platform, and the method comprises the steps: carrying out the key content extraction of question information fed back by a user, carrying out the weighting algorithm processing of the extracted information, and classifying common questions; defining labels for members of the observation group; intelligent matching is carried out on the common questions and labels of members of the observation group, and the matched members of the observation group reply and solve the questions; and converting the network comments after the problems with the generality are solved into quantifiable satisfaction. The method has the advantages that the technical problems that a traditional method is low in processing efficiency and poor in pertinence are solved, and the efficiency, accuracy and scientificity of feedback problem processing are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of problem processing platforms, and in particular to a data processing method applied to a feedback platform. Background Art

[0002] This project is an online government inquiry platform for the media industry. Users provide feedback and expose issues through the platform, which then transfers the issues to government departments. The departments then respond online and respond. Finally, users rate their satisfaction based on the responses from the departments. The general evaluation results are: satisfied, average, or dissatisfied.

[0003] The existing solution provides an online platform for online government affairs, which lacks intelligent analysis of user feedback. The problems reported by netizens are large and small, numerous and complex, and the platform fails to extract the commonality of the problems well, that is, it is unable to extract and summarize the key content of the problem information reported by users. In this way, the feedback problems cannot be deeply analyzed, quickly located, and professionally tracked and processed. The traditional methods of the existing online platform for online government affairs are inefficient and not very targeted, and the efficiency and accuracy of problem processing are low. Summary of the invention

[0004] The purpose of the present invention is to provide a data processing method applied to a feedback platform to solve the above technical problems. The specific technical solution is as follows:

[0005] A data processing method applied to a feedback platform includes the following methods:

[0006] Extract key content from user feedback, process the extracted information using a weighted algorithm, and classify common problems;

[0007] Define labels for members of the observation group;

[0008] Intelligently match common questions with the tags of members of the observation group, and have the matched members of the observation group answer and solve the questions;

[0009] Convert online reviews after common problems are solved into quantifiable satisfaction;

[0010] The method for extracting key content from the problem information fed back by the user includes:

[0011] Establishing a database for the feedback platform, the database collects user feedback information, including: picture information and text information;

[0012] Get a list of key information;

[0013] Purify the key information in the key information list to obtain the keyword information of each user's feedback question information, and define the part of speech for it to form a keyword information database;

[0014] The method of performing weighted algorithm processing on the extracted information to classify common problems includes:

[0015] Taking into account the number of user feedback questions, page views, comments, and likes, a popularity score is calculated for each keyword. The specific formula is as follows:

[0016] Popularity score = (number of questions corresponding to the keyword / number of questions corresponding to the highest keyword in the current period) * 0.6 + (number of views of the corresponding question / highest number of views of the corresponding question) * 0.2 + (number of comments on the corresponding question / highest number of comments on the question) * 0.15 + (number of likes on the corresponding question / highest number of likes on the question) * 0.05,

[0017] Select mainstream common issues based on their popularity scores.

[0018] Furthermore, the data processing method applied to the feedback platform also includes:

[0019] The extracted information is processed to obtain emergent problems;

[0020] Intelligently match sudden issues with the tags of members of the observation group, and have the matched members of the observation group respond to and resolve the issues;

[0021] Convert online reviews after urgent issues are resolved into quantifiable satisfaction.

[0022] Furthermore, the specific method of processing the extracted information to obtain the sudden problem is:

[0023] The growth rate of the questions corresponding to each keyword is counted. If the counted growth rate exceeds the preset rate, the questions corresponding to the keyword are classified as sudden questions.

[0024] Furthermore, the specific method of counting the growth rate of questions corresponding to each keyword is:

[0025] Count the number of questions corresponding to each keyword and the release time of each question, calculate the time span of the release time of these questions, and then calculate the ratio of the number of questions and the time span, and use it as the growth rate.

[0026] Furthermore, after calculating the time span of the release time of these issues, determine whether the time span is greater than the preset minimum span value. If the time span is greater than the preset minimum span value, further calculate the ratio of the number of issues and the time span and use it as the growth rate.

[0027] Furthermore, the keyword information of each user's feedback problem information generates key tags of people, things, places, objects, and organizations for managers to quickly understand and decide whether to adopt them.

[0028] Furthermore, after forming the keyword information database, the method of extracting key content from the question information fed back by the user also includes:

[0029] Establish subject databases at the department and issue levels;

[0030] In the early stage, the corresponding information is input into the corresponding subject database for classification through user filling and manual correction;

[0031] In the later stage, by using the accumulated data input in the early stage, the underlying vocabulary library is formed, that is, different subject libraries correspond to different keyword information libraries, so as to realize automatic classification of user questions and exposed content, and continuously optimize and adjust during actual use.

[0032] Furthermore, the method of defining labels for members of the observation group includes:

[0033] Data Collection:

[0034] Collect the basic information, professional background, hobbies and relevant experience of the observation team members. Basic information includes: gender, age, and permanent residence area. Professional background includes: education, major and main research direction. Hobbies include: hobby type, certificates obtained, and competition projects. Relevant experience includes: projects participated in and papers published.

[0035] Tag definition:

[0036] Build personal portrait tags based on members’ basic information;

[0037] Build industry expertise labels based on members’ professional backgrounds, interests, hobbies, and relevant experience;

[0038] Label verification and optimization:

[0039] After defining personal portrait tags and industry expertise tags, verify and optimize the personal portrait tags and industry expertise tags of members;

[0040] By extracting and classifying the data information of the professional background, hobbies and relevant experience of the observation group members, the industry expertise information words that can reflect the professional ability and interests of the members are summarized, and the industry expertise labels are formed based on the industry expertise information words;

[0041] Verify and optimize the member's industry expertise label by comparing it with the information of other members of the observation group;

[0042] The method for intelligently matching common questions with labels of members of the observation group includes:

[0043] Using text matching technology, we match user feedback questions with the tags of observation group members for analysis, including:

[0044] Industry mapping and matching: According to the problem classification, determine the industry or field to which it belongs, and search for tags related to the industry or field in the tags of the observation group members, so as to screen out members with relevant industry background or expertise;

[0045] Interest mapping matching: Use the keywords in the question to match the interests, hobbies, professional certificates and other labels of the observation group members;

[0046] Finally, based on the results of the matching evaluation, select one or more appropriate members of the observation team to follow up and handle the problem.

[0047] Furthermore, the method of intelligently matching common questions with labels of members of the observation group also includes:

[0048] Using text matching technology, we match and analyze user feedback questions with the tags of observation team members. After obtaining preliminary matching results, we evaluate the relevance, verify the accuracy and optimize the algorithm.

[0049] Relevance assessment: The relevance of the matched observation team members to the problem is assessed by comprehensively considering their experience in the field, published papers or research results;

[0050] Accuracy verification: Verify the matching results through manual verification or machine learning algorithms to ensure the accuracy of the matching;

[0051] Optimization algorithm: Based on feedback from actual applications, member tag information and keyword information library are continuously adjusted to improve matching accuracy and efficiency.

[0052] Furthermore, the method of converting the network comments after the common problems are solved into quantifiable satisfaction includes:

[0053] Collect online comments from feedback platforms and pre-process the data;

[0054] According to the preprocessed data, the model for calculating satisfaction is as follows:

[0055] Sentiment score assignment: sentiment recognition of the review content, using sentiment analysis algorithms to identify the sentiment tendency in the review, and classifying the review as positive, negative or neutral. Positive reviews are scored higher, negative reviews are deducted, and neutral reviews are scored zero.

[0056] Level of detail of the review: Count the number of words in the review content and set different length intervals. Each interval corresponds to a different score. Shorter reviews have lower scores, while longer and more detailed reviews have higher scores.

[0057] Mentioned keywords and topics: Using keyword extraction technology, we identify the keywords and topics mentioned in the comments and score them based on their positive and negative attributes and importance.

[0058] Total satisfaction score = (positive and negative dimension score * 0.5) + (comment detail score * 0.3) + (keyword and topic score * 0.2).

[0059] Beneficial effects: The data processing method applied to the feedback platform provided by this application has achieved remarkable results in solving the processing efficiency and quality of mass feedback problems. This solution uses big data technology to enable us to conduct in-depth analysis, rapid positioning and professional tracking of feedback issues. It not only overcomes the technical difficulties of low processing efficiency and lack of pertinence of traditional methods, but also greatly improves the efficiency and accuracy of problem handling. At the same time, by constructing labels for the observation group and applying intelligent matching algorithms, the automatic connection between problems and industry experts is achieved. This innovative method not only quickly combines professional knowledge with practical problems, but also significantly improves the scientificity and accuracy of problem solving. Therefore, the data processing method applied to the feedback platform of this application has shown excellent results in handling mass feedback issues and has made important contributions to improving the quality and efficiency of public services. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a schematic diagram of a data processing method applied to a feedback platform of the present invention; DETAILED DESCRIPTION

[0061] In order to make the technical solution of the present invention clearer, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0062] like Figure 1 As shown, a data processing method applied to a feedback platform of the present solution includes the following method steps:

[0063] Step 1: Extract key content from user feedback information, process the extracted information using a weighted algorithm, and classify common problems;

[0064] Step 2: Define labels for members of the observation group;

[0065] Step 3: Intelligently match common questions with the labels of the observation group members, and have the matched observation group members answer and solve the questions;

[0066] Step 4: Convert online reviews after common problems are resolved into quantifiable satisfaction.

[0067] In step 1 of the above method, the specific method for extracting key content from the question information fed back by the user is as follows:

[0068] First, a database is established for the feedback platform. The database collects user feedback information, including image information and text information. Then, based on the above image information and text information, a key information list is obtained. Here, the key information list can be extracted through the three-party capability interface and the media AI model. Next, the key information in the above key information list is purified to obtain the keyword information of each user's feedback problem information, and the part of speech is defined to form a keyword information library. In other words, the keyword information library includes qualitative keyword information extracted from problems reported by multiple users.

[0069] The keyword information of each user's feedback information generates key tags of people, things, places, objects, and organizations, so that managers can quickly understand and decide whether to adopt the user's feedback. By filtering key tags, corresponding problems that need to be adopted and solved can be quickly and accurately filtered out, because different management departments solve corresponding social problems. In this way, problems corresponding to the areas handled by management departments can be quickly and effectively filtered out.

[0070] As a further method, after forming a keyword information library, the method for extracting key content from the problem information fed back by the user also includes: establishing two-level theme libraries of departments and problems. After establishing the department theme library and the problem theme library, in the early stage of using the data processing method applied to the feedback platform, the corresponding information is input into the corresponding theme library for classification by means of user filling and manual correction (early data accumulation). In the later stage of use, the platform applied to the feedback forms an underlying vocabulary by using the data accumulation input in the early stage. Different theme libraries correspond to different keyword information libraries, that is, the department theme library corresponds to the keyword information theme library of the relevant information of the department management field, and the problem theme library corresponds to the keyword information library extracted and qualitatively analyzed by the problem fed back by the user. In this way, the platform applied to the feedback can realize automatic classification of the content of user questions and exposure, that is, when the feedback question is raised, the corresponding department theme library and the corresponding problem theme library are directly classified, and the management department and the problem direction are locked quickly, which is efficient and accurate in classification, and is convenient for accurate solution in the later stage, and provides accurate solutions. In the actual use process, such a classification method is continuously optimized and adjusted according to the information changes of the department theme library, the problem theme library and the keyword information library.

[0071] In step 1 of the above method, the extracted information is processed by a weighted algorithm, and the specific method of classifying common problems includes:

[0072] Taking into account the number of user feedback questions, views, comments, likes and other dimensions, a popularity score is calculated for each keyword. The specific formula is as follows:

[0073] Popularity score = (number of questions corresponding to the keyword / number of questions corresponding to the highest keyword in the current period) * 0.6 + (number of views of the corresponding question / highest number of views of the corresponding question) * 0.2 + (number of comments on the corresponding question / highest number of comments on the question) * 0.15 + (number of likes on the corresponding question / highest number of likes on the question) * 0.05,

[0074] According to the above-mentioned heat scores, the mainstream common problems are selected. In other words, the degree of resolution of the reported problems is ranked according to the heat scores. The higher the heat score of the problem, the higher its social impact, which needs to be urgently resolved. This can optimize the resolution of problems with greater social impact reported by users, and improve social stability attributes.

[0075] Methods for defining labels for the members of the observation group in step 2 above (members of the observation group refer to personnel in the platform who can propose professional solutions to corresponding problems, and the observation group includes professionals in multiple management fields) include:

[0076] Data Collection:

[0077] Collect the basic information, professional background, hobbies and relevant experience of the observation team members. Basic information includes: gender, age, and permanent residence area; professional background includes: education, major and main research direction; hobbies include: hobby type, certificates obtained, and competition projects; relevant experience includes: projects participated in and papers published;

[0078] Tag definition:

[0079] Based on the above basic information of the member, a personal portrait label is constructed. For example, a member constructs a personal portrait label of "male", "30-35 years old", and "East China" based on the basic information;

[0080] Construct industry specialty labels based on the members' professional background, hobbies and related experience. For example, by extracting and classifying the data information of a member's professional background, hobbies and related experience, summarize the industry specialty information words that can reflect the member's professional ability and interest, and construct industry specialty labels based on the industry specialty information words;

[0081] Label verification and optimization:

[0082] After defining personal portrait tags and industry specialty tags, verify and optimize the personal portrait tags and industry specialty tags of members. Specifically, the personal portrait tags of members can be verified and optimized by communicating and confirming with the members themselves, and the industry specialty tags of members can be verified and optimized by comparing and verifying with the information of other team members. If the tags are found to be inaccurate or need to be adjusted, the platform will optimize and update them in a timely manner.

[0083] In step 3 of the above method, the specific method of intelligently matching the common questions with the labels of the members of the observation group is as follows:

[0084] Using text matching technology, we match and analyze the user feedback questions with the tags of the observation team members. The specific matching methods include:

[0085] Industry mapping and matching: According to the problem classification, determine the industry or field to which it belongs, and search for tags related to the industry or field in the tags of the observation group members, so as to screen out members with relevant industry background or expertise;

[0086] Interest mapping matching: Use the keywords in the question to match the interests, hobbies, professional certificates and other tags of the observation group members;

[0087] Finally, based on the results of the matching evaluation, select one or more appropriate members of the observation team to follow up and handle the problem.

[0088] Step 3 of the above solution can accurately and quickly match the problem to an observation team member with professional problem-solving capabilities through the above-mentioned industry mapping matching and interest mapping matching, so that the problems raised by users can be solved scientifically, improving the user's problem-solving strength and accuracy, and meeting social needs.

[0089] Furthermore, the specific method of intelligently matching common problems with labels of members of the observation group also includes:

[0090] Using text matching technology, we match and analyze the user feedback questions with the tags of the observation team members. After obtaining the preliminary matching results, we conduct relevance evaluation, accuracy verification and algorithm optimization on the matching results, as follows;

[0091] Relevance assessment: The relevance of the matched observation team members to the problem is assessed based on their experience in the field, published papers or research results, etc.

[0092] Accuracy verification: Verify the matching results through manual verification or machine learning algorithms to ensure the accuracy of the matching;

[0093] Optimization algorithm: Based on feedback from actual applications, member tag information and keyword information library are continuously adjusted to improve matching accuracy and efficiency.

[0094] In step 4 of the above method, the specific methods for converting online reviews with common problems solved into quantifiable satisfaction include:

[0095] Collect online comments from the feedback platform and preprocess the data. The preprocessing methods here include: removing irrelevant characters, punctuation marks, special symbols, etc., performing word segmentation, removing stop words, etc.

[0096] According to the above preprocessed data, the model for calculating satisfaction is as follows:

[0097] Sentiment score assignment: sentiment recognition of the review content, using sentiment analysis algorithms to identify the sentiment tendency in the review, and classifying the review as positive, negative or neutral. Positive reviews are scored higher, negative reviews are deducted, and neutral reviews are scored zero.

[0098] Level of detail of the review: Count the number of words in the review content and set different length intervals. Each interval corresponds to a different score. Shorter reviews have lower scores, while longer and more detailed reviews have higher scores.

[0099] Mentioned keywords and topics: Using keyword extraction technology, we identify the keywords and topics mentioned in the comments and score them based on their positive and negative attributes and importance.

[0100] Total satisfaction score = (positive and negative dimension scores*0.5) + (review detail score*0.3) + (keyword and topic score*0.2).

[0101] According to the above-mentioned satisfaction calculation, the satisfaction of social users with problem solving can be obtained in a timely and accurate manner, so as to understand the social feedback of the common problem based on the satisfaction, so that the relevant management departments can more accurately understand the social situation and impact related to the problem, and thus timely adjust the management methods of the relevant social management departments, play a better social regulation role, and improve the quality of public services.

[0102] In summary, the data processing method applied to the feedback platform provided by this application has achieved remarkable results in solving the processing efficiency and quality of mass feedback problems. This solution uses big data technology to enable us to conduct in-depth analysis, rapid positioning and professional tracking of feedback issues. It not only overcomes the technical difficulties of low processing efficiency and lack of pertinence of traditional methods, but also greatly improves the efficiency and accuracy of problem handling. At the same time, by constructing the labels of the observation group and applying the intelligent matching algorithm, the automatic connection between problems and industry experts is realized. This innovative method not only quickly combines professional knowledge with practical problems, but also significantly improves the scientificity and accuracy of problem solving. Therefore, the data processing method applied to the feedback platform of this application has shown excellent results in handling mass feedback issues, and has made important contributions to improving the quality and efficiency of public services.

[0103] It is understandable that the identification of common problems is based on the heat value of keywords. The calculation of heat value refers to multiple dimensions such as the number of feedback questions, views, comments and likes. However, in some cases, for some sudden hot or urgent events, the newly released questions have not been widely viewed, commented or liked, and the calculated heat value is small, which cannot be identified as a common problem and thus does not receive attention. However, the number of these reflected problems is large and generally more important.

[0104] In this application, in order to solve such a problem, as an optional implementation, the data processing method applied to the feedback platform further includes:

[0105] The extracted information is processed to obtain sudden problems, which are intelligently matched with the labels of the members of the observation group, and the matched members of the observation group respond to the problems and solve them, and the online comments after the sudden problems are solved are converted into quantifiable satisfaction. Specifically, the processing method of converting the online comments after the sudden problems are solved into quantifiable satisfaction is the same as the statistical method of the common problems mentioned above, which will not be repeated here.

[0106] In the implementation manner of the present application, the specific method for processing the extracted information to obtain the sudden problem is:

[0107] The growth rate of the questions corresponding to each keyword is counted. If the counted growth rate exceeds the preset rate, the questions corresponding to the keyword are classified as sudden questions.

[0108] Specifically, the number of questions corresponding to each keyword and the publishing time of each question are counted, and the time span of the publishing time of these questions is calculated. For example, the publishing time of the most recent question at the current time is 3:10 on 2022.11.10, and the publishing time of the question farthest from the current time is 2:10 on 2022.11.10, then the time span is one hour. Then calculate the ratio of the number of questions and the time span, and use it as the growth rate. If the calculated growth rate is too large, it means that a large number of people have concentrated on reflecting this problem in a short period of time, and it should be taken seriously. In this application, it is selected as an emergency problem for priority processing. The subsequent processing flow can refer to the aforementioned processing method for common problems. Here, the preset rate can be specifically set as needed, and this application is not limited.

[0109] As a preferred implementation, after calculating the time span of the release time of these problems, it is determined whether the time span is greater than the preset minimum span value. If the time span is greater than the preset minimum span value, the ratio of the number of problems and the time span is further calculated and used as the growth rate.

[0110] It is understandable that in some extreme cases, the release time of two questions may be very short. According to the above calculation method, even if there are only two questions, a very large growth rate will be calculated. However, the sudden problems that this application actually wants to count are those that are posted by a large number of users in a short period of time. Only two questions obviously do not meet expectations. Therefore, this application sets a judgment logic and introduces a minimum span value. The calculated time span must be greater than this minimum span value before the subsequent growth rate calculation is performed.

[0111] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A data processing method applied to a feedback platform, characterized in that: The following methods are included: Extract key content from user feedback, process the extracted information using a weighted algorithm, and classify common problems; Define labels for members of the observation group; Intelligently match common questions with the tags of members of the observation group, and have the matched members of the observation group answer and solve the questions; Convert online reviews after common problems are solved into quantifiable satisfaction; The method for extracting key content from the problem information fed back by the user includes: Establishing a database for the feedback platform, the database collects user feedback information, including: picture information and text information; Get a list of key information; Purify the key information in the key information list to obtain the keyword information of each user's feedback question information, and define the part of speech for it to form a keyword information database; The method of performing weighted algorithm processing on the extracted information to classify common problems includes: Taking into account the number of user feedback questions, page views, comments, and likes, a popularity score is calculated for each keyword. The specific formula is as follows: Popularity score = (number of questions corresponding to the keyword / number of questions corresponding to the highest keyword in the current period) * 0.6 + (number of views of the corresponding question / highest number of views of the corresponding question) * 0.2 + (number of comments on the corresponding question / highest number of comments on the question) * 0.15 + (number of likes on the corresponding question / highest number of likes on the question) * 0.05, Select mainstream common issues based on their popularity scores.

2. The data processing method applied to the feedback platform according to claim 1, characterized in that: The data processing method applied to the feedback platform also includes: The extracted information is processed to obtain emergent problems; Intelligently match sudden issues with the tags of members of the observation group, and have the matched members of the observation group respond to and resolve the issues; Convert online reviews after urgent issues are resolved into quantifiable satisfaction.

3. The data processing method applied to the feedback platform according to claim 2, characterized in that: The specific method of processing the extracted information to obtain the sudden problem is: The growth rate of the questions corresponding to each keyword is counted. If the counted growth rate exceeds the preset rate, the questions corresponding to the keyword are classified as sudden questions.

4. The data processing method applied to the feedback platform according to claim 3, characterized in that: The specific method for counting the growth rate of questions corresponding to each keyword is: Count the number of questions corresponding to each keyword and the release time of each question, calculate the time span of the release time of these questions, and then calculate the ratio of the number of questions and the time span, and use it as the growth rate.

5. The data processing method applied to the feedback platform according to claim 4, characterized in that: After calculating the time span of the release time of these issues, determine whether the time span is greater than the preset minimum span value. If the time span is greater than the preset minimum span value, further calculate the ratio of the number of issues and the time span and use it as the growth rate.

6. The data processing method applied to the feedback platform according to claim 1, characterized in that: The keyword information of each user's feedback problem information generates key tags of people, things, places, objects, and organizations, so that managers can quickly understand and decide whether to adopt them.

7. The data processing method applied to the feedback platform according to claim 1, characterized in that: After forming the keyword information database, the method of extracting key content from the question information fed back by the user also includes: Establish subject databases at the department and issue levels; In the early stage, the corresponding information is input into the corresponding subject database for classification through user filling and manual correction; In the later stage, by using the accumulated data input in the early stage, the underlying vocabulary library is formed, that is, different subject libraries correspond to different keyword information libraries, so as to realize automatic classification of user questions and exposed content, and continuously optimize and adjust during actual use.

8. The data processing method applied to the feedback platform according to claim 1, characterized in that: The method for defining labels for members of an observation group includes: Data Collection: Collect the basic information, professional background, hobbies and relevant experience of the observation team members. Basic information includes: gender, age, and permanent residence area. Professional background includes: education, major and main research direction. Hobbies include: hobby type, certificates obtained, and competition projects. Relevant experience includes: projects participated in and papers published. Tag definition: Build personal portrait tags based on members’ basic information; Build industry expertise labels based on members’ professional backgrounds, interests, hobbies, and relevant experience; Label verification and optimization: After defining personal portrait tags and industry expertise tags, verify and optimize the personal portrait tags and industry expertise tags of members; By extracting and classifying the data information of the professional background, hobbies and relevant experience of the observation group members, the industry expertise information words that can reflect the professional ability and interests of the members are summarized, and the industry expertise labels are formed based on the industry expertise information words; Verify and optimize the member's industry expertise label by comparing it with the information of other members of the observation group; The method for intelligently matching common questions with labels of members of the observation group includes: Using text matching technology, we match user feedback questions with the tags of observation group members for analysis, including: Industry mapping and matching: According to the problem classification, determine the industry or field to which it belongs, and search for tags related to the industry or field in the tags of the observation group members, so as to screen out members with relevant industry background or expertise; Interest mapping matching: Use the keywords in the question to match the interests, hobbies, professional certificates and other labels of the observation group members; Finally, based on the results of the matching evaluation, select one or more appropriate members of the observation team to follow up and handle the problem.

9. The data processing method applied to the feedback platform according to claim 8, characterized in that: The method for intelligently matching common questions with labels of members of the observation group also includes: Using text matching technology, we match and analyze user feedback questions with the tags of observation team members. After obtaining preliminary matching results, we evaluate the relevance, verify the accuracy and optimize the algorithm. Relevance assessment: The relevance of the matched observation team members to the problem is assessed by comprehensively considering their experience in the field, published papers or research results; Accuracy verification: Verify the matching results through manual verification or machine learning algorithms to ensure the accuracy of the matching; Optimization algorithm: Based on feedback from actual applications, member tag information and keyword information library are continuously adjusted to improve matching accuracy and efficiency.

10. The data processing method applied to the feedback platform according to claim 1, characterized in that: The method of converting online reviews with common problems solved into quantifiable satisfaction includes: Collect online comments from feedback platforms and pre-process the data; According to the preprocessed data, the model for calculating satisfaction is as follows: Sentiment score assignment: sentiment recognition of the review content, using sentiment analysis algorithms to identify the sentiment tendency in the review, and classifying the review as positive, negative or neutral. Positive reviews are scored higher, negative reviews are deducted, and neutral reviews are scored zero. Level of detail of the review: Count the number of words in the review content and set different length intervals. Each interval corresponds to a different score. Shorter reviews have lower scores, while longer and more detailed reviews have higher scores. Mentioned keywords and topics: Using keyword extraction technology, we identify the keywords and topics mentioned in the comments and score them based on their positive and negative attributes and importance. Total satisfaction score = (positive and negative dimension scores*0.5) + (review detail score*0.3) + (keyword and topic score*0.2).

Images