A system for intelligent question-answering and learning for researchers

Abstract

The present application discloses a system for intelligent question-answering and learning for scientific researchers, and relates to the field of intelligent question-answering technology; the intelligent learning module can more accurately determine the research field to which the scientific research question belongs and find the scientific researcher end related thereto by matching the fields and keywords of the scientific research questions and calculating the correlation; at the same time, by accumulating monitoring of the scientific research capabilities of the scientific researchers and evaluating the professional index of the scientific researchers end for specific scientific research questions, it can objectively reflect the professionalism of the scientific researchers on the scientific research questions, and help the system to more accurately push the scientific research questions to the relevant final selection ends, thereby improving the correctness and efficiency of problem solving; after the question-answering output module receives the output instruction sent by the intelligent learning module, it sorts the target answers according to their confidence levels and sends them to the question-asking end of the scientific research questions, thereby ensuring that the scientific researchers obtain the most likely accurate answers and improving the credibility and reliability of the answers.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent question answering, and in particular to a system for intelligent question answering and learning for scientific researchers. Background Art

[0002] In scientific research, we often encounter various problems, including theoretical problems, experimental design, data analysis, literature reading, etc. If we do not get timely answers, the scientific research project will stagnate. The intelligent question-answering and learning system for scientific researchers can solve this problem very well. Therefore, the intelligent question-answering and learning system for scientific researchers is very necessary.

[0003] At present, intelligent question-answering and learning systems for scientific researchers usually answer related questions by collecting various scientific research questions and their corresponding solutions. This method may easily lead to questions that are not included in the knowledge base not being effectively answered. In addition, due to the different sources and quality of the knowledge base content, the credibility and reliability of various questions and their corresponding solutions in the knowledge base cannot be judged and identified, which will result in the accuracy of the answers given by the system cannot be guaranteed. Summary of the invention

[0004] The purpose of the present invention is to solve the problems mentioned in the above background technology and to propose a system for intelligent question answering and learning for scientific researchers.

[0005] The purpose of the present invention can be achieved by the following technical solutions: A system for intelligent question-answering and learning for scientific researchers, comprising an intelligent robot end and a user end; the intelligent robot end comprises a cloud question knowledge base, an intelligent learning module and a question-answering output module;

[0006] When receiving scientific research questions from researchers, the intelligent robot will learn and answer the scientific research questions to improve the cloud knowledge base and output answers. Specifically:

[0007] The Yunwen Knowledge Base uses natural language understanding to traverse the scientific research questions and answers stored in it. If there is no question consistent with the scientific research question in the Yunwen Knowledge Base, the scientific research question will be integrated into a learning instruction and sent to the intelligent learning module; if there is a question consistent with the scientific research question in the Yunwen Knowledge Base, the question consistent with the scientific research question and its corresponding correct answer will be retrieved and sent to the question-asking end corresponding to the scientific research question;

[0008] When the intelligent learning module receives a learning instruction, it uses CNN and Apriori algorithms to determine the target field corresponding to the scientific research question, the relevance of each common key, and locates and analyzes it to obtain the final selection end of the scientific research question and the relevance index and professional index corresponding to the final selection end; pushes the scientific research question to the final selection end, and integrates and learns the received scientific research answers to obtain several target answers to the scientific research question and their corresponding confidence levels, and integrates them into output instructions and sends them to the question-answering output module;

[0009] When the question-and-answer output module receives an output instruction, it parses the output instruction to obtain several target answers to the scientific research question and their corresponding confidence levels, and sorts each answer in order of its confidence level and sends it to the corresponding question-asking end of the scientific research question as the final answer to the scientific research question; the question-asking end of the scientific research question can provide feedback on several answers in the final answer to the scientific research question, where the feedback includes adoption or rejection; when the feedback is received, a cumulative analysis is performed on each source end to obtain a professional cumulative coefficient, and the answer that is adopted is used as the correct answer to the corresponding scientific research question, and is sent to the Yunwen knowledge base for storage.

[0010] As a preferred embodiment of the present invention, the integration process of sending the output instruction is specifically as follows:

[0011] The received scientific research answers to scientific research questions are sorted into several replies using natural language understanding. Specifically, when the descriptions of one or more scientific research answers to scientific research questions are the same, the corresponding one or more scientific research answers to scientific research questions are one reply, and the final selection end corresponding to the one or more scientific research answers is recorded as the source end of the reply as p, P = 1, 2, 3...P, P is a positive integer, P represents the total number of source segments corresponding to the reply, and p represents the sequence number of any source end in the reply; thus, several source ends of each reply can be obtained, and the correlation index Dhp and professional index Lhp of the several source ends corresponding to each reply are substituted into the set formula The confidence LD corresponding to the answer is calculated, where h6 and h7 are the set weight coefficients respectively; when the confidence is greater than the set confidence threshold, the answer corresponding to the confidence is recorded as the target answer, thereby obtaining several target answers to scientific research questions and their corresponding confidences, and integrating them into output instructions and sending them to the question and answer output module.

[0012] As a preferred embodiment of the present invention, the specific process of cumulative analysis is as follows:

[0013] Obtain behavioral data from each source, where the behavioral data includes the number of answers, the answer time and feedback status corresponding to each answer, and divide the corresponding answers into valid answers and invalid answers according to the feedback status, count the number of valid answers and invalid answers respectively, and record them as D1 and D2 respectively; calculate the time difference between the answer time corresponding to the valid answer and the invalid answer and the system time to obtain the time interval corresponding to each valid answer and the invalid answer, and record them as D3j and D4k respectively, where j = 1, 2, 3 ... J, k = 1, 2, 3 ... K, J and K are both positive integers, J represents the total number of valid answers, K represents the total number of invalid answers; j represents the sequence number of any valid answer, and k represents the sequence number of any invalid answer; through the set formula The professional cumulative coefficient β of each source end is calculated and sent to the intelligent learning module; among them, f1, f2, f3, and f4 are the set weight coefficients respectively.

[0014] As a preferred embodiment of the present invention, the specific process of positioning analysis is:

[0015] 401: Record the researcher end whose research field is consistent with the target field corresponding to the scientific research question as the preliminary selection end, and perform correlation analysis on the research field of the preliminary selection end and the keywords corresponding to the research field to obtain a correlation index about the scientific research question;

[0016] 402: Record the online preliminary selection end as the re-selection end. If the correlation index of the re-selection end is greater than the set correlation threshold, record the re-selection end as the final selection end. Thus, the final selection end of the scientific research question and the correlation index corresponding to each final selection end can be obtained, and the scientific research question is pushed to the final selection end.

[0017] 403: Retrieve the registration information of the final selection terminal and measure its professional ability to obtain the professional index of the final selection terminal.

[0018] As a preferred implementation of the present invention, the specific process of performing association analysis on the research field of the primary selection end and the keywords corresponding to the research field is as follows:

[0019] Retrieve the research field and the keywords corresponding to the research field from the preliminary selection end, and compare them with the common keywords corresponding to the target field to obtain the keywords with common points between the two and record them as common keywords. The correlation degree corresponding to the common keywords that can be matched is recorded as Di, i=1,2,3...I, I is a positive integer, I represents the number of common keywords in the preliminary selection end, and i represents the sequence number of any common keyword among them;

[0020] Compare and analyze the relevance of common keywords with the set relevance interval to classify the relevance into high relevance, medium relevance and low relevance; count the cumulative times of high relevance, medium relevance and low relevance respectively, and record them as C1, C2 and C3 respectively;

[0021] Substitute the correlation degree Di, the cumulative number of highly correlated values ​​C1, the cumulative number of moderately correlated values ​​C2, and the cumulative number of lowly correlated values ​​C3 into the set formula The correlation index Dh of the preliminary selection end regarding scientific research issues is calculated, where h1, h2, and h3 are the set weight coefficients, and h1>h2>h3>0.

[0022] As a preferred implementation of the present invention, the specific process of measuring the professional ability based on the registration information of the terminal selection end is:

[0023] Set a base value for each educational background, and compare the educational background of the final selection end with all the set educational backgrounds to match the corresponding educational background value, which is recorded as L1;

[0024] Set a professional base value corresponding to different professional titles, compare the professional title corresponding to the final selection end with all the set professional titles to match the corresponding professional base value, which is recorded as L2;

[0025] Each user terminal is set to correspond to a cumulative professional coefficient, and the final selected terminal is compared with all user terminals to match the corresponding professional cumulative coefficient, which is recorded as β;

[0026] Substitute the academic base value L1, occupation base value L2 and professional cumulative coefficient β into the given formula Lh=β×(h4×L1+h5×L2) to calculate the professional index Lh, where h4 and h5 are the set weight coefficients respectively.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] 1. The intelligent learning module can more accurately determine the research field to which the scientific research question belongs and find the relevant scientific researchers by matching the fields and keywords of the scientific research questions and calculating the correlation. This improves the matching and correlation between the scientific research questions and the scientific researchers. At the same time, by cumulatively monitoring and evaluating the scientific research capabilities of the scientific researchers and analyzing the feedback of effective and invalid answers, the professional index of the scientific researchers for specific scientific research questions can be obtained. This can objectively reflect the professionalism of the scientific researchers for the scientific research questions, help the system to more accurately push the scientific research questions to the relevant final selection end, and improve the correctness and efficiency of problem solving.

[0029] 2. After receiving the output instruction sent by the intelligent learning module, the question-answering output module sorts the target answers according to their confidence and sends them to the question-asking end of the scientific research question, which can ensure that the scientific researchers get the most likely and accurate answers, and improve the credibility and reliability of the answers; Feedback mechanism and knowledge base update: After receiving the feedback, the question-answering output module can perform cumulative analysis on each source end and generate a professional cumulative coefficient to feed back to the intelligent learning module; through such a feedback mechanism, the system can continuously learn and optimize, update the cloud question knowledge base, and improve the accuracy and credibility of the system's answers to scientific research questions;

[0030] In summary, compared with the traditional method of collecting problem-solving solutions, the intelligent question-answering and learning system has higher personalization and accuracy, can effectively solve the problem of inconsistent content sources and quality of the knowledge base, and improve the accuracy and credibility of the answers given by the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0032] Figure 1 This is the overall system block diagram of the present invention. DETAILED DESCRIPTION

[0033] The technical scheme of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] It should be noted that the relevant information (including but not limited to user registration information, etc.) and data (including but not limited to data used for display, data for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties.

[0035] See also Figure 1 As shown, a system for intelligent question-answering and learning for scientific researchers, the system includes an intelligent machine end and a user end; wherein the intelligent machine end includes a cloud question knowledge base 111, a question-answering output module 112 and an intelligent learning module 110;

[0036] The user terminal refers to the mobile terminal of the successfully registered researchers (including but not limited to smart phones, smart tablets and smart watches). The researchers ask questions or get answers from the intelligent question-answering robot terminal through the user terminal;

[0037] When receiving a scientific research question from a user terminal of a scientific researcher, the intelligent robot answers the scientific research question and learns to improve the cloud question knowledge base 111, so as to achieve a fast and accurate answer to the scientific research question; specifically:

[0038] The cloud knowledge base 111 uses natural language understanding to traverse the scientific research questions and answers stored therein. If there is no question consistent with the scientific research question in the cloud knowledge base 111, the scientific research question is integrated into a learning instruction and sent to the intelligent learning module 110; if there is a question consistent with the scientific research question in the cloud knowledge base 111, the question consistent with the scientific research question and the corresponding correct answer are retrieved and sent to the questioning end corresponding to the scientific research question;

[0039] When the intelligent learning module 110110 receives the learning instruction, it monitors and evaluates all successfully registered user terminals to measure the research capabilities of each researcher and locate research problems; specifically:

[0040] Obtain the registration information of scientific researchers, including name, age, education, professional title, research field and corresponding keywords. The research field is specifically manifested in the fields of chemistry, physics, biology, agronomy, pharmacy, etc., and each research field is set with several keywords. For example, the common keywords in the field of chemistry usually include polymer chemistry, analytical chemistry, inorganic chemistry, organic chemistry, etc., and the common keywords in the field of pharmacy usually include pharmacology, pharmacy, drug analysis, drug toxicology, etc.; scientific researchers choose the corresponding keywords according to their own research and expertise; it can be seen that the setting of field plus keyword can accurately locate the research direction of scientific researchers;

[0041] CNN (convolutional neural network) is used to determine the research field to which the scientific research problem belongs, and it is recorded as the target field; the Apriori algorithm (association rule mining algorithm) is used to mine and analyze the correlation between the common keywords in the target field, thereby obtaining the correlation between the scientific research problem and its corresponding target field and each common keyword;

[0042] The researcher end whose research field is consistent with the target field corresponding to the scientific research question is recorded as the preliminary selection end (it should be noted that the preliminary selection end does not include the user end that raises the scientific research question), and the research field of the preliminary selection end and the corresponding keywords of the research field are retrieved, and compared with the common keywords corresponding to the target field to obtain the keywords with common points between the two and record them as common keywords. The correlation degree corresponding to the common keywords that can be matched is recorded as Di, where i=1,2,3...I, I is a positive integer, I represents the number of common keywords in the preliminary selection end, and i represents the sequence number of any common keyword; the correlation degree of the common keywords is compared and analyzed with the set correlation interval. When the correlation degree is greater than the maximum value in the set correlation interval, a high correlation is accumulated; when the correlation degree is within the set correlation interval, a moderate correlation is accumulated; when the correlation degree is less than the minimum value in the set correlation interval, a low correlation is accumulated; the cumulative number of high correlation, moderate correlation and low correlation is counted respectively, and they are recorded as C1, C2 and C3 respectively; using the set formula Calculate and obtain the correlation index Dh of the preliminary selection end with respect to the scientific research problem, where h1, h2, and h3 are respectively the set weight coefficients, and h1>h2>h3>0; thereby, the correlation index corresponding to each preliminary selection end can be obtained; the online preliminary selection end is recorded as the check-in end, and the correlation index of the check-in end is compared and analyzed with the set correlation threshold. If the correlation index of the check-in end is greater than the set correlation threshold, it means that the scientific researcher corresponding to the check-in end has a relatively large correlation with the scientific research problem, and the check-in end is recorded as the final selection end; thereby, the final selection end of the scientific research problem and the correlation index corresponding to each final selection end can be obtained, and the scientific research problem is pushed to the final selection end;

[0043] Set a base value for each educational background, and the base values ​​for different educational backgrounds are different. For example, the base value for a bachelor's degree is smaller than that for a master's degree, and the base value for a master's degree is smaller than that for a doctorate degree. Compare the educational background of the final selection end with all the set educational backgrounds to match the corresponding base value, and record it as L1.

[0044] Set a career base value corresponding to different professional titles, because different professional titles usually correspond to different scientific research levels, and the promotion of professional titles often requires certain professional experience and achievements; therefore, the career base values ​​corresponding to different professional titles can be set as indicators that reflect their professional levels and contributions; for example, the career base value of a teaching assistant is smaller than the career base value corresponding to a lecturer, which is smaller than the career base value corresponding to an associate professor, and the career base value corresponding to a lecturer is smaller than the career base value corresponding to a full professor; compare the professional title corresponding to the final selection end with all the set professional titles to match the corresponding career base value, and record it as L2;

[0045] Set each user end to correspond to a cumulative professional coefficient, compare the final selection end with all user ends to match the corresponding professional cumulative coefficient, record it as β; use the set formula Lh=β×(h4×L1+h5×L2) to calculate and obtain the professional index Lh, where h4 and h5 are the set weight coefficients respectively;

[0046] The received scientific research answers are integrated and learned to obtain accurate answers to scientific research questions. The specific integration process is as follows:

[0047] The received scientific research answers to scientific research questions are sorted into several replies using natural language understanding. Specifically, when the descriptions of one or more scientific research answers to scientific research questions are the same, the corresponding one or more scientific research answers to scientific research questions are regarded as one reply, and the final selection end corresponding to the one or more scientific research answers is regarded as the source end of the reply; thus, the source end of each reply is recorded as p, where P = 1, 2, 3...P, P is a positive integer, P represents the total number of source segments corresponding to the reply, and p represents the sequence number of any source end in the reply; retrieve the correlation index and professional index of the source end, and record them as Dhp and Lhp respectively, and use the set formula The confidence level corresponding to the answer is calculated, where h6 and h7 are respectively the set weight coefficients LD; when the confidence level is greater than the set confidence threshold, the answer corresponding to the confidence level is recorded as the target answer, thereby obtaining several target answers to the scientific research question and their corresponding confidence levels, and integrating them into an output instruction and sending them to the question and answer output module 112;

[0048] By matching and calculating the correlation between fields and keywords of scientific research questions, we can more accurately determine the research field to which the scientific research question belongs and find the relevant scientific researchers, thereby improving the matching and correlation between scientific research questions and scientific researchers; at the same time, by cumulatively monitoring and evaluating the scientific research capabilities of scientific researchers, and analyzing the feedback of effective and invalid answers, we can obtain the professional index of the scientific researcher for specific scientific research questions, which can objectively reflect the professionalism of the scientific researcher on the scientific research problem, and help the system push scientific research questions to the relevant final selection end more accurately.

[0049] When the question-answer output module 112 receives the output instruction, it parses the output instruction to obtain several target answers to the scientific research question and their corresponding confidence levels, and sorts the answers in order according to their confidence levels and sends them to the corresponding scientific research question-asking end as the final answers to the scientific research question; the scientific research question-asking end can provide feedback on several answers in the final answers to the scientific research question, where the feedback includes whether they are adopted or not;

[0050] When feedback is received, a cumulative analysis is performed on each source end, wherein the specific steps of the cumulative analysis are:

[0051] Obtain behavioral data from each source end, where the behavioral data includes the number of answers, the answer time corresponding to each answer, and the feedback situation, and the specific feedback situation includes adoption and rejection; record the answer with the feedback situation of adoption as a valid answer, and record the answer with the feedback situation of rejection as an invalid answer, count the number of valid answers and invalid answers respectively, and record them as D1 and D2 respectively; calculate the time difference between the answer time corresponding to the valid answer and the invalid answer and the system time to obtain the time interval corresponding to each valid answer and the invalid answer, and record them as D3j and D4k respectively, where j = 1, 2, 3 ... J, k = 1, 2, 3 ... K, J and K are both positive integers, J represents the total number of valid answers, and K represents the total number of invalid answers; j represents the serial number of any valid answer, and k represents the serial number of any invalid answer. It should be noted that when the time interval is larger, it means that the answer corresponding to the time interval is farther away from the current time, and the impact on the current source end (scientific researchers) is smaller; use the set formula Calculate and obtain the professional cumulative coefficient β of each source end, and send it to the intelligent learning module 110; wherein f1, f2, f3, and f4 are respectively set weight coefficients;

[0052] The answers that are fed back as adopted are regarded as the correct answers to the corresponding scientific research questions, and are sent to the Yunwen Knowledge Base 111 for storage;

[0053] After the question-and-answer output module receives the output instructions sent by the intelligent learning module, it sorts the target answers according to their confidence levels and sends them to the question-asking end of the scientific research question, which can ensure that researchers obtain the most likely accurate answers, thereby improving the credibility and reliability of the answers; feedback mechanism and knowledge base update: After receiving the feedback, the question-and-answer output module can perform a cumulative analysis of each source end and generate a professional cumulative coefficient to feed back to the intelligent learning module; through such a feedback mechanism, the system can continuously learn and optimize, update the cloud question knowledge base, and improve the accuracy and credibility of the system's answers to scientific research questions.

[0054] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A system for intelligent question-answering and learning for scientific researchers, comprising an intelligent robot end and a user end; characterized in that: The intelligent robot side includes an intelligent learning module and a question-answering output module; When receiving scientific research questions from researchers, the intelligent robot will learn and answer the scientific research questions to improve the cloud knowledge base and output answers. Specifically: The Cloud Question Knowledge Base uses natural language understanding to traverse the scientific research questions and answers stored in it. If there is no question consistent with the scientific research question in the Cloud Question Knowledge Base, the scientific research question will be integrated into a learning instruction and sent to the intelligent learning module; If there is a question consistent with the scientific research question in the Yunwen knowledge base, the question consistent with the scientific research question and its corresponding correct answer will be retrieved and sent to the question-asking end corresponding to the scientific research question; When the intelligent learning module receives the learning instruction, it uses CNN and Apriori algorithms to determine the target field corresponding to the scientific research problem and the relevance of each common key, and locates and analyzes them to obtain the final selection end of the scientific research problem and the relevance index and professional index corresponding to the final selection end; Push scientific research questions to the final selection end, integrate and learn the received scientific research answers to obtain several target answers to the scientific research questions and their corresponding confidence levels, and integrate them into output instructions and send them to the question-answering output module; When the question-answer output module receives an output instruction, it parses the output instruction to obtain several target answers to the scientific research question and their corresponding confidence levels, and sorts each answer in order of its confidence level and sends it to the corresponding scientific research question-asking end as the final answer to the scientific research question; the scientific research question-asking end can provide feedback on several answers in the final answer to the scientific research question, where the feedback includes adoption or rejection; when the feedback is received, a cumulative analysis is performed on each source end to obtain a professional cumulative coefficient, and the answer that is adopted is used as the correct answer to the corresponding scientific research question, and is sent to the Cloud Question Knowledge Base for storage; The integration process of sending output instructions is as follows: The received scientific research answers to scientific research questions are sorted into several replies using natural language understanding. Specifically, when the descriptions of one or more scientific research answers to scientific research questions are the same, the corresponding one or more scientific research answers to scientific research questions are one reply, and the final selection end corresponding to the one or more scientific research answers is recorded as the source end of the reply as p, P = 1, 2, 3 ... P, P is a positive integer, P represents the total number of source segments corresponding to the reply, and p represents the sequence number of any source end in the reply; thus, several source ends of each reply can be obtained, and the correlation index Dhp and professional index Lhp of the several source ends corresponding to each reply are calculated by the set formula The confidence LD corresponding to the answer is calculated, where h6 and h7 are the set weight coefficients respectively; when the confidence is greater than the set confidence threshold, the answer corresponding to the confidence is recorded as the target answer, thereby obtaining several target answers to scientific research questions and their corresponding confidences, and integrating them into output instructions and sending them to the question-answering output module; The specific process of cumulative analysis is as follows: Obtain behavioral data from each source, where the behavioral data includes the number of answers, the answer time and feedback status corresponding to each answer, and divide the corresponding answers into valid answers and invalid answers according to the feedback status, count the number of valid answers and invalid answers respectively, and record them as D1 and D2 respectively; calculate the time difference between the answer time corresponding to the valid answer and the invalid answer and the system time to obtain the time interval corresponding to each valid answer and the invalid answer, and record them as D3j and D4k respectively, where j = 1, 2, 3 ... J, k = 1, 2, 3 ... K, J and K are both positive integers, J represents the total number of valid answers, K represents the total number of invalid answers; j represents the sequence number of any valid answer, and k represents the sequence number of any invalid answer; through the set formula Calculate the professional cumulative coefficient β of each source end and send it to the intelligent learning module; Among them, f1, f2, f3, and f4 are the set weight coefficients respectively; The specific process of positioning analysis is as follows: 401: Record the researcher end whose research field is consistent with the target field corresponding to the scientific research question as the preliminary selection end, and perform correlation analysis on the research field of the preliminary selection end and the keywords corresponding to the research field to obtain a correlation index about the scientific research question; 402: Record the online preliminary selection end as the re-selection end. If the correlation index of the re-selection end is greater than the set correlation threshold, record the re-selection end as the final selection end. Thus, the final selection end of the scientific research question and the correlation index corresponding to each final selection end can be obtained, and the scientific research question is pushed to the final selection end. 403: Retrieve the registration information of the final selection terminal and measure its professional ability to obtain the professional index of the final selection terminal; The specific process of conducting correlation analysis on the research fields at the preliminary end and the keywords corresponding to the research fields is as follows: Retrieve the research field and the keywords corresponding to the research field from the preliminary selection end, and compare them with the common keywords corresponding to the target field to obtain the keywords with common points between the two and record them as common keywords. The correlation degree corresponding to the common keywords that can be matched is recorded as Di, i=1,2,3...I, I is a positive integer, I represents the number of common keywords in the preliminary selection end, and i represents the sequence number of any common keyword among them; Compare and analyze the relevance of common keywords with the set relevance interval to classify the relevance into high relevance, medium relevance and low relevance; count the cumulative times of high relevance, medium relevance and low relevance respectively; The correlation degree Di, the cumulative number of highly correlated cases C1, the cumulative number of moderately correlated cases C2, and the cumulative number of lowly correlated cases C3 are calculated by the formula The correlation index Dh of the preliminary selection end on scientific research issues is calculated, where h1, h2, and h3 are the set weight coefficients, and h1>h2>h3>0; The specific process of measuring professional capabilities based on the registration information of the final selection end is as follows: Set a base value for each educational background, and compare the educational background of the final selection end with all the set educational backgrounds to match the corresponding base value; Set a professional base value corresponding to different professional titles, and compare the professional title corresponding to the final selection end with all the set professional titles to match the corresponding professional base value; Each user terminal is set to correspond to a cumulative professional coefficient, and the final selected terminal is compared with all user terminals to match the corresponding professional cumulative coefficient; The professional index Lh is obtained by calculating the academic base value L1, occupational base value L2 and professional cumulative coefficient β through the set formula Lh=β×(h4×L1+h5×L2), where h4 and h5 are the set weight coefficients for formulaic calculation and analysis to obtain the professional index.

Images