Expert recommendation method and system, electronic equipment and storage medium

Abstract

The invention provides an expert recommendation method and system, electronic equipment and a storage medium, and the method comprises the steps: receiving project text information of a target project and a screening condition of the target project sent by a user side, and carrying out the feature extraction of the project text information, the extracted item feature vectors are matched with expert feature vectors of all experts in a pre-created expert vector database, so that an initial expert recommendation list is generated according to a matching result, and credibility information of all the experts in the initial expert recommendation list is inquired; sorting experts in the initial expert recommendation list based on the creditworthiness information, screening the sorted experts in the expert recommendation list according to the screening condition of the target project, obtaining a final expert recommendation list, and sending the final expert recommendation list to the user side; according to the method, quick and accurate expert recommendation can be realized, the user satisfaction is effectively improved, the requirements of different users are better met, and the personalized degree of recommendation is enhanced.

Description

Technical Field

[0001] The present application relates to the technical field of expert recommendation, and in particular to an expert recommendation method, system, electronic device and storage medium. Background Art

[0002] In the field of scientific and technological innovation and technology research and development, the success of a project often depends to a large extent on the participation and contribution of professional talents or experts. In order to ensure the smooth progress of the project and the resolution of technical problems, it is necessary to accurately match experts with corresponding professional backgrounds and rich experience.

[0003] However, traditional expert recommendation methods mostly rely on manual screening and recommendation. This manual recommendation method is not only inefficient and difficult to adapt to the fast-paced R&D needs, but is also easily affected by limitations such as the subjective cognitive bias and incomplete information of the recommenders. It is difficult to meet the needs of fast and accurate matching of experts, which to a certain extent restricts the progress and effectiveness of scientific and technological innovation and technology R&D. Summary of the invention

[0004] In view of the shortcomings of the prior art mentioned above, the present application provides an expert recommendation method, system, electronic device and storage medium to solve the technical problems that the above-mentioned manual expert recommendation method is inefficient and is easily affected by the limitations of the recommender's subjective cognitive bias and incomplete information, making it difficult to meet the needs of quickly and accurately matching experts.

[0005] The present application provides an expert recommendation method, which includes: receiving project text information of a target project and screening conditions of the target project sent by a user terminal; performing feature extraction on the project text information, and matching the extracted project feature vector with the expert feature vector of each expert in a pre-created expert vector database, so as to generate an initial expert recommendation list according to the matching result; querying the credibility information of each expert in the initial expert recommendation list, sorting the experts in the initial expert recommendation list based on the credibility information, and screening the experts in the sorted expert recommendation list according to the screening conditions of the target project, so as to obtain a final expert recommendation list and send it to the user terminal.

[0006] In one embodiment of the present application, feature extraction is performed on the project text information, including: preprocessing the project text information to obtain preprocessed project text information; extracting keywords of the target project from the preprocessed project text information, and identifying the research field of the target project based on the preprocessed project text information; performing text splicing on the keywords and research fields of the target project, and converting the spliced ​​text into a vector to obtain the project feature vector.

[0007] In one embodiment of the present application, the extracted project feature vector is matched with the expert feature vectors of each expert in a pre-created expert vector database, including: inputting the project feature vector and the expert feature vector of each expert in the expert vector database into a semantic matching model, calculating the similarity between the project feature vector and the expert feature vector of each expert in the expert vector database through the semantic matching model, wherein the semantic matching model is pre-trained based on the historical project feature vectors of each expert's historical projects and the expert feature vectors of each expert; comparing a preset similarity threshold with each similarity, determining experts related to the target project as relevant experts based on the comparison result, and querying the expert text information of each relevant expert to generate the initial expert recommendation list based on all relevant experts and the expert text information of each relevant expert.

[0008] In one embodiment of the present application, the experts in the initial expert recommendation list are sorted based on the credibility information, including: determining the comprehensive score of each expert in the initial expert recommendation list based on the credibility information and the corresponding similarity of each expert in the initial expert recommendation list, so as to sort the experts in the initial expert recommendation list according to the comprehensive score to obtain the sorted expert recommendation list; wherein, determining the comprehensive score of an expert based on the credibility information and the corresponding similarity of the expert includes extracting the activity information, historical cooperation record information and user feedback information of the expert from the credibility information of the expert; determining the expert's activity information based on the activity information of the expert The activity score of the expert is determined based on the expert's historical cooperation record information, and the expert's user feedback score is determined based on the expert's user feedback information, the activity information includes at least one of the number of research activities, the number of papers, and the number of scientific research projects within a preset time period, the historical cooperation record information includes at least one of the cooperation frequency, the cooperation success rate, and the project completion quality, and the user feedback information includes at least one of user satisfaction surveys, user evaluations, and user suggestions; the expert's activity score, cooperation score, user feedback score, and corresponding similarity are weightedly calculated according to preset allocation weights to obtain a comprehensive score of the expert.

[0009] In one embodiment of the present application, the experts in the sorted expert recommendation list are screened according to the screening conditions of the target project, including: constructing a query statement for the target project according to the screening conditions of the target project, the screening conditions including at least one of the technical field, expected results, and project stage; searching the expert text information of each expert in the sorted expert recommendation list according to the query statement of the target project, and generating the final expert recommendation list according to the search results.

[0010] In one embodiment of the present application, after obtaining the final expert recommendation list and sending it to the user terminal, the method includes: receiving recommendation feedback information on the final expert recommendation list sent by the user terminal; performing parameter adjustment according to the recommendation feedback information, wherein the parameter adjustment includes adjusting at least one of the similarity threshold, adjusting the allocation weight and adjusting the semantic matching model.

[0011] In one embodiment of the present application, before matching the extracted project feature vector with the expert feature vector of each expert in a pre-created expert vector database, the method includes: obtaining expert text information of different experts; preprocessing the expert text information of each expert to obtain processed expert text information of each expert; constructing an expert portrait of each expert based on the academic information of each expert, and converting the expert portrait of each expert into a vector, so as to create the expert vector database based on the expert feature vectors of all experts obtained by conversion, wherein the expert text information at least includes the academic information.

[0012] In one embodiment of the present application, an expert recommendation system is also provided, the system comprising: a data receiving module, used to receive project text information of a target project and screening conditions of the target project sent by a user terminal; an information processing module, used to extract features of the project text information, and match the extracted project feature vector with the expert feature vector of each expert in a pre-created expert vector database, so as to generate an initial expert recommendation list according to the matching results; query the credibility information of each expert in the initial expert recommendation list, sort the experts in the initial expert recommendation list based on the credibility information, and screen the experts in the sorted expert recommendation list according to the screening conditions of the target project to obtain a final expert recommendation list; a data sending module, used to send the final expert recommendation list to the user terminal.

[0013] In one embodiment of the present application, an electronic device is also provided, comprising: one or more processors; a storage device for storing one or more programs, and when the one or more programs are executed by the one or more processors, the electronic device implements the expert recommendation method as described above.

[0014] In one embodiment of the present application, a computer-readable storage medium is further provided, on which a computer program is stored. When the computer program is executed by a processor of a computer, the computer executes the expert recommendation method as described above.

[0015] Beneficial effects of the present invention: The present invention provides an expert recommendation method, system, electronic device and storage medium. The method extracts features from the project text information reported by the user to analyze the user's project needs and preferences, and then matches the appropriate experts, which can effectively avoid the subjective cognitive bias and information missing limitations of manual recommendation, and achieve fast and accurate expert recommendation. In addition, the expert recommendation list is sorted according to the reputation information of the expert himself, which can effectively improve user satisfaction. The expert recommendation list is screened according to the screening conditions customized by the user, which can better meet the needs of different users and enhance the personalization of the recommendation.

[0016] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a schematic diagram of an implementation environment of an expert recommendation method shown in an exemplary embodiment of the present application;

[0018] Figure 2 is a flow chart of an expert recommendation method shown in an exemplary embodiment of the present application;

[0019] Figure 3 is a schematic diagram of an expert recommendation process according to a specific embodiment of the present application;

[0020] Figure 4 is a block diagram of an expert recommendation system shown in an exemplary embodiment of the present application;

[0021] Figure 5 It is a structural schematic diagram of an electronic device shown in an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0022] The following describes the embodiments of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0023] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application, and thus the drawings only show components related to the present application rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed at will, and the component layout may also be more complicated.

[0024] It should be noted that in this application, "first", "second", etc. are only used to distinguish similar objects, and are not used to limit the order or precedence of similar objects. The variations of "including", "having", etc. described above indicate that the scope covered by the subject of the word is not exclusive except for the examples shown by the word.

[0025] It is understood that the various numbers, step numbers, and other reference numerals recorded in this application are distinguished for the convenience of description and are not intended to limit the scope of this application. The size of the reference numerals in this application does not mean the order of execution. The execution order of each process should be determined by its function and internal logic.

[0026] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application. However, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present application difficult to understand.

[0027] The embodiments of the present application respectively propose an expert recommendation method, an expert recommendation system, an electronic device, a computer-readable storage medium, and a computer program product, which will be described in detail below.

[0028] See also Figure 1 , Figure 1 It is a schematic diagram of an implementation environment of an expert recommendation method shown in an exemplary embodiment of the present application.

[0029] like Figure 1 As shown, the implementation environment may include a user terminal 110 and a server terminal 120, wherein the user terminal 110 may be at least one of a desktop computer, a laptop computer, a tablet computer, a mobile phone, a smart TV, etc., which is not limited here. The server terminal 120 may be an independent physical server, or a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms, which is not limited here. The user can input the project text information and screening conditions of the target project through the user terminal 110, and send them to the server terminal 120, so that the server terminal 120 can implement expert recommendations.

[0030] Schematically, the server 120 receives the project text information of the target project and the screening conditions of the target project sent by the user end 110; extracts features from the project text information, and matches the extracted project feature vector with the expert feature vector of each expert in the pre-created expert vector database to generate an initial expert recommendation list according to the matching result; queries the credibility information of each expert in the initial expert recommendation list, sorts the experts in the initial expert recommendation list based on the credibility information, and screens the experts in the sorted expert recommendation list according to the screening conditions of the target project, obtains the final expert recommendation list and sends it to the user end 110. It can be seen that the technical solution of the embodiment of the present application can effectively avoid the subjective cognitive bias and information loss of manual recommendation by extracting features from the project text information reported by the user to analyze the user's project needs and preferences, and then match the appropriate experts, so as to achieve fast and accurate expert recommendation. In addition, sorting the expert recommendation list according to the expert's own credibility information can effectively improve user satisfaction, and screening the expert recommendation list according to the user's personalized screening conditions can better adapt to the needs of different users and enhance the personalized degree of recommendation.

[0031] It should be noted that the expert recommendation method provided in the embodiment of the present application is generally specifically executed by the server 120 .

[0032] See also Figure 2 , Figure 2 is a flowchart of an expert recommendation method shown in an exemplary embodiment of the present application. The expert recommendation method can be applied to Figure 1 The implementation environment shown in FIG. 1 is specifically implemented by the server 120 in the implementation environment. It should be understood that the expert recommendation method can also be applied to other exemplary implementation environments and be specifically implemented by devices in other implementation environments. This embodiment does not limit the implementation environment to which the expert recommendation method is applicable. Figure 2 As shown, in an exemplary embodiment, the expert recommendation method includes at least steps S210 to S230, which are described in detail as follows:

[0033] Step S210: receiving the project text information of the target project and the screening condition of the target project sent by the user terminal.

[0034] In one embodiment of the present application, a user interaction interface may be designed to collect project text information of the target project filled in by the user, including at least one of the project name, project description, and cooperation requirements. The user interaction interface may also be designed to provide clear input fields and options in the interface so that the user can input or select corresponding recommended parameters according to the specific needs of the target project, including at least one of the technical field, expected results, project stage, etc., to form screening conditions for the target project. After the user fills in the project text information of the target project through the user interaction interface in the user end according to the needs, and inputs or selects the recommended parameters of the target project, the user end sends the screening conditions formed by the project text information and the recommended parameters to the server end.

[0035] In a specific embodiment of the present application, for the user interaction interface related to the filtering conditions, first, the filterable recommended parameters can be clearly defined, such as technical fields, expected results, project stages, etc., and clear options or input rules can be defined for each recommended parameter. Then, an intuitive user interaction interface is developed to allow the user to input or select recommended parameters through drop-down menus, check boxes, text boxes and other controls on the user side, thereby forming filtering conditions.

[0036] Step S220 , extracting features from the project text information, and matching the extracted project feature vectors with the expert feature vectors of each expert in a pre-created expert vector database, so as to generate an initial expert recommendation list according to the matching results.

[0037] In one embodiment of the present application, NLP (Natural Language Processing) technology, including part-of-speech tagging, NER (Named Entity Recognition), dependency syntax analysis, etc., and data mining technology, including cluster analysis, classification analysis, etc., are used to deeply analyze project text information, extract key features and terms, and convert them into quantifiable feature vectors as project feature vectors. In addition, the information of multiple experts can be obtained in advance and feature extraction can be performed to obtain the expert feature vectors of each expert, so as to create an expert vector database based on the expert feature vectors of all experts. Then, the project feature vector of the target project is matched with the expert feature vectors of each expert in the expert vector database, and experts suitable for the target project are obtained to form an initial expert recommendation list.

[0038] In one embodiment of the present application, feature extraction is performed on project text information, including: preprocessing the project text information to obtain preprocessed project text information; extracting keywords of the target project from the preprocessed project text information, and identifying the research field of the target project based on the preprocessed project text information; text splicing the keywords and research fields of the target project, and converting the spliced ​​text into a vector to obtain a project feature vector.

[0039] In this embodiment, the project text information submitted by the user through the user terminal is first preprocessed, including at least one of text cleaning, word segmentation, part-of-speech tagging, named entity recognition, etc., to ensure the quality and consistency of the text data, wherein text cleaning specifically includes removing useless symbols, removing stop words, etc. Then, the TextRank algorithm can be used to identify the keywords in the preprocessed project text information, which can represent the core content of the target project. The topic model algorithm can also be applied to discover the hidden topic distribution from the description of the target project in the project text information to identify the main research field or direction of the target project. Finally, the keywords and research fields of the target project are integrated or spliced ​​from the text, and the NLP technology can be used to convert the spliced ​​text from the text format into a numerical vector to form a comprehensive feature vector as the project feature vector.

[0040] In one embodiment of the present application, before matching the extracted project feature vector with the expert feature vector of each expert in a pre-created expert vector database, the method includes: obtaining expert text information of different experts; preprocessing the expert text information of each expert to obtain processed expert text information of each expert; constructing an expert portrait of each expert based on the academic information of each expert, and converting the expert portrait of each expert into a vector, so as to create an expert vector database based on the expert feature vectors of all experts obtained by the conversion, wherein the expert text information includes at least academic information.

[0041] In this embodiment, first, according to the needs of the field of science and technology innovation, the key fields required for the expert database can be determined, including basic information work experience, academic information, etc., wherein the basic information includes at least one of name, title, educational background and contact information, and the academic information includes at least one of research field, research direction, professional skills, academic achievements (such as: academic papers, research results), etc. Of course, academic information can also involve expert publications, patents, project reports, project experience, etc., to ensure the comprehensiveness and accuracy of the expert recommendation system, and key fields can also include the expert's available time, geographical location and cooperation preferences, so as to better match project needs. Then, data collection tools can be used and according to the determined key fields from various public resources such as academic papers, professional websites, social media, etc., to automatically capture relevant information as expert text information, and pre-process the expert text information, including cleaning and formatting. Extract academic information such as research results, research fields, academic papers, etc. from the pre-processed expert text information, so as to conduct in-depth analysis of the expert's research results, research fields, etc. Academic information, extract keywords, themes and concepts, etc., construct expert portraits, and convert expert portraits from text format to numerical feature vectors as expert feature vectors. Create an expert vector database and import the expert feature vectors of all experts into the expert vector database.

[0042] In addition, an expert text database may be created, and the preprocessed expert text information and / or expert portraits may be imported into the expert text database, so as to read the expert text information and / or expert portraits from the expert text database to generate an initial expert recommendation list. Schematically, a relational database model may be used for storage of the expert text database to ensure the structuring and queryability of the data.

[0043] In addition, corresponding identification information can be configured for each expert, which can be a number or the expert's ID number, so that the expert's information can be read from the database based on the expert's identification information. A data update mechanism can also be set up to regularly maintain and update the expert's text information to ensure the timeliness and accuracy of the database.

[0044] In one embodiment of the present application, the extracted project feature vector is matched with the expert feature vectors of each expert in a pre-created expert vector database, including: inputting the project feature vector and the expert feature vector of each expert in the expert vector database into a semantic matching model, calculating the similarity between the project feature vector and the expert feature vector of each expert in the expert vector database through the semantic matching model, wherein the semantic matching model is pre-trained based on historical project feature vectors of historical projects of each expert and the expert feature vectors of each expert; comparing a preset similarity threshold with each similarity, determining experts related to the target project as relevant experts based on the comparison result, and querying the expert text information of each relevant expert to generate an initial expert recommendation list based on all relevant experts and the expert text information of each relevant expert.

[0045] In this embodiment, the semantic matching model is trained in advance, and the historical project text information of the expert can be extracted from the expert text information of each expert, or the historical project text information of each expert can be read from the pre-created historical project database, and then feature extraction is performed to obtain the historical project feature vectors of the historical projects of each expert, based on the ESimCSE (Enhanced SimCSE, a semantic similarity matching model based on momentum contrast learning) algorithm, and the historical project feature vectors of the historical projects of each expert and the expert feature vectors of each expert are used to match the semantic matching model. During the training process, the algorithm learns how to identify the expert feature vectors most relevant to it through the project feature vectors. The model performance can be improved by adjusting hyperparameters such as the learning rate, batch size, and number of training rounds, and the performance of the model is monitored and tuned using a validation set. The performance of the model is evaluated on an independent test set, and the model is optimized according to the evaluation results. Finally, the trained semantic matching model is deployed to the server, and it is ensured that the model can accept new inputs and provide accurate matching results.

[0046] The server inputs the project feature vector of the target project and the expert feature vectors of each expert in the expert vector database into the deployed semantic matching model, and calculates the similarity between the target project and each expert through the semantic matching model, which involves calculating the similarity index between the two vectors. Specifically, the semantic matching model can use metrics such as cosine similarity to accurately evaluate the degree of match between the target project and each expert.

[0047] A similarity threshold can be predetermined to screen out experts who are highly relevant to the target project requirements, wherein the setting of the similarity threshold can be based on statistical analysis of historical data or manual evaluation of experts, and can be dynamically adjusted later. Experts corresponding to similarities greater than or equal to the similarity threshold can be determined as experts related to the target project, i.e., relevant experts, and expert text information of each relevant expert can be queried from a pre-created expert text database based on the identification information of the relevant experts, so as to generate an initial expert recommendation list based on all relevant experts and the expert text information of the relevant experts. When generating the preliminary expert recommendation list, all relevant experts can also be ranked according to the calculated similarity.

[0048] Step S230, query the credibility information of each expert in the initial expert recommendation list, sort the experts in the initial expert recommendation list based on the credibility information, and filter the experts in the sorted expert recommendation list according to the filtering conditions of the target project to obtain the final expert recommendation list and send it to the user end.

[0049] In one embodiment of the present application, the reputation information includes at least one of the expert's activity, historical cooperation records, user ratings or satisfaction, user suggestions or opinions, etc. A rating mechanism can be established based on the reputation information to calculate the comprehensive score of each expert, so as to sort the experts in the expert recommendation list according to the comprehensive score, thereby improving the user's satisfaction with the recommended experts. The experts in the expert recommendation list can also be screened according to the screening conditions input by the user, so as to recommend experts that are more suitable for the user's personalized needs, thereby enhancing the personalization of the recommendation.

[0050] In one embodiment of the present application, ranking the experts in the initial expert recommendation list based on the reputation information includes: determining the comprehensive score of each expert in the initial expert recommendation list based on the reputation information and the corresponding similarity of each expert in the initial expert recommendation list, and ranking each expert in the initial expert recommendation list according to the comprehensive score to obtain a ranked expert recommendation list;

[0051] Wherein, based on the credibility information and corresponding similarity of an expert, determining the comprehensive score of the expert includes extracting the expert's activity information, historical cooperation record information and user feedback information from the expert's credibility information; determining the expert's activity score according to the expert's activity information, determining the expert's cooperation score according to the expert's historical cooperation record information, and determining the expert's user feedback score according to the expert's user feedback information, the activity information including at least one of the number of research activities, the number of papers and the number of scientific research projects within a preset time period, the historical cooperation record information including at least one of the frequency of cooperation, the success rate of cooperation and the quality of project completion, and the user feedback information including at least one of user satisfaction survey, user evaluation and user suggestion; weighted calculation is performed on the expert's activity score, cooperation score, user feedback score and corresponding similarity according to preset allocation weights to obtain the expert's comprehensive score.

[0052] In this embodiment, for the evaluation of activity scores, expert activity indicators can be defined, including the number of recent research activities, the number of published papers, the number of scientific research projects participated in, etc., and the activity of each expert can be scored through big data analysis. For the evaluation of cooperation scores, a historical project database can be established in advance to store data information of historical cooperation projects between users and experts, and the cooperation records of experts can be extracted from the historical project database, including cooperation frequency, cooperation success rate, project completion quality, etc., which are used as an important basis for recommendation to score the degree of cooperation of each expert. For the evaluation of user feedback scores, a user interaction interface can be designed in advance to collect user feedback information on recommended experts and / or cooperative experts, including user satisfaction surveys, user evaluation scores, user suggestions or opinions, etc., and convert user feedback into quantitative indicators to form user feedback scores.

[0053] Pre-configure multi-dimensional weight distribution and assign weights to each dimension of activity, cooperation record, user feedback and similarity. Build a comprehensive scoring model, combine similarity with the corresponding scores of activity, cooperation record, and user feedback, and calculate the comprehensive score of each expert through a weighted average algorithm. Sort the experts in the initial expert recommendation list based on the comprehensive score to generate a sorted expert recommendation list. By comprehensively considering the similarity between expert information and project requirements, expert activity, expert professional field relevance, and historical project participation success rate, etc., use a weighted algorithm to sort the matching results to ensure that the experts in the recommendation list are not only highly matched with project requirements in terms of semantics, but also most suitable in terms of professional ability and experience, thereby improving the efficiency and success rate of project team formation.

[0054] In one embodiment of the present application, experts in the sorted expert recommendation list are screened according to the screening conditions of the target project, including: constructing a query statement for the target project according to the screening conditions of the target project, the screening conditions including at least one of the technical field, expected results, and project stage; searching the expert text information of each expert in the sorted expert recommendation list according to the query statement of the target project, and generating a final expert recommendation list according to the search results.

[0055] In this embodiment, according to the screening conditions input by the user, a database query statement is dynamically constructed to retrieve matching expert records from the expert recommendation list, wherein when the screening conditions are converted into a query statement, fuzzy matching or multi-condition combination query can be supported. To ensure that the screening tool responds quickly, indexing, caching and other technologies can also be used to improve query efficiency for processing large amounts of data.

[0056] In addition, a real-time feedback mechanism can be established, including preview of filtering results, suggestions for modifying filtering conditions, etc., to enhance the user's filtering experience.

[0057] In one embodiment of the present application, after obtaining the final expert recommendation list and sending it to the user end, the method includes: receiving recommendation feedback information on the final expert recommendation list sent by the user end; adjusting parameters according to the recommendation feedback information, wherein the parameter adjustment includes adjusting at least one of a similarity threshold, adjusting an allocation weight, and adjusting a semantic matching model.

[0058] In this embodiment, a user feedback interface can be pre-designed to enable users to easily evaluate the expert recommendation results. The interface may include a rating system, a comment box, and possible improvement suggestion fields. Then, the user's recommendation feedback information is collected and stored, including satisfaction or acceptance ratings, text comments, and any other relevant inputs. The user's recommendation feedback information is analyzed mainly with a sentiment analysis algorithm to extract key opinions and common problems so as to make targeted improvements to the expert recommendation. For example, according to the feedback analysis results, parameters including similarity thresholds, weight allocation, etc. can be dynamically adjusted to optimize the relevance and accuracy of the expert recommendation. The semantic matching model can also be retrained regularly with new user recommendation feedback information to ensure that the model can learn and adapt to changes in user preferences. In addition, an effect evaluation mechanism can be implemented to evaluate the impact of parameter adjustment on recommendation performance by comparing the recommendation results before and after the adjustment. After the recommendation algorithm is updated, the user is notified through the user end to enhance user participation and the interactivity of the expert recommendation system.

[0059] In one embodiment of the present application, a user interaction interface may also be designed to provide quick access to detailed information of experts and a function of directly communicating with experts.

[0060] See also Figure 3 , Figure 3 It is a schematic diagram of the expert recommendation process shown in a specific embodiment of the present application.

[0061] like Figure 3 As shown, the expert recommendation process is as follows:

[0062] 1. Construct an expert database, including an expert text database and an expert vector database. Among them, the expert text database is used to collect basic information of experts, including name, title, educational background and contact information, etc. It can also cover key information such as the expert's research field, professional skills, main research direction and key achievements as academic information. In order to ensure the depth and breadth of the database, the expert's published results, including academic papers, patents, project reports, etc., can also be included as academic information. Through natural language processing and machine learning algorithms, in-depth analysis of academic information such as experts' research results, research fields, academic papers, etc. is carried out to extract keywords, themes and concepts, build expert portraits of experts, and convert them into expert feature vectors to be added to the expert vector database. This process not only includes direct analysis of text content, but also involves the evaluation of the influence and relevance of expert results, such as the number of citations, cooperation networks and rankings in the research field, etc., which can provide accurate, detailed and highly relevant expert information for the expert recommendation system, and provide a solid data foundation for subsequent semantic matching and expert recommendation.

[0063] 2. Project requirements analysis. Through text analysis algorithms, the project text information such as project name, project description and cooperation requirements entered by the user is carefully parsed to identify and extract key features such as key words, phrases and concepts in the project description, and then construct the project feature vector of the project requirements. During the processing, the text is first preprocessed, including word segmentation, stop word removal, part-of-speech tagging and named entity recognition, to ensure the quality and consistency of the text data. Subsequently, by applying technologies such as topic modeling, sentiment analysis and semantic role labeling, the user's demand background, goals and expectations, as well as potential cooperation intentions are deeply understood.

[0064] 3. Perform in-depth semantic analysis and similarity calculation on project requirements and expert information to achieve accurate matching and generate a preliminary recommendation list. Under the deep learning framework, the semantic representation between project requirements and expert information is learned through the neural network structure to understand the deep semantics of the text, including word meaning, syntactic structure and contextual relationship. And through a sophisticated similarity calculation mechanism, the semantic similarity between scientific and technological innovation project requirements and expert information can be quantified.

[0065] 4. Integrate information from multiple dimensions to optimize the preliminary recommendation list and provide more accurate matching results. Comprehensively consider information such as the expert's activity, historical cooperation records, and user feedback to achieve accurate recommendations. In terms of activity, the frequency of experts' activities on the platform, including but not limited to the number of logins and project participation, can be tracked as a basis for evaluating their current activity. Historical cooperation records cover the experts' past successful cases, cooperation evaluations, and completion quality of projects. This information helps predict the experts' potential performance in similar projects. In the user feedback section, the system continuously adjusts the recommendation strategy by collecting users' evaluations of expert services, including satisfaction, professionalism, and response speed. Finally, a personalized and dynamically updated expert recommendation list is generated to provide users with customized expert recommendation services.

[0066] 5. Provide screening tools to support users to input key parameters such as technical fields and expected results to form screening conditions to customize the recommendation list.

[0067] 6. Establish a user feedback and optimization mechanism. Through the feedback mechanism, users can evaluate the expert recommendation results. These evaluations are used as input to optimize and adjust the performance of the expert recommendation system and improve the quality of recommendations. By analyzing the user feedback data, the relevance and accuracy issues of the recommendation results are identified, and then the recommendation logic is fine-tuned, including similarity thresholds, weight allocation, semantic matching models, etc. This feedback-based optimization method enables the expert recommendation system to adapt to changes in user needs, thereby providing more personalized and accurate recommendation services.

[0068] Prior to this, a user interaction interface can also be designed, which is designed with clarity and logic as the principles to ensure that users can interact with the expert recommendation system without obstacles. The interface can provide clear input fields and options, allowing users to enter or select corresponding recommendation parameters to customize the recommendation results according to the specific needs of scientific and technological innovation projects, including key parameters such as technical fields, expected results, and project stages. In addition, the interface can also provide quick access to detailed information about experts, including the qualifications of experts, past projects, and user evaluations, so that users can fully understand the background of recommended experts. At the same time, the interface can also integrate the function of communicating with experts, and users can establish contact and communicate with experts in the form of online messages or emails.

[0069] For the detailed process of this specific embodiment, please refer to the records in the aforementioned embodiments, which will not be repeated here. This specific embodiment achieves accurate expert recommendation by deeply analyzing the scientific and technological innovation project needs and preferences reported by users. The semantic matching model improves the accuracy of the recommendation, and the expert recommendation list is optimized through multi-dimensional comprehensive scoring and user-entered screening conditions, thereby enhancing the personalization of the recommendation. In addition, through the user feedback mechanism, the expert recommendation system can continuously learn and optimize according to user feedback, so as to adapt to the ever-changing user needs and market environment. Therefore, this specific embodiment provides a new solution for expert recommendation in the field of scientific and technological innovation, greatly improves the matching efficiency of expert resources and user satisfaction, and provides strong support for technological progress and innovation in related fields.

[0070] See also Figure 4 , Figure 4 is a block diagram of an expert recommendation system shown in an exemplary embodiment of the present application. The system can be applied to Figure 1 The system can also be applied to other exemplary implementation environments, and this embodiment does not limit the implementation environment to which the system is applied. Figure 4 As shown, the exemplary expert recommendation system includes: a data receiving module 410, which is used to receive project text information of a target project and screening conditions of the target project sent by a user terminal; an information processing module 420, which is used to extract features from the project text information, and match the extracted project feature vector with the expert feature vector of each expert in a pre-created expert vector database to generate an initial expert recommendation list according to the matching results; query the credibility information of each expert in the initial expert recommendation list, sort the experts in the initial expert recommendation list based on the credibility information, and screen the experts in the sorted expert recommendation list according to the screening conditions of the target project to obtain a final expert recommendation list; a data sending module 430, which is used to send the final expert recommendation list to the user terminal.

[0071] In this embodiment, the expert recommendation system is formed by a data receiving module 410, an information processing module 420 and a data sending module 430, and is integrated in the server. The data receiving module 410 can be a data receiver, a network card, etc., the information processing module 420 can be a computer device such as a computer, a computing cluster, a microcomputer, an embedded computer, a neural network computer, etc., and the data sending module 430 can be a data transmitter, a network card, etc., which are not limited here.

[0072] It should be noted that the expert recommendation system provided in the above embodiment and the expert recommendation method provided in the above embodiment belong to the same concept, wherein the specific manner in which each module and unit performs the operation has been described in detail in the method embodiment, and will not be repeated here. In practical applications, the expert recommendation system provided in the above embodiment can distribute the above functions to different functional modules as needed, that is, divide the internal structure of the system into different functional modules to complete all or part of the functions described above, and this is not limited here.

[0073] This embodiment also provides an electronic device, including: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the electronic device implements the expert recommendation method provided in the above-mentioned embodiments.

[0074] See also Figure 5 , Figure 5 is a schematic diagram of a structure of an electronic device shown in an exemplary embodiment of the present application. It should be noted that: Figure 5 The electronic device 500 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0075] like Figure 5 As shown, the electronic device 500 includes a processor 501, a memory 502 and a communication bus 503; the communication bus 503 is used to connect the processor 501 and the memory 502; the processor 501 is used to execute the computer program stored in the memory 502 to implement one or more methods in the above embodiments.

[0076] This embodiment also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor of a computer, the computer executes the expert recommendation method as described above. The computer-readable storage medium may be included in the electronic device described in the above embodiment, or may exist independently without being assembled into the electronic device.

[0077] This embodiment also provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the expert recommendation method provided in each of the above embodiments.

[0078] The electronic device provided in this embodiment includes a processor, a memory, a transceiver and a communication interface. The memory and the communication interface are connected to the processor and the transceiver and complete communication with each other. The memory is used to store computer programs, the communication interface is used to communicate, and the processor and the transceiver are used to run the computer program so that the electronic device executes each step of the above method.

[0079] In this embodiment, the memory may include a random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0080] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0081] The computer-readable storage medium in this embodiment can be understood by ordinary technicians in this field: all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, the execution includes the steps of the above-mentioned method embodiments; and the aforementioned storage medium includes: ROM (read-only memory), RAM (random access memory), magnetic disk or optical disk and other media that can store program codes.

[0082] The above embodiments are merely illustrative of the principles and effects of the present application, and are not intended to limit the present application. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed in the present application shall still be covered by the claims of the present application.

Claims

1. An expert recommendation method, characterized in that: The method comprises: Receiving project text information of a target project and screening conditions of the target project sent by a user terminal; Extracting features from the project text information, and matching the extracted project feature vectors with the expert feature vectors of each expert in a pre-created expert vector database, so as to generate an initial expert recommendation list according to the matching results; The credibility information of each expert in the initial expert recommendation list is queried, the experts in the initial expert recommendation list are sorted based on the credibility information, and the experts in the sorted expert recommendation list are screened according to the screening conditions of the target project to obtain a final expert recommendation list and send it to the user end.

2. The expert recommendation method according to claim 1, characterized in that: Extracting features from the project text information includes: Preprocessing the project text information to obtain preprocessed project text information; Extracting keywords of the target project from the preprocessed project text information, and identifying the research field of the target project based on the preprocessed project text information; The keywords and research fields of the target project are concatenated, and the concatenated text is converted into a vector to obtain the project feature vector.

3. The expert recommendation method according to claim 1, characterized in that: Match the extracted project feature vector with the expert feature vector of each expert in the pre-created expert vector database, including: Inputting the project feature vector and the expert feature vector of each expert in the expert vector database into a semantic matching model, and calculating the similarity between the project feature vector and the expert feature vector of each expert in the expert vector database by the semantic matching model, wherein the semantic matching model is pre-trained based on the historical project feature vector of each expert's historical project and the expert feature vector of each expert; The preset similarity threshold is compared with each similarity, and experts related to the target project are determined as relevant experts based on the comparison results, and the expert text information of each relevant expert is queried to generate the initial expert recommendation list based on all relevant experts and the expert text information of each relevant expert.

4. The expert recommendation method according to claim 3, characterized in that: Sorting the experts in the initial expert recommendation list based on the credibility information includes: Determining a comprehensive score of each expert in the initial expert recommendation list based on the credibility information and the corresponding similarity of each expert in the initial expert recommendation list, and sorting each expert in the initial expert recommendation list according to the comprehensive score to obtain the sorted expert recommendation list; Wherein, based on the reputation information of an expert and the corresponding similarity, the comprehensive score of the expert is determined, including: Extracting the expert's activity information, historical cooperation record information and user feedback information from the expert's reputation information; Determine the activity score of the expert according to the activity information of the expert, determine the cooperation score of the expert according to the historical cooperation record information of the expert, and determine the user feedback score of the expert according to the user feedback information of the expert, wherein the activity information includes at least one of the number of research activities, the number of papers, and the number of scientific research projects within a preset time period, the historical cooperation record information includes at least one of the cooperation frequency, the cooperation success rate, and the project completion quality, and the user feedback information includes at least one of the user satisfaction survey, user evaluation, and user suggestion; The expert's activity score, cooperation score, user feedback score and corresponding similarity are weightedly calculated according to preset allocation weights to obtain the expert's comprehensive score.

5. The expert recommendation method according to claim 3, characterized in that: The experts in the sorted expert recommendation list are screened according to the screening criteria of the target project, including: Constructing a query statement for the target project according to the screening condition of the target project, wherein the screening condition includes at least one of a technical field, an expected result, and a project stage; The expert text information of each expert in the ranked expert recommendation list is retrieved according to the query statement of the target project, and the final expert recommendation list is generated according to the retrieval result.

6. The expert recommendation method according to claim 4, characterized in that: After obtaining the final expert recommendation list and sending it to the user terminal, the method includes: Receiving recommendation feedback information on the final expert recommendation list sent by the user terminal; Parameter adjustment is performed according to the recommendation feedback information, wherein the parameter adjustment includes at least one of adjusting the similarity threshold, adjusting the allocation weight, and adjusting the semantic matching model.

7. The expert recommendation method according to any one of claims 1 to 6, characterized in that: Before matching the extracted project feature vector with the expert feature vectors of each expert in the pre-created expert vector database, the method includes: Get expert text information from different experts; Preprocessing the expert text information of each expert to obtain processed expert text information of each expert; An expert portrait of each expert is constructed based on the academic information of each expert, and the expert portrait of each expert is converted into a vector, so as to create the expert vector database based on the expert feature vectors of all experts obtained by the conversion, wherein the expert text information at least includes the academic information.

8. An expert recommendation system, characterized in that: The system comprises: A data receiving module, used for receiving the project text information of the target project and the screening conditions of the target project sent by the user end; An information processing module is used to extract features from the project text information, and match the extracted project feature vector with the expert feature vector of each expert in a pre-created expert vector database, so as to generate an initial expert recommendation list according to the matching result; query the credibility information of each expert in the initial expert recommendation list, sort the experts in the initial expert recommendation list based on the credibility information, and screen the experts in the sorted expert recommendation list according to the screening conditions of the target project to obtain a final expert recommendation list; A data sending module is used to send the final expert recommendation list to the user terminal.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; A storage device, used for storing one or more programs, when the one or more programs are executed by the one or more processors, enables the electronic device to implement the expert recommendation method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor of a computer, the computer is caused to execute the expert recommendation method according to any one of claims 1 to 7.

Images