Data resource recommendation matching method and device, equipment and storage medium

Abstract

The invention provides a data resource recommendation matching method and device, equipment and a storage medium, and the method comprises the steps: obtaining a data resource and a target demand of a user, carrying out the analysis of the demand and the data resource through employing at least two preset data models, extracting the respective key features, forming demand feature information and resource feature information, and carrying out the recommendation matching of the demand feature information and the resource feature information; performing matching according to the extracted feature information to generate a plurality of data resource recommendation lists, and performing integration processing on different recommendation lists to generate a comprehensive recommendation list so as to recommend data resources to the user; according to the method, the recommendation accuracy is improved, the flexibility and adaptability of the system are enhanced, and diversified requirements of different users can be better met.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a data resource recommendation and matching method, device, equipment and storage medium. Background Art

[0002] In the digital age, data has become an indispensable resource for corporate and personal decision-making. With the rapid development of information technology, the amount of data has shown an explosive growth trend. Both companies and individuals have to face a huge amount of information every day. In this case, how to quickly and accurately find data resources that meet specific needs from the huge ocean of data has become one of the key issues that need to be solved urgently. For enterprises, efficient resource acquisition capabilities directly affect the speed and quality of decision-making; for individual users, whether they can quickly find the required information is directly related to the convenience of life and work.

[0003] However, traditional data search and recommendation methods often rely on keyword matching or simple classification and filtering techniques. Although these methods can meet basic search needs to a certain extent, they have exposed many limitations in practical applications. For example, the recommendation accuracy is not high, which makes it difficult to meet the personalized needs of users; the user experience is not good, and there may be a large deviation between the search results and the actual needs of users. More importantly, due to its own limitations, a single data model may not be able to fully capture the complex characteristics of user needs and data resources, which further leads to unsatisfactory recommendation results. Summary of the invention

[0004] In view of the above-mentioned shortcomings of the prior art, the present invention provides a data resource recommendation and matching method, device, equipment and storage medium to solve the above-mentioned technical problems.

[0005] The present invention provides a data resource recommendation and matching method, the method comprising: acquiring data resources and user target requirements; based on at least two preset data models, respectively extracting features of the target requirements and data resources to obtain demand features and resource features; matching the target requirements and data resources according to the demand features and data resource features to generate multiple data resource recommendation lists; merging the multiple data resource recommendation lists generated by different data models to obtain a comprehensive recommendation list, and recommending data resources to the user based on the comprehensive recommendation list.

[0006] In one embodiment of the present invention, the preset data model includes at least two of a first data model, a second data model, and a third data model, wherein the first data model is used to extract technical hot words from target requirements and data resources, and generate a hot word data resource recommendation list corresponding to the target requirements by matching the technical hot words of the data resources with the technical hot words of the target requirements; the second data model is used to determine the target technology field to which the target requirements belong according to preset field division rules, and generate a domain data resource recommendation list matching the target requirements based on the influence list of the target technology field; the third data model is used to convert the description information of the target requirements and data resources into semantic vectors, and generate a semantic data resource recommendation list matching the target requirements by calculating the similarity between the semantic vector of the target requirements and the semantic vector of the data resources.

[0007] In one embodiment of the present invention, a hot word data resource recommendation list corresponding to the target demand is generated, including: extracting the keywords of the target demand and the data resource respectively to obtain the target technology hot words and the data resource technology hot words; calculating the hot word similarity between the target technology hot words and the data resource technology hot words; determining the data resource whose hot word similarity is greater than a preset hot word similarity threshold as the first candidate data resource, and generating a hot word data resource recommendation list based on the hot word similarity sorting of the first candidate data resource.

[0008] In one embodiment of the present invention, a recommendation list of domain data resources matching the target demand is generated, including: based on a preset technical field division rule, dividing the data resources into a plurality of different technical fields, and determining the target technical field to which the target demand belongs; identifying the influence of each data resource in the technical field to which it belongs, and generating an influence list; selecting a preset number of data resources from the influence list in the target technical field as candidate data resources, and generating a recommendation list of data resources in the technical field based on the influence ranking of the candidate data resources.

[0009] In one embodiment of the present invention, a recommendation list of semantic data resources matching the target demand is generated, including: obtaining description information of the target demand and the data resource, and performing data preprocessing on the description information, wherein the data preprocessing at least includes text cleaning, word segmentation, and removal of stop words; converting the description information of the target demand and the data resource into a target semantic vector and a data resource semantic vector, respectively; calculating the vector similarity between the target semantic vector and the data resource semantic vector; determining the data resource whose vector similarity is greater than a preset vector similarity threshold as a third candidate data resource, and generating a recommendation list of semantic data resources based on the vector similarity sorting of the third candidate data resource.

[0010] In one embodiment of the present invention, multiple data resource recommendation lists generated by different data models are merged to obtain a comprehensive recommendation list, including: obtaining the total number of candidate data resource items in the comprehensive recommendation list; selecting the same number of candidate data resources from each data resource recommendation list, or selecting candidate data resources from each data resource recommendation list based on preset model weights; deduplicating the candidate data resources obtained after the selection to generate a comprehensive recommendation list based on the deduplicated candidate data resources, wherein the sum of the number of data resource items selected in each data resource recommendation list is equal to the total number of candidate data resources.

[0011] In one embodiment of the present invention, multiple data resource recommendation lists generated by different data models are merged to obtain a comprehensive recommendation list, which also includes: combining candidate data resources from different data resource recommendation lists; or, taking the number of candidate data resources in any data resource recommendation list as a benchmark, adjusting the number of candidate data resources generated by other data models according to preset model weights, selecting a corresponding number of candidate data resources from each data resource recommendation list and combining them; deduplicating the candidate data resources obtained after the combination to generate a comprehensive recommendation list based on the deduplicated candidate data resources.

[0012] The present application provides a data resource recommendation and matching device, which includes: a data acquisition module, which obtains data resources and user target needs; a recommendation list generation module, which is used to extract features of target needs and data resources based on at least two preset data models to obtain demand features and resource features; according to the demand features and data resource features, the target needs and data resources are matched to generate multiple data resource recommendation lists; a recommendation information integration module, which is used to merge multiple data resource recommendation lists generated by different data models to obtain a comprehensive recommendation list, so as to recommend data resources to users based on the comprehensive recommendation list.

[0013] The present application provides an electronic device, characterized in that it includes a processor, a memory and a communication bus; the communication bus is used to connect the processor and the memory; the processor is used to execute a computer program stored in the memory to implement the data resource recommendation and matching method as described above.

[0014] The present application provides a computer-readable storage medium, characterized in that a computer program is stored thereon, and the computer program is used to enable a computer to execute the data resource recommendation and matching method as described above.

[0015] Beneficial effects of the present invention: The data resource recommendation and matching method of the present invention obtains data resources and user target needs, uses at least two preset data models to conduct in-depth analysis of needs and data resources, extracts their respective key features, and forms demand feature information and resource feature information, ensuring a comprehensive understanding and capture of user needs and data resource characteristics from multiple angles, avoiding the limitations brought by a single model. Based on the extracted feature information, multiple data resource recommendation lists are generated, each list is based on the analysis results of different models, and then these different recommendation lists are integrated and processed to finally generate a comprehensive recommendation list, which recommends to users the data resources that best meet their needs, which not only improves the accuracy and diversity of recommendations, but also enhances the flexibility and adaptability of the system, and can better meet the personalized needs of different users in different scenarios.

[0016] In addition, the present invention effectively solves the problems of low recommendation accuracy and poor user experience in traditional recommendation methods through multi-model fusion, providing users with more accurate and efficient data resource recommendation services, thereby greatly improving user satisfaction and resource utilization efficiency.

[0017] 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

[0018] The drawings herein are incorporated into the specification and constitute a part of the specification, showing embodiments consistent with the present application, and together with the specification, are used to explain the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0019] Figure 1 is a schematic diagram of an implementation environment of a data resource recommendation and matching method shown in an exemplary embodiment of the present application;

[0020] Figure 2 is an implementation flow chart of a data resource recommendation and matching method shown in an exemplary embodiment of the present application;

[0021] Figure 3 is a schematic diagram of a first data model structure shown in an exemplary embodiment of the present application;

[0022] Figure 4 is a schematic diagram of a second data model structure shown in an exemplary embodiment of the present application;

[0023] Figure 5 is a schematic diagram of a third data model structure shown in an exemplary embodiment of the present application;

[0024] Figure 6 is a block diagram of a data resource recommendation and matching device shown in an exemplary embodiment of the present application;

[0025] Figure 7 A schematic diagram of the structure of a computer system suitable for implementing an electronic device of an embodiment of the present application is shown. DETAILED DESCRIPTION

[0026] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention 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 invention. It should be understood that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0027] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention 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 arbitrarily, and the component layout may also be more complicated.

[0028] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention 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 invention difficult to understand.

[0029] Figure 1 It is a schematic diagram of an implementation environment of a data resource recommendation and matching method shown in an exemplary embodiment of the present application.

[0030] like Figure 1As shown, the implementation environment of the data resource recommendation matching method includes a data acquisition device 101 and a computer module 102. In this environment, the data acquisition device 101 is mainly responsible for collecting data resources related to user target needs. These data resources may come from various legal channels such as the Internet, databases, and social media. Specifically, the data acquisition device 101 can be a technical means such as a software interface and API call, or it can be implemented through an application on a user terminal device (such as a smart phone, a computer, etc.). The computer module 102 is the core part of the entire recommendation system, which is responsible for processing and analyzing the information collected by the data acquisition device 101. First, the module will use at least two preset data models to conduct an in-depth analysis of the collected data resources and user needs, aiming to extract their respective key features and form demand feature information and resource feature information. The data model here can be an advanced algorithm model such as content-based filtering, collaborative filtering, and deep learning. Through the precise matching of these feature information, the computer module 102 can generate multiple preliminary data resource recommendation lists.

[0031] In order to further improve the accuracy of recommendations and user experience, the computer module 102 will also integrate different recommendation lists. This process not only takes into account the quality of each list, but also combines factors such as user feedback and popularity to ultimately generate a comprehensive recommendation list. The comprehensive recommendation list can better meet the personalized needs of users, while also ensuring the diversity and novelty of the recommendation results. In this way, the data resource recommendation matching method can effectively help users find the data resources that best meet their needs in a vast amount of information.

[0032] In summary, the implementation environment of the data resource recommendation matching method not only covers a wide range of data sources and efficient collection technologies, but also integrates advanced data analysis and processing capabilities to jointly build an intelligent and personalized recommendation system.

[0033] Figure 2 It is an implementation flow chart of a data resource recommendation and matching method shown in an exemplary embodiment of the present application.

[0034] like Figure 2 As shown, in an exemplary embodiment, the data resource recommendation matching method at least includes steps S210 to S230, which are described in detail as follows:

[0035] Step S210, obtaining data resources and user target requirements.

[0036] In one embodiment of the present application, it is assumed that there is an enterprise service platform that aims to provide consulting and support services in professional fields to enterprises. There are experts from all walks of life on the platform, who have different professional knowledge and experience. At the same time, there are many corporate customers seeking professional advice on the platform, and their business needs and concerns are different. In order to improve the quality of service, the platform plans to use the data resource recommendation matching method proposed in this application to optimize the expert recommendation system. It is necessary to obtain expert-related information as data resources, obtain the target needs of users, that is, enterprises, and store them for subsequent use. The specific method is as follows:

[0037] Regarding obtaining expert information: First, the platform needs to collect basic information from the registered experts' personal profiles, including name, gender, age, education, professional background, areas of expertise, past project experience, etc. Secondly, the professional ability and industry influence of each expert is evaluated through case studies, customer reviews, peer recommendations, etc. provided by the experts. In addition, it is also necessary to use natural language processing technology to process the text content such as the expert's self-introduction and case description, and extract keywords as the expert's ability feature information; at the same time, combined with the evaluation results, construct an expert feature vector, including but not limited to professional fields, number of successful cases, customer satisfaction scores, etc.

[0038] Regarding obtaining the target needs of users: First, through online forms, telephone interviews or face-to-face meetings, we can deeply understand the specific situation and problems faced by the enterprise, including industry background, development stage, main challenges, and problems expected to be solved. Secondly, we analyze the activity records of the enterprise on the platform, such as search keywords, browsed pages, experts consulted, etc., to supplement the enterprise demand information. Finally, based on the collected enterprise information and activity data, we use machine learning algorithms to identify and extract the specific demand patterns of the enterprise, and form the enterprise demand feature information, such as the professional field of the required experts, the urgency of solving the problem, and the budget range.

[0039] Step S220, based on at least two preset data models, feature extraction is performed on target demand and data resources to obtain demand features and resource features respectively; according to the demand features and data resource features, the target demand and data resources are matched to generate multiple data resource recommendation lists.

[0040] It should be emphasized that the various preset data models proposed above are usually data models that are screened and matched through different dimensions, such as keyword dimensions, semantic vector dimensions, situational context dimensions, multimodal data dimensions, etc. This application does not make specific restrictions on the data processing dimensions of the data model.

[0041] In one embodiment of the present application, the preset data model includes at least two of a first data model, a second data model, and a third data model, wherein the first data model is used to extract technical hot words from target requirements and data resources, and generate a hot word data resource recommendation list corresponding to the target requirements by matching the technical hot words of the data resources with the technical hot words of the target requirements; the second data model is used to determine the target technology field to which the target requirements belong according to preset field division rules, and generate a domain data resource recommendation list matching the target requirements based on the influence list of the target technology field; the third data model is used to convert the description information of the target requirements and data resources into semantic vectors, and generate a semantic data resource recommendation list matching the target requirements by calculating the similarity between the semantic vector of the target requirements and the semantic vector of the data resources.

[0042] In one embodiment of the present application, a hot word data resource recommendation list corresponding to the target demand is generated, including: extracting keywords of the target demand and the data resource respectively to obtain the target technology hot words and the data resource technology hot words; calculating the hot word similarity between the target technology hot words and the data resource technology hot words; determining the data resources whose hot word similarity is greater than a preset hot word similarity threshold as the first candidate data resources, and generating a hot word data resource recommendation list based on the hot word similarity sorting of the first candidate data resources.

[0043] Figure 3 It is a schematic diagram of the first data model structure shown in an exemplary embodiment of the present application.

[0044] like Figure 3 As shown, take recommending relevant experts to enterprises as an example. The working scenario of the first data model is divided into two stages: offline and online. When in the offline stage, the model needs to complete the keyword extraction of the corpus data to form technical hot word mining. The corpus data includes: papers, patents, books, etc. Keyword extraction models include but are not limited to: TextRank, TopicRank, SalienceRank. Technical hot words: represents the highest frequency technical keywords extracted from corpus data in various fields. When in the online stage, the technical hot words extracted in the offline stage are used to extract technical hot words from the enterprise description information and expert description information, thereby generating the relationship between the enterprise and the expert, as well as the hot word information corresponding to each expert, so as to facilitate the subsequent recall of related experts based on technical hot words as the enterprise's expert recommendation list.

[0045] like Figure 3As shown in the figure, in the application scenario of recommending relevant experts for enterprises, the operation of the first data model can be divided into two main stages: offline processing and online service. In the offline processing stage, the main task of the model is to extract key technical hot words from a large amount of corpus data. These corpus data cover a wide range of sources, such as academic papers, patent documents and professional books. In order to achieve efficient and accurate keyword extraction, a variety of algorithm models can be used, such as TextRank, TopicRank and SalienceRank. Technical hot words refer to words with high frequency and important technical significance extracted from various professional literature. After entering the online service stage, the model uses the technical hot words obtained in the offline stage to conduct in-depth analysis of specific enterprise needs and expert background information. By matching the technical hot words in the enterprise and expert information, the model can establish a connection between the two and generate a detailed technical hot word file for each expert. This process helps to accurately identify experts who are highly matched with enterprise needs, and finally form a high-quality expert recommendation list to support enterprises in finding suitable partners or consulting objects in specific technical fields. This process not only improves the efficiency and accuracy of expert recommendation, but also promotes the effective circulation and application of knowledge and technical resources.

[0046] In a specific implementation of this application, take providing consulting services in professional fields to enterprises as an example. In order to improve the accuracy of recommendations, the platform decided to adopt a method based on hot word matching to optimize the expert recommendation system. When corporate customers submit consulting requests on the platform, they will fill out a detailed description of the needs, including the required expert fields, specific problems, etc. The platform first uses natural language processing technology (such as TF-IDF, word embedding, etc.) to analyze this demand description and extract key technical vocabulary, which are the target technical hot words. At the same time, the platform will also perform similar processing on the personal information of registered experts, including the expert's profile, professional field, description of successful cases, etc., and extract data resource technology hot words from them.

[0047] Next, the platform uses the cosine similarity algorithm to calculate the similarity between the target technology hot words and the data resource technology hot words. For example, if an enterprise's demand description mentions "the application of artificial intelligence in the field of medical health", then the target technology hot words may include "artificial intelligence", "medical health", and "application". If an expert's profile mentions "focusing on the research and development of AI technology, especially intelligent solutions in the medical industry", the data resource technology hot words may be "AI technology", "medical industry", and "intelligent solutions". By calculating the similarity between these hot words, the degree of match between enterprise needs and expert expertise can be quantified.

[0048] In addition, to ensure the quality of recommendations, the platform sets a preset hot word similarity threshold, such as 0.7. Only those experts whose hot word similarity exceeds this threshold will be selected as the first candidate data resources. For example, in the above example, if the calculation shows that the hot word similarity between the expert and the company's needs is 0.85, which is higher than the preset threshold, then the expert will be included in the first candidate list. For all qualified first candidate experts, the platform will sort them according to their hot word similarity with the company's needs. The higher the hot word similarity, the higher the ranking. Finally, the platform generates a recommended list of hot word data resources based on this ranking and displays it to corporate customers to help them quickly find the experts who best meet their needs, thereby improving consulting efficiency and service quality.

[0049] In one embodiment of the present application, a recommendation list of domain data resources that match the target demand is generated, including: based on a preset technical field division rule, dividing the data resources into multiple different technical fields, and determining the target technical field to which the target demand belongs; identifying the influence of each data resource in the technical field to which it belongs, and generating an influence list; selecting a preset number of data resources from the influence list in the target technical field as candidate data resources, and generating a recommendation list of data resources in the technical field based on the influence ranking of the candidate data resources.

[0050] Figure 4 It is a schematic diagram of the second data model structure shown in an exemplary embodiment of the present application.

[0051] like Figure 4 As shown in the figure, in the context of recommending relevant experts to enterprises, the workflow of the second data model can be divided into two stages: offline processing and online service. In the offline processing stage, the goal of the model is to mine popular technical topics and figures from different national standards (GB) fields. For each GB field i, the model will evaluate the influence of all experts in the field and build a hot list of experts in GB field i based on this. The influence here can be comprehensively evaluated by a variety of indicators, such as the number and quality of published papers, patent holdings, industry recognition, etc. When entering the online service stage, the system will use the hot list built offline to quickly respond to the needs of enterprises. Specifically, when an enterprise j proposes the need to find a suitable expert, the GB field i to which the enterprise belongs is first determined. Then, the system will select the experts who best meet the needs of the enterprise from the pre-built hot list of experts in GB field i as recommendation candidates. This method not only speeds up the speed of expert recommendation, but also ensures the relevance and quality of the recommendation results, thereby effectively helping enterprises find the most suitable partners.

[0052] It is understandable that through such a two-stage design, the second data model can significantly improve processing efficiency while ensuring recommendation accuracy, thus meeting the needs of enterprises in a rapidly changing technological environment.

[0053] In a specific embodiment of the present application, we take the example of providing consulting services in professional fields to enterprises. There are many experts from different industries on the platform, each of whom has a unique professional background and rich practical experience. In order to improve the accuracy of recommendations, the platform decided to use a method based on technical field matching to optimize the expert recommendation system. The following is the specific implementation process:

[0054] First, the platform defines a series of technical fields, such as "artificial intelligence", "big data", "cloud computing", "Internet of Things", "financial technology", etc. Each expert needs to choose the technical field in which he or she is good at when registering, and provide corresponding detailed information, such as professional background, success cases, published articles, etc. When corporate customers submit consultation requests on the platform, they will fill out a detailed description of the needs, including the required expert field, specific problems, etc. The platform analyzes the demand description through natural language processing technology (such as TF-IDF, word embedding, etc.), extracts key technical vocabulary, and determines the technical field to which the target demand belongs according to the preset technical field division rules. For example, if the company's demand description mentions "how to use big data technology to optimize supply chain management", the target technology field is "big data".

[0055] Next, the platform evaluates the influence of each expert in his or her technical field in a variety of ways, including professional background (education, work experience, industry certification, etc.), success cases (number of completed projects, project size and complexity, customer satisfaction evaluation, etc.), academic contributions (number of published papers, number of citations, academic conferences attended, etc.) and industry recognition (awards, honorary titles, industry reputation, etc.). Based on the above evaluation indicators, the platform generates a comprehensive influence score for each expert and generates an influence list in each technical field, with the experts in the list sorted from high to low according to their influence scores.

[0056] Assume that the number of preset items set by the platform is 10, that is, the top 10 experts with the highest influence scores in each technical field are selected as candidate data resources. For example, in the field of "big data", the platform will select the top 10 experts with the highest influence scores as candidate experts. Specifically, the comprehensive influence score of expert A is 9.5, expert B is 9.0, expert C is 8.8, expert D is 8.5, expert E is 8.3, expert F is 8.1, expert G is 7.9, expert H is 7.8, expert I is 7.6, and expert J is 7.5. These 10 experts will be candidate experts.

[0057] Finally, the platform will select the top 10 candidate experts from the target technology field (such as "big data") and sort them from high to low according to their influence scores to generate a final recommendation list of data resources in the technology field. This recommendation list will be displayed to corporate customers to help them quickly find the experts who best meet their needs. For example, the recommendation list is as follows: Expert A, Expert B, Expert C, Expert D, Expert E, Expert F, Expert G, Expert H, Expert I, Expert J.

[0058] Understandably, through this approach, the platform can more accurately recommend experts to enterprises that are highly matched with their needs, thereby improving service quality and user satisfaction.

[0059] In one embodiment of the present application, a recommendation list of semantic data resources matching a target requirement is generated, including: obtaining description information of the target requirement and the data resource, and performing data preprocessing on the description information, wherein the data preprocessing includes at least text cleaning, word segmentation, and removal of stop words; converting the description information of the target requirement and the data resource into a target semantic vector and a data resource semantic vector, respectively; calculating the vector similarity between the target semantic vector and the data resource semantic vector; determining a data resource whose vector similarity is greater than a preset vector similarity threshold as a third candidate data resource, and generating a recommendation list of semantic data resources based on the vector similarity sorting of the third candidate data resource.

[0060] Figure 5 It is a schematic diagram of the third data model structure shown in an exemplary embodiment of the present application.

[0061] like Figure 5 As shown in the figure, in the process of recommending relevant experts to enterprises, the workflow of the second data model is mainly divided into two stages: offline processing and online service. In the offline processing stage, the system uses the semantic matching model to semantically vectorize the description information of each expert and stores the generated vectors in a special vectorization library. The semantic vectorization models used here include but are not limited to advanced algorithms such as SimCSE and ESimCSE, which can effectively capture the deep semantic features of the text and improve the accuracy of the vector representation.

[0062] In the online service stage, when an enterprise makes a request for expert recommendation, the model will immediately convert the enterprise's description information into a semantic vector. Then, by calculating the similarity between the vector and the expert vector stored in the vector library, a group of experts who are highly matched with the enterprise's needs are quickly located. Finally, the system organizes these highly matched experts into a recommendation list and provides it to the enterprise as a reference. This search method based on semantic similarity not only greatly improves the accuracy of the recommendation, but also ensures the efficiency of the recommendation process, providing strong support for enterprises to quickly find suitable partners.

[0063] In a specific embodiment of the present application, taking the example of providing consulting services in a professional field for enterprises. There are multiple experts from different industries on the platform, and each expert has a unique professional background and rich practical experience. To improve the accuracy of recommendations, the platform decides to adopt a semantic matching-based method to optimize the expert recommendation system. The following is the specific implementation process:

[0064] First, when enterprise customers submit consulting requests on the platform, they will fill out a detailed requirement description, including the required expert fields and specific questions. For example, the requirement description submitted by an enterprise may be: "We are a supply chain management company and hope to use big data technology to optimize our inventory management and logistics distribution systems." At the same time, each expert on the platform needs to provide detailed personal information when registering, including professional background, successful cases, published articles, etc. For example, the personal information of an expert may be: "Focusing on the research and development of big data technology, especially having rich experience in supply chain management and logistics optimization."

[0065] Next, the platform performs data preprocessing on these description information. The specific steps include text cleaning, removing non-text content such as special characters, numbers, punctuation marks, etc. in the description information, and retaining the pure text part; word segmentation, splitting the text into words or phrases, for example, using a Chinese word segmentation tool (such as jieba) to split the sentence into words; removing stop words, removing common meaningless words, such as "de", "shi", "zai", etc., to reduce noise.

[0066] Then, the platform converts the description information of the target requirement and data resources into target semantic vectors and data resource semantic vectors respectively. The specific method is to use a pre-trained word vector model (such as Word2Vec, GloVe or BERT) to convert each word into a vector representation, and then aggregate all the word vectors in the description information to generate the semantic vector of the entire description. Common aggregation methods include taking the average, weighted average or using an attention mechanism.

[0067] Next, the platform calculates the vector similarity between the target semantic vector and the data resource semantic vector. Using cosine similarity, Euclidean distance or other similarity calculation methods, calculate the similarity between the semantic vector of the target requirement and the semantic vector of each expert.

[0068] In addition, to ensure the quality of the recommendation, the platform also sets a preset vector similarity threshold, such as 0.7. The vector similarities of all experts and enterprise requirements are compared, and experts with similarities greater than the preset threshold are selected as the third candidate data resources. For example, suppose the demand description submitted by the enterprise is: "We are a supply chain management company and hope to use big data technology to optimize our inventory management and logistics distribution system." The description information of the target demand after preprocessing is: supply chain management, big data technology, inventory management, logistics distribution system. The description information of the three experts is: Expert A (big data technology, supply chain management, logistics optimization, project experience), Expert B (artificial intelligence, machine learning, data analysis, project management), Expert C (big data technology, supply chain optimization, logistics management, successful cases). After calculating the similarity, the results are: Expert A's similarity is 0.85, Expert B's similarity is 0.50, and Expert C's similarity is 0.80. Since the preset similarity threshold is 0.7, Expert A and Expert C are selected as the third candidate data resources.

[0069] Finally, the platform sorts the third candidate experts from high to low according to vector similarity and generates the final semantic data resource recommendation list. In this example, the generated recommendation list is as follows: Expert A, Expert C, Expert E, Expert B.

[0070] It should be noted that the above-mentioned technical buzzwords and technical fields are all implemented based on conventional division methods, such as keyword extraction, topic modeling, machine learning classifiers, knowledge graph construction, etc. This application is implemented based on pre-divided technical fields and pre-defined technical buzzwords. This application does not impose any specific restrictions on the division method of its technical fields, the specific division method and content of technical buzzwords.

[0071] Step S230 , merging multiple data resource recommendation lists generated by different data models to obtain a comprehensive recommendation list, so as to recommend data resources to the user based on the comprehensive recommendation list.

[0072] In one embodiment of the present application, multiple data resource recommendation lists generated by different data models are merged to obtain a comprehensive recommendation list, including: obtaining the total number of candidate data resource items in the comprehensive recommendation list; selecting the same number of candidate data resources from each data resource recommendation list, or selecting candidate data resources from each data resource recommendation list based on a preset model weight distribution; deduplicating the candidate data resources obtained after the selection to generate a comprehensive recommendation list based on the deduplicated candidate data resources, wherein the sum of the number of data resource items selected in each data resource recommendation list is equal to the total number of candidate data resource items.

[0073] In a specific embodiment of the present application, it is assumed that there is an enterprise service platform that aims to provide consulting services in professional fields for enterprises. There are many experts from different industries on the platform, each of whom has a unique professional background and rich practical experience. In order to improve the accuracy of the recommendation, the platform decides to use three different data models: the first data model, the second data model, and the third data model to generate the expert recommendation list. The following is the specific implementation process:

[0074] First, use different data models to generate multiple data resource recommendation lists. Among them, based on the first data model, extract the key hot words in the enterprise needs and technical fields, calculate the hot word similarity, and generate a hot word data resource recommendation list as follows: expert A, expert B, expert C, expert D, expert E. Through the second data model, based on the preset technical field division rules, evaluate the influence of experts in specific fields, and generate an influence list as follows: expert B, expert C, expert F, expert G, expert H. Using the third data model, perform data preprocessing on enterprise needs and expert description information, generate semantic vectors, calculate vector similarity, and generate a semantic data resource recommendation list as follows: expert A, expert C, expert D, expert I, expert J.

[0075] Case 1: Assume that the total number of candidate data resources in the comprehensive recommendation list set by the platform is 10, and the weights assigned by the platform to the three models are: the first data model (0.4), the second data model (0.3), and the third data model (0.3). The number of data resource items selected by each model is calculated based on the weights, which are 4 items for the first data model, 3 items for the second data model, and 3 items for the third data model. Therefore, the first 4 items are selected from the hot word recommendation list, expert A, expert B, expert C, and expert D. The first 3 items are selected from the hot recommendation list, expert B, expert C, and expert F. The first 3 items are selected from the semantic recommendation list, expert A, expert C, and expert D. Then, the candidate data resources obtained after selection are deduplicated, and the candidate data resources after deduplication are obtained as follows: expert A, expert B, expert C, expert D, and expert F. Since there are only 5 candidate data resources after deduplication, which is less than the set 10 items, the remaining 5 items need to be supplemented from the remaining recommendation list. Expert E is added from the hot word recommendation list, experts G and H are added from the popular recommendation list, and experts I and J are added from the semantic recommendation list. The final comprehensive recommendation list is: expert A, expert B, expert C, expert D, expert F, expert E, expert G, expert H, expert I, and expert J.

[0076] Case 2: Assuming that the total number of candidate data resources in the comprehensive recommendation list set by the platform is 12, in order to ensure that the recommendation results of each model can be fully reflected in the comprehensive recommendation list, the platform decides to select the same number of candidate data resources from the recommendation list generated by each data model. Then select the top 4 experts A, B, C, and D from the hot word recommendation list. Select the top 4 experts B, C, F, and G from the popular recommendation list. Select the top 4 experts A, C, D, and I from the semantic recommendation list. Then, the platform deduplicates the candidate data resources obtained after selection, and the deduplicated candidate data resources are experts A, B, C, D, F, G, and I. Finally, the platform generates a comprehensive recommendation list. Since there are only 7 candidate data resources after deduplication, which is less than the set 12 items, the remaining 5 items need to be supplemented from the remaining recommendation lists, namely expert E from the hot word recommendation list, expert H from the popular recommendation list, and expert J, expert K, and expert L from the semantic recommendation list. The final comprehensive recommendation list is: expert A, expert B, expert C, expert D, expert F, expert G, expert I, expert E, expert H, expert J, expert K, and expert L.

[0077] It can be understood that, based on the method proposed in this embodiment, the platform can comprehensively consider the advantages of multiple data models and generate a more comprehensive and accurate expert recommendation list, thereby improving service quality and user satisfaction.

[0078] In another embodiment of the present application, multiple data resource recommendation lists generated by different data models are merged to obtain a comprehensive recommendation list, which also includes: combining candidate data resources from different data resource recommendation lists; or, taking the number of candidate data resources in any data resource recommendation list as a benchmark, adjusting the number of candidate data resources generated by other data models according to preset model weights, selecting a corresponding number of candidate data resources from each data resource recommendation list and combining them; deduplicating the candidate data resources obtained after the combination to generate a comprehensive recommendation list based on the deduplicated candidate data resources.

[0079] In another specific embodiment of the present application, based on the aforementioned embodiments, the recommended list of hot word data resources is: expert A, expert B, expert C, expert D, expert E; the influence list is: expert B, expert C, expert F, expert G, expert H; the recommended list of semantic data resources is: expert A, expert C, expert D, expert I, and expert J.

[0080] Case 1: Directly combine and deduplicate the candidate data resources in the three recommendation lists, and obtain the final recommendation list: expert A, expert B, expert C, expert D, expert E, expert F, expert G, expert H, expert I, expert J.

[0081] Case 2: In order to more accurately control the number of recommendation lists, the platform uses the number of candidate data resources in any data resource recommendation list as a benchmark and adjusts the number of candidate data resources generated by other data models according to the preset model weights. The specific steps are as follows:

[0082] Assume that the 5 items in the recommendation list generated by the first data model are used as the benchmark. The platform assigns weights to the three models respectively: first data model (0.4), second data model (0.3), and third data model (0.3). First data model: 5 items; second data model: 5*0.3 / 0.4=3.75≈4 items; third data model: 5*0.3 / 0.4=3.75≈4 items. Select the top 5 items from the hot word recommendation list: expert A, expert B, expert C, expert D, expert E; select the top 4 items from the hot recommendation list: expert B, expert C, expert F, expert G; select the top 4 items from the semantic recommendation list: expert A, expert C, expert D, expert I. Then, the platform deduplicates the candidate data resources obtained after the combination, and the deduplicated candidate data resources are: expert A, expert B, expert C, expert D, expert E, expert F, expert G, and expert I.

[0083] In summary, in the data resource recommendation matching method proposed in this application, when merging multiple data resource recommendation lists generated by different data models, either weighted merging or unweighted merging can be selected. In addition, the number of final recommendation lists can be pre-limited or unrestricted. The specific merging processing method can be flexibly combined and used according to actual needs.

[0084] It should be noted that the above embodiment is an exemplary description of the method of jointly processing the first data model, the second data model, and the third data model. In actual application, considering specific needs, it can be a combination of the first data model and the second data model, or a combination of the first data model and the third data model, or a combination of the second data model and the third data model, and this application does not impose any specific restrictions on it.

[0085] In addition, it should be emphasized that the data resource recommendation and matching method proposed in this application is not only used to recommend relevant experts to enterprises, but can also be applied to many other related scenarios. The following are some specific implementation scenario examples:

[0086] Course recommendation scenarios on online education platforms: There are a large number of course resources on online education platforms, covering various disciplines and skill areas. Take the example of the platform's need to recommend the most suitable courses to users based on their learning goals and interests. The first data model extracts key hot words in the learning goals submitted by users, such as "Programming Basics", "Python", "Data Structure", etc., calculates the hot word similarity, and generates a hot word recommendation list. The second data model generates a list of popular course recommendations based on the popularity of the course and user reviews. The third data model performs data preprocessing on the user's learning goals and course descriptions, generates semantic vectors, calculates vector similarity, and generates a semantic recommendation list. Finally, the above three recommendation lists are weighted and merged to generate the final comprehensive course recommendation list, which is displayed to the user.

[0087] Product recommendation scenario on e-commerce platforms: There are a large number of products on e-commerce platforms, covering various categories. Take the example of the platform needing to recommend the most suitable products to users based on their shopping history and interest preferences. The first data model extracts key hot words in user search keywords and shopping history, such as "smart watches", "sports shoes", "laptops", etc., calculates the hot word similarity, and generates a hot word recommendation list. The second data model generates a hot product recommendation list based on product sales data and user reviews. The third data model performs data preprocessing on the user's shopping history and product description, generates a semantic vector, calculates vector similarity, and generates a semantic recommendation list. Finally, the above three recommendation lists are weighted and merged to generate the final comprehensive product recommendation list, which is displayed to the user.

[0088] Article recommendation scenario of news information platform: There are a large number of articles on the news information platform, covering various types of news and special reports. Take the example of the platform's need to recommend the most suitable news articles to users based on their reading habits and interest preferences. The first data model extracts key hot words in the user's reading history, such as "international news", "technological frontiers", "financial analysis", etc., calculates the similarity of hot words, and generates a hot word recommendation list. The second data model generates a recommendation list of popular articles based on the click-through rate and user feedback of the article. The third data model performs data preprocessing on the user's reading history and article content, generates semantic vectors, calculates vector similarity, and generates a semantic recommendation list. Finally, the above three recommendation lists are weighted and merged to generate the final comprehensive article recommendation list, which is displayed to the user.

[0089] Through these specific implementation scenarios, it can be seen that the data resource recommendation and matching method proposed in this application has broad application prospects and flexibility, and can effectively improve the accuracy of recommendations and user satisfaction.

[0090] Finally, it should be emphasized that the data resource recommendation and matching method proposed in this application uses a variety of data models to conduct multi-dimensional analysis and matching of data resources from different sources, which significantly improves the accuracy and flexibility of recommendations. The following are the main benefits of this method:

[0091] Improve the accuracy of recommendations: The data resource recommendation and matching method proposed in this application analyzes and matches data resources from multiple perspectives through the comprehensive use of the first data model, the second data model, and the third data model to ensure the comprehensiveness and accuracy of the recommendation results. In addition, the third data model is used to deeply understand the semantic information of user needs and data resources, avoiding the limitations of relying solely on keyword matching and improving the relevance and accuracy of recommendations.

[0092] Enhance the flexibility and adaptability of the system: The data resource recommendation matching method proposed in this application supports weighted merging and unweighted merging, and can flexibly select the merging method according to the actual application scenario and needs, thereby improving the adaptability of the system. In addition, when finally generating the recommendation list, the number of the final recommendation list can be limited or not, which meets the needs of different users and scenarios and provides more flexibility.

[0093] Improve user experience: The data resource recommendation and matching method proposed in this application can more accurately capture the personalized needs of users through multi-dimensional analysis and matching, provide recommendation results that are more in line with user expectations, and improve user satisfaction. It also combines the advantages of different data models to ensure the diversity and richness of recommendation results, and avoid the over-concentration or singleness of recommendation results that may be caused by a single model.

[0094] In summary, the data resource recommendation and matching method proposed in this application significantly improves the accuracy of recommendations and user experience through multi-model fusion, semantic understanding and flexible merging processing. At the same time, it has wide applicability and scalability, providing effective technical support for various application scenarios.

[0095] Figure 6 is a block diagram of a device shown in an exemplary embodiment of the present application. The device can be applied to Figure 1 The device may also be applicable to other exemplary implementation environments and specifically configured in other devices, and this embodiment does not limit the implementation environment to which the device is applicable.

[0096] like Figure 6 As shown, the exemplary data resource recommendation and matching device includes: a data collection module 610 , a recommendation list generation module 620 , and a recommendation information integration module 630 .

[0097] Among them, the data collection module 610 obtains data resources and user's target needs; the recommendation list generation module 620 is used to extract features of the target needs and data resources based on at least two preset data models to obtain demand features and resource features; according to the demand features and data resource features, the target needs and data resources are matched to generate multiple data resource recommendation lists; the recommendation information integration module 630 is used to merge multiple data resource recommendation lists generated by different data models to obtain a comprehensive recommendation list, so as to recommend data resources to users based on the comprehensive recommendation list.

[0098] It should be noted that the data resource recommendation and matching device provided in the above embodiment and the data resource recommendation and matching 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 actual applications, the data resource recommendation and matching device provided in the above embodiment can distribute the above functions to different functional modules as needed, that is, divide the internal structure of the device into different functional modules to complete all or part of the functions described above, and this is not limited here.

[0099] An embodiment of the present application also provides an electronic device, comprising: 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 data resource recommendation and matching method provided in the above-mentioned embodiments.

[0100] Figure 7 The structure diagram of the computer system suitable for implementing the electronic device of the embodiment of the present application is shown. It should be noted that: Figure 7 The computer system 700 of the electronic device 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.

[0101] like Figure 7As shown, the computer system 700 includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 702 or the program loaded from the storage part 708 to the random access memory (RAM) 703, such as executing the method described in the above embodiment. In the RAM 703, various programs and data required for system operation are also stored. The CPU 701, ROM 702 and RAM 703 are connected to each other through a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0102] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, a mouse, etc.; an output section 707 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 710 as needed so that a computer program read therefrom is installed into the storage section 708 as needed.

[0103] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication section 709, and / or installed from a removable medium 711. When the computer program is executed by a central processing unit (CPU) 701, various functions defined in the system of the present application are executed.

[0104] It should be noted that the computer-readable medium shown in the embodiment of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a computer-readable computer program is carried. This propagated data signal can take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. A computer program contained on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0105] The flowchart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operations of the system, method and computer program product according to various embodiments of the present application. Wherein, each box in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0106] The units involved in the embodiments described in this application may be implemented by software or hardware, and the units described may also be set in a processor. The names of these units do not, in some cases, constitute limitations on the units themselves.

[0107] Another aspect of the present application 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 data resource recommendation and matching 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.

[0108] Another aspect of the present application also provides a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the data resource recommendation and matching method provided in each of the above embodiments.

[0109] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present invention. 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 by the present invention shall still be covered by the claims of the present invention.

Claims

1. A data resource recommendation and matching method, characterized in that: The method comprises: Obtain data resources and user target needs; Based on at least two preset data models, feature extraction is performed on target demand and data resources to obtain demand features and resource features; According to the demand characteristics and data resource characteristics, target demand and data resources are matched to generate multiple data resource recommendation lists; Multiple data resource recommendation lists generated by different data models are merged to obtain a comprehensive recommendation list, so as to recommend data resources to users based on the comprehensive recommendation list.

2. The data resource recommendation and matching method according to claim 1, characterized in that: The preset data model includes at least two of a first data model, a second data model, and a third data model, wherein: The first data model is used to extract technical hot words from target requirements and data resources, and generate a hot word data resource recommendation list corresponding to the target requirements by matching the technical hot words of the data resources with the technical hot words of the target requirements; The second data model is used to determine the target technology field to which the target demand belongs according to the preset field division rules, and generate a recommended list of domain data resources matching the target demand based on the influence list of the target technology field; The third data model is used to convert the description information of the target demand and the data resource into a semantic vector, and generate a recommendation list of semantic data resources matching the target demand by calculating the similarity between the semantic vector of the target demand and the semantic vector of the data resource.

3. The data resource recommendation and matching method according to claim 2, characterized in that: Generate a recommended list of hot word data resources corresponding to the target needs, including: Extracting keywords of the target demand and the data resource respectively to obtain target technology hot words and data resource technology hot words; Calculating the hot word similarity between the target technology hot word and the data resource technology hot word; The data resources whose hot word similarity is greater than a preset hot word similarity threshold are determined as first candidate data resources, and a hot word data resource recommendation list is generated based on the hot word similarity ranking of the first candidate data resources.

4. The data resource recommendation and matching method according to claim 2, characterized in that: Generate a recommended list of domain data resources that match the target requirements, including: Based on the preset technical field division rules, data resources are divided into multiple different technical fields, and the target technical field to which the target demand belongs is determined; Identify the influence of each data resource in the technical field to which it belongs and generate an influence list; A preset number of data resources are selected from the influence list of the target technical field as candidate data resources, and a recommendation list of data resources in the technical field is generated based on the influence ranking of the candidate data resources.

5. The data resource recommendation and matching method according to claim 2, characterized in that: Generate a list of semantic data resource recommendations that match the target requirements, including: Obtaining description information of the target demand and the data resource, and performing data preprocessing on the description information, wherein the data preprocessing at least includes text cleaning, word segmentation, and stop word removal; Convert the target requirements and data resource description information into a target semantic vector and a data resource semantic vector respectively; Calculating the vector similarity between the target semantic vector and the data resource semantic vector; The data resources whose vector similarity is greater than a preset vector similarity threshold are determined as third candidate data resources, and a semantic data resource recommendation list is generated based on the vector similarity sorting of the third candidate data resources.

6. The data resource recommendation and matching method according to any one of claims 1 to 5, characterized in that: Multiple data resource recommendation lists generated by different data models are merged to obtain a comprehensive recommendation list, including: Get the total number of candidate data resources in the comprehensive recommendation list; Selecting candidate data resources of the same number of items from each data resource recommendation list, or selecting candidate data resources from each data resource recommendation list based on a preset model weight; The candidate data resources obtained after the selection are deduplicated to generate a comprehensive recommendation list based on the deduplicated candidate data resources, and the sum of the number of selected data resource items in each data resource recommendation list is equal to the total number of candidate data resource items.

7. The data resource recommendation and matching method according to any one of claims 1 to 5, characterized in that: The multiple data resource recommendation lists generated by different data models are merged to obtain a comprehensive recommendation list, which also includes: Combining candidate data resources from different data resource recommendation lists; or, taking the number of candidate data resources in any data resource recommendation list as a benchmark, adjusting the number of candidate data resources generated by other data models according to preset model weights, selecting a corresponding number of candidate data resources from each data resource recommendation list and combining them; The candidate data resources obtained after the combination are deduplicated to generate a comprehensive recommendation list based on the deduplicated candidate data resources.

8. A data resource recommendation and matching device, characterized in that: The device comprises: Data collection module, used to obtain target requirements and data resources; A recommendation list generation module is used to extract the characteristics of target requirements and data resources based on at least two preset data models to obtain requirement characteristics and resource characteristics; according to the requirement characteristics and data resource characteristics, the target requirements and data resources are matched to generate multiple data resource recommendation lists; The recommendation information integration module is used to merge multiple data resource recommendation lists generated by different data models to obtain a comprehensive recommendation list, so as to recommend data resources to data resource demanders based on the comprehensive recommendation list.

9. An electronic device, characterized in that: It comprises a processor, a memory and a communication bus; the communication bus is used to connect the processor and the memory; the processor is used to execute the computer program stored in the memory to implement the data resource recommendation and matching method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and the computer program is used to enable a computer to execute the data resource recommendation and matching method as described in any one of claims 1-7.

Images