Expert information management method based on double-algorithm cooperation and related device

Abstract

The application discloses an expert information management method based on double-algorithm cooperation and related equipment, which can be applied to the field of artificial intelligence technology. The application automatically audits the to-be-processed expert information through a semantic contradiction detection algorithm to improve the accuracy of the expert information audit result, and automatically completes the to-be-processed expert information through a structured data alignment completion algorithm to improve the accuracy of the expert information completion result. Then, the expert information audit result and the expert information completion result generate a structured audit and completion report, so that related personnel can quickly determine the defects of the expert information in time, thereby improving the efficiency and accuracy of the related personnel in processing the related defects of the expert information, further improving the accuracy of the expert information audit result and the completion result, and improving the authenticity, logicality and integrity of the expert information.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to an expert information management method and related equipment based on dual-algorithm collaboration. Background Technology

[0002] In existing expert information relationship systems, the information submitted by experts carries risks such as logical inconsistencies and false statements, including conflicting educational backgrounds and inaccurate achievement information. These systems primarily rely on manual review, resulting in low efficiency and inaccurate results. Furthermore, existing expert profiles often contain missing fields, such as incomplete records of academic achievements like papers, patents, and projects. These systems rely heavily on manual completion, leading to low efficiency and accuracy. While existing technologies can use keyword matching or regular expression-based validation to review or complete expert information, these methods remain inaccurate when dealing with multi-source, heterogeneous expert information, failing to simultaneously guarantee the authenticity, logical coherence, and completeness of the expert information.

[0003] In summary, the technical problems existing in the relevant technologies need to be improved. Summary of the Invention

[0004] The main objective of this application is to propose an expert information management method and related equipment based on dual-algorithm collaboration, which can effectively improve the accuracy of expert information review results and completion results, thereby enhancing the authenticity, logic, and completeness of expert information.

[0005] To achieve the above objectives, one aspect of this application proposes an expert information management method based on dual-algorithm collaboration, the method comprising the following steps: Obtain information on experts awaiting processing; The expert information to be processed is automatically reviewed using a semantic contradiction detection algorithm to obtain the expert information review result; The expert information to be processed is automatically completed using a structured data alignment and completion algorithm to obtain the expert information completion result; A structured review and completion report is generated based on the expert information review results and the expert information completion results.

[0006] In some embodiments, the automatic review of the expert information to be processed using a semantic contradiction detection algorithm includes: Construct a large-scale model for text logic judgment; The text logic judgment model is used to assess the logical completeness and compliance of the expert information to be processed. Construct a multidimensional semantic similarity comparison model; The authenticity of the expert information to be processed is verified by the multidimensional semantic similarity comparison model.

[0007] In some embodiments, constructing a large model for text logic judgment includes: Use an open-source large language model as the base model; Training data is obtained from a preset corpus, which includes academic paper information, patent specification information, industry standard information, or professional title evaluation document information. The base model is trained using the training data to learn word embedding representations. Construct logical constraint relationships, including temporal logical constraint relationships, hierarchical dependency constraint relationships, or compliance constraint relationships; The trained base model learns the logical constraint relationships through supervised fine-tuning, thus obtaining the large-scale text logic judgment model.

[0008] In some embodiments, the step of verifying the authenticity of the expert information to be processed using the multidimensional semantic similarity comparison model includes: Obtain candidate entry information retrieved from the first external database; The expert information to be processed is converted into a first high-dimensional semantic vector; The candidate entry information is converted into a second high-dimensional semantic vector; Calculate the cosine similarity between the first high-dimensional semantic vector and the second high-dimensional semantic vector; The semantic correlation between the expert information to be processed and the candidate item information is quantified based on the cosine similarity metric. The expert information to be processed is matched with the candidate item information by field matching; The authenticity verification result is generated based on the semantic relevance quantification result and the field matching result.

[0009] In some embodiments, the automatic completion of the expert information to be processed using a structured data alignment and completion algorithm to obtain the expert information completion result includes: Use the first keyword in the expert information to be processed as the query term; The initial candidate dataset is obtained by querying data in a second external database using the query terms. A comprehensive score is generated based on a two-stage entity disambiguation and matching mechanism to correspond to the initial candidate dataset. Based on the comprehensive score, several candidate results are selected from the initial candidate dataset, and a candidate result ranking table is generated based on all the candidate results. The information of the experts to be processed is supplemented based on the candidate results ranking table.

[0010] In some embodiments, generating a comprehensive score corresponding to the initial candidate dataset based on a two-segment entity disambiguation and matching mechanism includes: Extract keywords to be processed from the expert information to be processed; The keyword to be processed is compared with the keyword to be matched in the initial candidate dataset; Based on the comparison results, the data in the initial candidate dataset are filtered to obtain the candidate result dataset; The comprehensive score of each result in the candidate result dataset is calculated based on preset key dimensions.

[0011] In some embodiments, the preset key dimensions include the matching degree of institutional affiliation, semantic consistency of expert research fields, and the rationality of career chronology.

[0012] To achieve the above objectives, another aspect of this application proposes an expert information management device based on dual-algorithm collaboration, the device comprising: The acquisition module is used to acquire information about the experts to be processed. The automatic review module is used to automatically review the expert information to be processed using a semantic contradiction detection algorithm to obtain the expert information review result; The automatic completion module is used to automatically complete the expert information to be processed using a structured data alignment and completion algorithm to obtain the expert information completion result; The report generation module is used to generate a structured review and completion report based on the expert information review results and the expert information completion results.

[0013] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0014] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the methods described above.

[0015] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer program product, including a computer program that, when executed by a processor, implements the aforementioned method.

[0016] The embodiments of this application include at least the following beneficial effects: This application provides an expert information management method and related equipment based on dual-algorithm collaboration. This scheme automatically reviews the expert information to be processed through a semantic contradiction detection algorithm to improve the accuracy of the expert information review results. At the same time, it automatically completes the expert information to be processed through a structured data alignment and completion algorithm to improve the accuracy of the expert information completion results. Then, the expert information review results and expert information completion results generate a structured review and completion report, which allows relevant personnel to quickly and promptly identify defects in expert information, thereby improving the efficiency and accuracy of relevant personnel in handling defects in expert information, further improving the accuracy of expert information review results and completion results, and thus improving the authenticity, logic, and completeness of expert information. Attached Figure Description

[0017] Figure 1 This is a flowchart of the expert information management method based on dual-algorithm collaboration provided in the embodiments of this application; Figure 2 This is a schematic diagram of the structure of the expert information management device based on dual-algorithm collaboration provided in the embodiments of this application; Figure 3 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0019] It is understood that the terms “first,” “second,” etc., used in this application may be used herein to describe various concepts, but unless otherwise stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the words “if,” “when,” or “in response to a determination” as used herein may be interpreted as “when…” or “when…” or “in response to a determination.”

[0020] As used in this application, the terms "at least one", "multiple", "each", "any", etc., "at least one" includes one, two or more, "multiple" includes two or more, "each" refers to each of the corresponding multiples, and "any" refers to any one of the multiples.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0022] In existing expert information relationship systems, the information submitted by experts carries risks such as logical inconsistencies and false statements, including conflicting educational backgrounds and inaccurate achievement information. These systems primarily rely on manual review, resulting in low efficiency and inaccurate results. Furthermore, existing expert profiles often contain missing fields, such as incomplete records of academic achievements like papers, patents, and projects. These systems rely heavily on manual completion, leading to low efficiency and accuracy. While existing technologies can use keyword matching or regular expression-based validation to review or complete expert information, these methods remain inaccurate when dealing with multi-source, heterogeneous expert information, failing to simultaneously guarantee the authenticity, logical coherence, and completeness of the expert information.

[0023] In view of this, this application provides an expert information management method and related equipment based on dual-algorithm collaboration, which can effectively improve the accuracy of expert information review results and completion results, thereby improving the authenticity, logic and completeness of expert information.

[0024] The expert information management method based on dual-algorithm collaboration provided in this application relates to the field of artificial intelligence technology. This method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or in-vehicle terminal, but is not limited to these. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application implementing the expert information management method based on dual-algorithm collaboration, but is not limited to the above forms.

[0025] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0026] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.

[0027] The embodiments of this application will be described in detail below with reference to the accompanying drawings: Figure 1 This is an optional flowchart of the expert information management method based on dual-algorithm collaboration provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S110 to S140: Step S110: Obtain information on the experts to be processed; Step S120: Automatically review the expert information to be processed using a semantic contradiction detection algorithm to obtain the expert information review result; Step S130: Automatically complete the expert information to be processed using the structured data alignment and completion algorithm to obtain the expert information completion result; Step S140: Generate a structured review and completion report based on the expert information review results and expert information completion results.

[0028] It is understandable that the expert information to be processed can be expert information entered in real time by relevant personnel on a preset platform interface. This expert information includes, but is not limited to, information such as name, identity information, educational background, professional resume, academic output, and professional qualification certifications.

[0029] It is understood that the semantic contradiction detection algorithm in this embodiment, based on established expert entry qualification standards and information submission specifications, performs multi-dimensional and in-depth automated review and cross-validation of the personalized information submitted by experts, thereby significantly reducing the complexity and workload of manual review and improving the quality and credibility of expert database data assets. Specifically, the data objects processed by the semantic contradiction detection algorithm in this embodiment have multi-source heterogeneity, which may include core information entities and external reference data sources. Core information entities include, but are not limited to, key information such as the expert's educational background, professional resume, academic output, and professional qualification certifications. To ensure the authority and timeliness of verification, the semantic contradiction detection algorithm in this embodiment needs to interface with and integrate multiple authoritative third-party databases. This requires that the external data interface have high availability and that a regular data update and synchronization mechanism be established.

[0030] It is understandable that, in the process of automatically reviewing expert information to be processed using the semantic contradiction detection algorithm in this embodiment, a large-scale text logic judgment model can be constructed to evaluate the logical completeness and compliance of the expert information to be processed; simultaneously, a multi-dimensional semantic similarity comparison model can be constructed to verify the authenticity of the expert information to be processed. Specifically, the large-scale text logic judgment model is used to simulate the logical reasoning ability of human experts and automatically evaluate the internal logical completeness and compliance of expert information according to preset rules. Therefore, the construction process of the large-scale text logic judgment model includes, but is not limited to, the following steps: Use an open-source large language model as the base model; Training data is obtained from a pre-set corpus, which includes academic paper information, patent specification information, industry standard information, or professional title evaluation document information. The base model is trained using word embedding representation learning based on training data; Construct logical constraints, which include temporal logical constraints, hierarchical dependency constraints, or compliance constraints. By using supervised fine-tuning, the trained base model learns the logical constraints, resulting in a large-scale text logic judgment model.

[0031] It is understood that the base model selected in this embodiment can be an architecture capable of processing long texts and complex logical chain reasoning. Therefore, this embodiment selects an open-source large language model with excellent performance, verified by public benchmark tests (such as MMLU and GSM8K) in terms of logical reasoning, text understanding, and instruction compliance, as the technical base model. Specifically, in order for the base model to accurately understand the professional terminology and implicit knowledge in the field of expert information review, this embodiment will, as appropriate, employ efficient parameter fine-tuning techniques, such as low-rank adaptation (LoRA), to continuously train on a massive, high-quality pre-set corpus (covering academic papers, patent specifications, industry standards, professional title evaluation documents, etc.) to optimize the word embedding representation within the base model, enabling the base model to form a deep understanding of the concepts and contexts specific to this field. Meanwhile, the review rules set by relevant personnel are formalized and structured to construct a micro knowledge base to define the logical constraints of entities. Then, the logical constraints are transformed into "instruction-response pairs" in natural language form. The trained base model is then trained to understand and strictly execute these complex logical judgment tasks through supervised fine-tuning, so that the base model learns to output structured judgment criteria, rather than simple "yes / no" conclusions.

[0032] For example, the temporal logic constraints in this embodiment can be that the "doctoral graduation time" must be later than the "master's graduation time"; and that two full-time work experiences cannot exist within the same time period. The hierarchical dependency constraints can be the time of obtaining the "senior engineer" professional title, which usually requires meeting the basic requirement of years of work experience. The compliance constraints can be checking for records of academic misconduct in accordance with relevant scientific research integrity management regulations.

[0033] In this embodiment, after constructing a large text logic judgment model using the above method, the model is used to evaluate the logical completeness and compliance of the expert information to be processed, thereby effectively improving the accuracy of the evaluation results.

[0034] It is understood that the multidimensional semantic similarity comparison model in this embodiment is used to accurately compare the key information in the expert information submitted by experts with records in a third-party authoritative database to verify its authenticity. Specifically, this embodiment obtains candidate entry information retrieved from a first external database, converts the candidate entry information into a second high-dimensional semantic vector, and simultaneously converts the expert information to be processed into a first high-dimensional semantic vector; then, it calculates the cosine similarity between the first and second high-dimensional semantic vectors, and quantifies the semantic association between the expert information to be processed and the candidate entry information based on the cosine similarity metric; after performing field matching between the expert information to be processed and the candidate entry information, it generates an authenticity verification result based on the semantic association quantification result and the field matching result.

[0035] It is understood that this embodiment can employ a pre-trained language model (such as SciBERT) or a general semantic representation model (such as SimCSE) on a scientific literature corpus to encode the expert information (such as paper titles and patent names) submitted by experts and the candidate entry information retrieved from external databases into a first high-dimensional semantic vector and a second high-dimensional semantic vector, respectively. Then, by calculating the cosine similarity between the first and second high-dimensional semantic vectors, the semantic relevance between texts is quantified through the relationship between the cosine similarity and a similarity threshold. In this embodiment, the similarity threshold can be set differently and calibrated according to the information type (such as title, abstract, author list). For example, a very high similarity (e.g., >0.95) is required for paper titles, while the similarity requirement for project descriptions can be appropriately relaxed.

[0036] It is understood that this embodiment, for core fields such as author's name, affiliation, journal name, and patent number, will undergo secondary verification in addition to semantic similarity, combining exact string matching or fuzzy matching algorithms to ensure the accuracy of key information. After completing semantic relevance quantification and field matching, this embodiment generates authenticity verification results based on the semantic relevance quantification results and field matching results. These authenticity verification results include, but are not limited to, results of logical contradiction quantification analysis, results of temporal conflict, discrepancies in qualifications, doubts about the attribution of achievements, and results of factual consistency deviation analysis. These results reveal potential risks and doubts in the expert information to be processed, thereby effectively improving the authenticity and logical consistency of the expert information.

[0037] It is understood that the structured data alignment and completion algorithm in this embodiment actively retrieves and integrates external authoritative academic database resources to accurately and automatically complete and correct fields in the expert database that have missing, outdated, or inconsistent information, thereby improving the completeness, authenticity, and timeliness of expert information and constructing a comprehensive and accurate digital archive of experts. Specifically, when automatically completing the expert information to be processed, this embodiment can use the first keyword in the expert information to be processed as the query term; after obtaining the initial candidate dataset by querying data in a second external database using the query term, a comprehensive score corresponding to the initial candidate dataset is generated based on a two-segment entity disambiguation and matching mechanism; then, several candidate results are selected from the initial candidate dataset based on the comprehensive score, and a candidate result ranking table is generated based on all the candidate results; finally, the information completion processing of the expert information to be processed is performed based on the candidate result ranking table.

[0038] Specifically, the first keyword in this embodiment includes, but is not limited to, the personal identity information of the expert corresponding to the expert information to be processed, such as the expert's name, current or former institution, etc. Using the first keyword as the query term, keyword retrieval is performed by calling the API interface of a second external database (external academic database) to obtain structured or semi-structured data to form an initial candidate dataset. The data in the initial candidate dataset includes, but is not limited to, data such as the title of the paper / patent, author list (including affiliation), publication or publication date, and source (journal, conference, or patent office).

[0039] Understandably, to achieve high-precision structured data alignment and completion, this embodiment employs a hierarchical "candidate entity generation and credibility alignment framework" to automatically complete the expert information to be processed. Specifically, this embodiment uses the expert's "name" and "institution" as core query terms, and simultaneously calls multiple pre-set third-party academic database APIs to collect relevant academic achievements (papers, patents, etc.) data to form an initial candidate dataset. Then, a comprehensive score corresponding to the initial candidate dataset is generated based on two entity disambiguation and matching mechanisms. Specifically, in the first stage, this embodiment extracts keywords to be processed from the expert information to be processed, compares the keywords to be processed with the keywords to be matched in the initial candidate dataset, and filters the data in the initial candidate dataset according to the comparison results to obtain a candidate achievement dataset; in the second stage, a comprehensive score is calculated for each achievement data in the candidate achievement dataset based on preset key dimensions.

[0040] Understandably, in the first stage, this embodiment performs coarse-grained filtering and preliminary identity differentiation of the expert information to be processed, in order to quickly eliminate obviously irrelevant noise data and solve the common problem of scholars having the same name. Specifically, this embodiment will combine the fuzzy matching score of the institution name (using the Levenshtein or Jaro-Winkler algorithm to handle minor differences in institution names, such as "Peking University" and "Peking University (Shenzhen)") with the semantic similarity of keywords in the research field (extracting keywords from the expert's existing achievements or self-described field and comparing them with the keywords of candidate achievements) to perform coarse-grained filtering and preliminary identity differentiation of the expert information to be processed, so as to initially screen out the candidate achievement dataset that is highly related to the target expert's identity.

[0041] Understandably, in the second stage, this embodiment uses a multi-dimensional weighted scoring and precise judgment mechanism to conduct a refined and comprehensive evaluation of the achievement data in the candidate achievement dataset. Specifically, this embodiment scores each achievement data by constructing a weighted scoring model. This weighted scoring model scores each "expert-candidate achievement" pair using configurable preset key dimensions to obtain a comprehensive score for each achievement data. The preset key dimensions include institutional affiliation matching degree, semantic consistency of the expert's research field, and career chronological reasonableness. Institutional affiliation matching degree refers to the degree of matching between the expert's institution corresponding to the candidate achievement data and the institution where the expert holds a position in their resume, both in terms of time and name. Semantic consistency of the expert's research field is calculated using the SciBERT model to determine the semantic similarity score between the title or abstract of the candidate achievement and the expert's core research direction. Career chronological reasonableness refers to whether the publication time of the achievement falls within the expert's active academic career window.

[0042] This embodiment generates a candidate result ranking list with confidence scores based on the calculated comprehensive score. Then, it selects several candidate results from the initial candidate dataset using a dynamically configurable confidence threshold and comprehensive score (for example, results with a comprehensive score > 0.9 are considered strong matches). A candidate result ranking table is generated based on all candidate results. Finally, an automated and high-precision information completion operation is performed on the expert information to be processed based on the candidate result ranking table, thereby improving the accuracy of information completion.

[0043] It is understood that structured audit and completion reports include, but are not limited to, audit reports and completion results. Audit reports include information such as risk type, location evidence, and violation standards, while completion results include information such as source API, match confidence level, and executed actions. Both audit reports and completion results can be output in JSON format. Specifically, the content of the audit report includes, but is not limited to, the following: Unique identifier for experts; Verification results (Passed, Pending review); Risk item details: An array containing all detected issues, each issue object containing: error_type: Error type (e.g., TEMPORAL_CONFLICT, PUBLICATION_NOT_FOUND); error_description: A detailed natural language description of the error; location: Precise location of the error message (e.g., education_history[1], publications[3].title); evidence: External evidence or internally contradictory data points that support the judgment; failed_criteria: Clearly indicates which specific expert inclusion criterion was violated.

[0044] Specifically, after reviewing the expert information to be processed, this embodiment can also have relevant personnel periodically conduct manual sampling reviews of cases that have passed the review and those marked as "pending review" to assess the false negative rate and false positive rate of the semantic contradiction detection algorithm's corresponding model. For implicit logical contradictions that are misjudged or missed by the semantic contradiction detection algorithm's corresponding model, this embodiment can have reviewers perform reverse annotation and, based on the actual situation, incrementally incorporate high-quality negative sample data into the next round of fine-tuning training of the model, thereby achieving continuous self-optimization and evolution of the model and improving the long-term stability and reliability of the semantic contradiction detection algorithm's corresponding model.

[0045] It is understood that the completion operation in this embodiment can be implemented using the following pseudocode: Algorithm output: A structured JSON object that clearly displays the information completed for a specific expert.

[0046] Generated json { "expert_id": "E1001", "completion_status": "SUCCESS", "completion_results": { "papers": [ { Title: Deep Learning-Based Medical Image Analysis "journal": "IEEE Transactions on Medical Imaging", "year": 2021, "coauthors": ["Zhang XX", "Li XX"], "source_api": "IEEE Xplore", "match_confidence": 0.96, "action": "APPEND" } ], "patents": [ { Title: "A Novel Image Denoising Method" "patent_id": "CN10XXXXXXA", "grant_date": "2022-05-18", "source_api": "CNIPA", "match_confidence": 0.98, "action": "APPEND" } ] } } This embodiment automates the expert information completion process using the pseudocode described above, thereby effectively improving the efficiency and accuracy of the completion results.

[0047] As can be seen from the above, the method of this application embodiment achieves automatic identification of deep-seated logical errors and false information through a semantic contradiction detection algorithm, thereby improving the efficiency and accuracy of review; it achieves high-precision and automated completion of missing expert information fields through a structured data alignment and completion algorithm, ensuring information integrity and timeliness; and through the collaborative work of the two algorithms, it jointly builds an intelligent management capability covering the entire process of expert information "review-completion", thereby reducing manual intervention and lowering management costs; in addition, it can output standardized and structured machine-readable results, thereby supporting downstream system integration and manual review, and improving the overall application system compatibility and availability.

[0048] Please see Figure 2 This application also provides an expert information management device based on dual-algorithm collaboration, the device comprising: The acquisition module is used to acquire information about the experts to be processed. The automatic review module is used to automatically review the expert information to be processed using a semantic contradiction detection algorithm, and obtain the expert information review results. The auto-completion module is used to automatically complete the expert information to be processed using a structured data alignment and completion algorithm, and obtain the expert information completion result; The report generation module is used to generate structured review and completion reports based on the expert information review results and expert information completion results.

[0049] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0050] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0051] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0052] Please see Figure 3 , Figure 3 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 310 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 320 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 320 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 320 and is called and executed by the processor 310 using the methods described in the embodiments of this application. Input / output interface 330 is used to realize information input and output; The communication interface 340 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 350 transmits information between various components of the device (e.g., processor 310, memory 320, input / output interface 330, and communication interface 340); The processor 310, memory 320, input / output interface 330 and communication interface 340 are connected to each other within the device via bus 350.

[0053] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0054] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0055] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0056] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0057] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0058] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0059] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0060] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0061] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0062] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0063] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0064] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0065] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for expert information management based on dual-algorithm collaboration, characterized in that, The method includes the following steps: Obtain information on experts awaiting processing; The expert information to be processed is automatically reviewed using a semantic contradiction detection algorithm to obtain the expert information review result; The expert information to be processed is automatically completed using a structured data alignment and completion algorithm to obtain the expert information completion result; A structured review and completion report is generated based on the expert information review results and the expert information completion results; The automatic review of the expert information to be processed using a semantic contradiction detection algorithm includes: Construct a large-scale model for text logic judgment; The text logic judgment model is used to assess the logical completeness and compliance of the expert information to be processed. Construct a multidimensional semantic similarity comparison model; The authenticity of the expert information to be processed is verified by the multidimensional semantic similarity comparison model. The construction of the large-scale text logic judgment model includes: Use an open-source large language model as the base model; Training data is obtained from a preset corpus, which includes academic paper information, patent specification information, industry standard information, or professional title evaluation document information. The base model is trained using the training data to learn word embedding representations. Construct logical constraint relationships, including temporal logical constraint relationships, hierarchical dependency constraint relationships, or compliance constraint relationships; The trained base model learns the logical constraint relationship through supervised fine-tuning, thus obtaining the large model for text logic judgment. The step of verifying the authenticity of the expert information to be processed using the multidimensional semantic similarity comparison model includes: Obtain candidate entry information retrieved from the first external database; The expert information to be processed is converted into a first high-dimensional semantic vector; The candidate entry information is converted into a second high-dimensional semantic vector; Calculate the cosine similarity between the first high-dimensional semantic vector and the second high-dimensional semantic vector; The semantic correlation between the expert information to be processed and the candidate item information is quantified based on the cosine similarity metric. The expert information to be processed is matched with the candidate item information by field matching; Generate authenticity verification results based on semantic relevance quantification results and field matching results; The automatic completion of the expert information to be processed using a structured data alignment and completion algorithm to obtain the expert information completion result includes: Use the first keyword in the expert information to be processed as the query term; The initial candidate dataset is obtained by querying data in a second external database using the query terms. A comprehensive score is generated based on a two-stage entity disambiguation and matching mechanism to correspond to the initial candidate dataset. Based on the comprehensive score, several candidate results are selected from the initial candidate dataset, and a candidate result ranking table is generated based on all the candidate results. Based on the candidate results ranking table, information completion processing is performed on the expert information to be processed; The process of generating a comprehensive score for the initial candidate dataset based on a two-stage entity disambiguation and matching mechanism includes: Extract keywords to be processed from the expert information to be processed; The keyword to be processed is compared with the keyword to be matched in the initial candidate dataset; Based on the comparison results, the data in the initial candidate dataset are filtered to obtain the candidate result dataset; The comprehensive score for each result in the candidate result dataset is calculated based on preset key dimensions. The preset key dimensions include the matching degree of institutional affiliation, semantic consistency of expert research fields, and the rationality of career chronology.

2. An expert information management device based on dual-algorithm collaboration, characterized in that, The apparatus is used in the method of claim 1, the apparatus comprising: The acquisition module is used to acquire information about the experts to be processed. The automatic review module is used to automatically review the expert information to be processed using a semantic contradiction detection algorithm to obtain the expert information review result; The automatic completion module is used to automatically complete the expert information to be processed using a structured data alignment and completion algorithm to obtain the expert information completion result; The report generation module is used to generate a structured review and completion report based on the expert information review results and the expert information completion results.

3. An electronic device, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor performs the method as described in claim 1.

4. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of claim 1.

Images