Performance determination method and device, equipment and medium

Abstract

The embodiment of the invention relates to a performance determination method and device, equipment and a medium. The method comprises the steps of obtaining performance source data of a to-be-evaluated person; performing structured processing and data cleaning processing on the performance source data by taking a to-be-evaluated person as an index item to obtain basic evaluation data corresponding to the to-be-evaluated person; wherein the basic evaluation data comprises a thesis reference topological graph and network interaction information; performing domain identification on the to-be-assessed person according to the thesis reference topological graph to obtain a cross-domain influence value of the to-be-assessed person; performing network influence analysis on the to-be-assessed person according to the network interaction information, and determining an interaction influence value of the to-be-assessed person; inputting the evaluation value into the trained performance measurement model to obtain a performance value of the to-be-evaluated person; wherein the evaluation value comprises a cross-domain influence value and an interaction influence value. Therefore, quantitative assessment of the influence of the to-be-assessed person is realized automatically, the consumed time is short, and the method comprises a plurality of dimensions representing the influence, and is objective and accurate.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a performance determination method, device, equipment and medium. Background Art

[0002] At present, when evaluating the performance of the personnel to be evaluated, the traditional evaluation method is to manually count the pre-set individual indicators such as the number of papers, citations, project completion, etc., and determine the performance based on the statistical results. However, this method takes a long time for a single evaluation, and it is difficult to dynamically and quantitatively evaluate the influence of the personnel to be evaluated. Summary of the invention

[0003] In order to solve the above technical problems or at least partially solve the above technical problems, the present disclosure provides a performance determination method, device, equipment and medium.

[0004] The present disclosure provides a performance determination method, including:

[0005] Obtain performance source data of the person to be evaluated;

[0006] Taking the person to be evaluated as an index item, the performance source data is structured and cleaned to obtain basic evaluation data corresponding to the person to be evaluated; wherein the basic evaluation data includes a paper citation topology map and network interaction information;

[0007] Performing field identification on the person to be evaluated according to the paper citation topology map to obtain the cross-field influence value of the person to be evaluated;

[0008] Performing a network impact analysis on the person to be evaluated according to the network interaction information to determine the interaction impact value of the person to be evaluated;

[0009] The evaluation value is input into the trained performance measurement model to obtain the performance value of the person to be evaluated; wherein the evaluation value includes the cross-domain impact value and the interactive impact value.

[0010] The present disclosure also provides a performance determination device, including:

[0011] An acquisition module is used to obtain the performance source data of the person to be evaluated;

[0012] A preprocessing module is used to perform structured processing and data cleaning processing on the performance source data with the person to be evaluated as an index item to obtain basic evaluation data corresponding to the person to be evaluated; wherein the basic evaluation data includes a paper citation topology map and network interaction information;

[0013] A first processing module is used to identify the field of the person to be evaluated according to the paper citation topology map to obtain the cross-field influence value of the person to be evaluated;

[0014] A second processing module is used to perform network impact analysis on the person to be evaluated according to the network interaction information to determine the interaction impact value of the person to be evaluated;

[0015] An input module is used to input the evaluation value into the trained performance measurement model to obtain the performance value of the person to be evaluated; wherein the evaluation value includes the cross-domain impact value and the interactive impact value.

[0016] An embodiment of the present disclosure also provides an electronic device, which includes: a processor; a memory for storing executable instructions of the processor; the processor is used to read the executable instructions from the memory and execute the instructions to implement the performance determination method provided in the embodiment of the present disclosure.

[0017] The embodiment of the present disclosure also provides a computer-readable storage medium, wherein the storage medium stores a computer program, and the computer program is used to execute the performance determination method provided by the embodiment of the present disclosure.

[0018] The technical solution provided by the embodiments of the present disclosure has the following advantages over the prior art: the performance determination solution provided in the embodiments of the present disclosure, the method comprising: obtaining the performance source data of the person to be evaluated; taking the person to be evaluated as the index item, performing structured processing and data cleaning processing on the performance source data, and obtaining basic evaluation data corresponding to the person to be evaluated; wherein the basic evaluation data includes a paper citation topology map and network interaction information; performing domain identification on the person to be evaluated based on the paper citation topology map, and obtaining the cross-domain influence value of the person to be evaluated; performing network impact analysis on the person to be evaluated based on the network interaction information, and determining the interactive influence value of the person to be evaluated; inputting the evaluation value into a trained performance measurement model, and obtaining the performance value of the person to be evaluated; wherein the evaluation value includes a cross-domain influence value and an interactive influence value. By adopting the above technical scheme, the performance source data is preprocessed based on the person to be evaluated, and the paper citation topology map and network interaction information of the person to be evaluated are obtained. The influence value of the person to be evaluated in the cross-field citation dimension of the paper is determined based on the paper citation topology map, and the influence value of the person to be evaluated in the network interaction dimension is determined based on the network interaction information. The performance value of the person to be evaluated is determined based on the influence values ​​of the above multiple dimensions through the model, which realizes the automatic dynamic quantitative evaluation of the influence of the person to be evaluated. It takes a short time and includes multiple dimensions that represent influence. The evaluation result is more objective and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0021] Figure 1 A flowchart of a performance determination method provided in an embodiment of the present disclosure;

[0022] Figure 2 A schematic diagram of a process for determining basic evaluation data provided by an embodiment of the present disclosure;

[0023] Figure 3 A schematic diagram of a process for determining a cross-domain impact value provided by an embodiment of the present disclosure;

[0024] Figure 4 A schematic diagram of a process for determining an interactive influence value provided by an embodiment of the present disclosure;

[0025] Figure 5 A schematic diagram of the architecture of a performance determination method provided in an embodiment of the present disclosure;

[0026] Figure 6 A schematic diagram of a performance determination method provided in an embodiment of the present disclosure;

[0027] Figure 7 A schematic diagram of the structure of a performance determination device provided in an embodiment of the present disclosure;

[0028] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0029] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0030] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0031] With the rapid development of the knowledge economy, scientific research activities have become a key force in promoting social progress and economic growth. However, in the practice of scientific research management, how to effectively evaluate the performance of personnel and rationally allocate scientific research resources is a problem that needs to be solved urgently. In traditional scientific research performance evaluation methods, performance is mostly determined manually based on multiple pre-set static indicators, including the number of papers, number of citations, project completion, etc. This evaluation method has one or more of the following problems:

[0032] One-sidedness: Specifically, performance is determined based on the single dimension of quantity, ignoring the application value of quality and scope of influence, which is relatively limited and one-sided. Lag: Specifically, the end of an event is used as the evaluation node, which makes it difficult to dynamically and real-time represent the dynamic changes in the influence of personnel during the process of the event. Difficult to standardize: Specifically, there are significant differences in scientific research projects in different disciplines and of different types. Quantitative standards based on quantity are difficult to fully cover the characteristics and needs of all scientific research activities, and the universality of evaluation standards is poor.

[0033] It can be seen that in the relevant technologies, a single evaluation takes a long time and it is difficult to dynamically and quantitatively evaluate the influence of the person being evaluated. A more scientific and refined performance evaluation method for scientific researchers is urgently needed.

[0034] In order to solve the above problems, the embodiments of the present disclosure provide a performance determination method, which is introduced below in conjunction with specific embodiments.

[0035] Figure 1 The present invention provides a flow chart of a performance determination method provided in an embodiment of the present invention. The performance determination method can be applied to a performance determination device. The performance determination device can be implemented by software and / or hardware. The performance determination device can generally be integrated in an electronic device. Figure 1 As shown, the performance determination method includes:

[0036] Step 101, obtaining the performance source data of the person to be evaluated.

[0037] Among them, the person to be evaluated may be a person to be subjected to performance evaluation, and this embodiment does not limit the type of the person to be evaluated. For example, the person to be evaluated may be a scientific researcher. The performance source data may be original data recording the work situation of the person to be evaluated. The performance source data may include data recorded in one or more scientific research systems, academic databases, and online social platforms. Among them, the scientific research system may include an internal system, a laboratory system, a project system, and the like. The academic database may include an internal database, an external database, and the like. The online social platform may include an official website, social media, and the like.

[0038] In the disclosed embodiment, different performance source data may have corresponding standardized data interfaces (Application Programming Interface, API), and the performance determination device may periodically capture the performance source data of the person to be evaluated through the data interface to keep the performance source data updated in real time. Alternatively, the performance source data may be obtained through an Extract Transform Load (ETL) tool, and the Extract Transform Load tool sends the performance source data to the performance determination device, which obtains the performance source data.

[0039] In some embodiments of the present disclosure, the obtaining of the performance data of the person to be evaluated includes: if the current user passes the identity authentication, obtaining the performance data of the person to be evaluated. The current user may be the user who currently requests for performance determination.

[0040] In this embodiment, a decentralized identity (DID) can be distributed to each user through blockchain calculation. The performance determination device first sends the identity of the current user to the data source where the performance source data is located. The data source verifies the identity. If the verification is successful, the performance source data is returned to the performance determination device, and the performance determination device obtains the performance source data. This reduces the risk of performance source data leakage, and does not rely on the central authentication server. The user's permission changes and access records are stored in the blockchain, ensuring the transparency and immutability of the access records, which is convenient for subsequent auditing and tracing.

[0041] Optionally, the performance determination method further includes: deleting performance source data that meets an abnormality judgment condition. Wherein, meeting the abnormality judgment condition may include at least one of the following:

[0042] Optionally, the abnormal judgment condition includes: the data characteristics of the performance source data are outside the normal characteristic interval, wherein the normal characteristic interval can be determined based on historical data, or the normal characteristic interval can be manually set. In this embodiment, the normal characteristic interval can be dynamically adjusted based on the historical value of the data or manual experience. For example, the performance determination device can determine the normal characteristic interval corresponding to the duration and the normal characteristic interval corresponding to the number of achievements through a machine learning model based on the average duration of historical projects, the number of achievements, etc., thereby achieving adaptive detection of the project cycle or the number of achievements corresponding to the performance source data. Optionally, the review period of the performance source data through the abnormal judgment condition can also be dynamically adjusted through a machine learning model. For example, for a data source that is frequently updated, the review frequency is automatically increased.

[0043] Optionally, the abnormality judgment condition includes: the presence of an abnormal label in the performance source data. The abnormal label can be a label that characterizes the accuracy abnormality of the performance source data as reported by the user. For example, if the user marks the performance source data with an abnormal label, the performance determination device can delete the performance source data if it is determined that the performance source data corresponds to the abnormal label. The abnormal label can be determined based on one or more of the usage history, feedback, and update records of the data source corresponding to the performance source data. For example, if data errors frequently occur in the data of a certain data source, the data generated by the data source can be removed from the performance source data by marking the abnormal label.

[0044] Optionally, the abnormality judgment condition includes: the performance source data fails to pass the cross-validation. In this embodiment, the performance determination device can cross-validate the performance source data of different data sources, and if the performance source data fails to pass the cross-validation, the performance source data is deleted. For example, the performance in the scientific research system, the popularity in the academic database, and the comments in the network social platform can be cross-validated for consistency. Thereby, the accuracy of abnormality detection is improved.

[0045] Optionally, the abnormal judgment condition includes: the comprehensive score of the performance source data is lower than a preset score threshold, and the comprehensive score can be a comprehensive score of the data's accuracy, timeliness, completeness, stability and other dimensions, and the data quality of the performance source data can be characterized by the comprehensive score. Based on the comprehensive score, low-quality performance source data can be regularly screened and eliminated.

[0046] Step 102, using the person to be evaluated as an index item, performs structured processing and data cleaning on the performance source data to obtain basic evaluation data corresponding to the person to be evaluated; wherein the basic evaluation data includes a paper citation topology map and network interaction information.

[0047] Among them, the index item can be the basis for processing the performance source data, and the index item can be understood as the primary key. Structured processing can be the processing of converting unstructured or semi-structured data into structured data. Data cleaning processing is also called data cleaning processing or data purification processing. Through data cleaning processing, erroneous data can be discovered and corrected, and duplicate data can be deleted. Optionally, the data cleaning processing can be implemented through stream processing technology (for example, Apache Kafka or Flink). In the process of data cleaning processing, the cleaning strategy can be dynamically adjusted according to user needs.

[0048] The basic evaluation data may be data that has been structured and cleaned, and the basic evaluation data may be the basic data used as the basis for the performance evaluation of the person to be evaluated. The paper citation topology diagram may represent the citation relationship between the paper with the person to be evaluated as the author and other papers. The paper citation topology diagram may include multiple nodes and lines between nodes, where the nodes represent the papers, and the lines between the nodes represent the citation relationship between the papers. The network interaction information may be information on network interaction with the person to be evaluated. In some embodiments of the present disclosure, the basic evaluation data also includes the scientific research projects, laboratories and technology platforms, scientific and technological achievements, achievement transformation, scientific and technological awards, talent training, etc. of the person to be evaluated.

[0049] In this embodiment, structured processing of performance source data can be achieved by uniformly converting performance source data into an intermediate format. This embodiment does not limit the format conversion method. An example is described as follows: the performance source data is automatically mapped to the format template of the intermediate format through a machine learning model to achieve format unification. This embodiment does not limit the intermediate format. For example, the intermediate format may include JSON, CSV, etc. Specifically, the machine learning model automatically learns the format conversion rules between different performance source data and the intermediate format based on field names, data distribution characteristics, context and other information, and automatically determines the mapping relationship between the fields of different performance source data and the format model. For example, the machine learning model can automatically identify the author sequence and convert it to a standard format. Thereby, an automatic format mapping mechanism is implemented.

[0050] Alternatively, the performance determination device can automatically identify the format of the performance source data through a process automation (Robotic Process Automation, RPA) tool, and perform batch formatting of the performance source data according to the format of the performance source data. The process automation tool can simulate user operations and automatically convert data formats between data sources of different formats.

[0051] Figure 2 A schematic diagram of a process for determining basic evaluation data provided by an embodiment of the present disclosure, such as Figure 2 As shown, in some embodiments of the present disclosure, the person to be evaluated is used as an index item, and the performance source data is structured and cleaned to obtain the basic evaluation data corresponding to the person to be evaluated, including:

[0052] Step 201, extract the names of the performance source data to obtain the names of the personnel to be evaluated.

[0053] The person name may be the name of the person to be evaluated.

[0054] In this embodiment, the performance determination device can input the performance source data into the name extraction model to obtain the name of the person to be evaluated recorded in the performance source data.

[0055] Step 202, determining the performance attribute information corresponding to the personnel name in the performance source data; wherein the performance attribute information includes an attribute name and an attribute value.

[0056] Among them, a performance attribute information can be used to record the performance of the person to be evaluated in one dimension. The attribute name can be a name that identifies the performance-related attribute of the person to be evaluated. The attribute name can be pre-set or extracted and determined from the performance source data. This embodiment does not limit the attribute name. For example, the attribute name can include attribute names extracted from scientific research systems such as project number, project name, and project start date. The attribute name can include attribute names extracted from academic databases such as publication number, title, author ranking, and publication date. The attribute value can be the data content corresponding to the attribute name, and the attribute value can be a specific value extracted from the performance source data. The attribute name can be understood as information used to record dimensions in the performance attribute information, and the attribute value can be understood as information used to record specific performance in the performance attribute information.

[0057] In this embodiment, after determining the personnel name, the performance determination device can extract data from multiple performance source data according to the personnel name, extract the attribute name associated with the personnel name and the attribute value under the attribute name from the multiple performance source data, combine a pair of attribute name and attribute value into a performance attribute information, and use the performance attribute information determined by different performance source data as the performance attribute information of the person to be evaluated. Thus, data fusion between different performance source data is achieved.

[0058] Optionally, the performance attribute information may be in an intermediate format. In this embodiment, the performance determination device may convert the performance source data into an intermediate format, thereby achieving structured processing of the performance source data through format conversion, and achieving unification of field naming and format, as well as mapping and standardization of fields.

[0059] Step 203, performing data cleaning processing on the performance attribute information to obtain basic evaluation data.

[0060] In this embodiment, after determining the performance attribute information, the performance determination device may perform data cleaning on the performance attribute information to correct erroneous information in the performance attribute information and delete redundant information in the performance attribute information to obtain basic evaluation data.

[0061] In the above scheme, the name of the performance source data as a whole is first extracted, and then the performance attribute information of the performance source data as a whole is extracted based on the personnel name, which improves the efficiency and accuracy of heterogeneous data integration.

[0062] In some embodiments of the present disclosure, data cleaning is performed on the performance attribute information to obtain basic evaluation data, including:

[0063] Step a1, clustering the performance attribute information according to the attribute name to obtain multiple information groups.

[0064] Among them, one information group may include one performance attribute information, or one information group may include multiple performance attribute information whose attribute name text similarity is less than the similarity threshold. This embodiment does not limit the type of the text similarity, for example, the type of the text similarity may include Jaccard similarity and cosine similarity.

[0065] In this embodiment, the performance determination device can extract the attribute name in the performance attribute information, and perform clustering processing on the performance attribute information according to the attribute name, and classify the performance attribute information whose text similarity of the attribute name is less than the similarity threshold into the same information group. For example, scientific researchers and project personnel can be classified into the same information group.

[0066] Step a2: among the multiple information groups, determine the information group including multiple attribute values ​​and the similarity of the multiple attribute values ​​being greater than a similarity threshold as the information group to be processed.

[0067] The information group to be processed may be an information group in which the performance attribute information is to be deleted. For example, the information group to be processed may include a plurality of substantially identical performance attribute information extracted from scientific research papers or reports.

[0068] In this embodiment, among multiple information groups, the performance determination device may determine an information group including multiple performance attribute information as an intermediate information group, and determine the similarity of multiple attribute values ​​corresponding to multiple performance attribute information in the intermediate information group. If the similarity is greater than the similarity threshold, it means that the repetition of the performance attribute information in the intermediate information group is high, and the intermediate information group is determined as the information group to be processed. This embodiment does not limit the method for determining the similarity. For example, the similarity can be determined by a pre-trained model (for example, a BERT model).

[0069] Step a3: determine a piece of performance attribute information in the information group to be processed as the first attribute information, and delete the performance attribute information in the information group to be processed except the first attribute information.

[0070] The first attribute information may be a performance attribute information in the information group to be processed, and the first attribute information may be the performance attribute information retained in the information group to be processed.

[0071] In this embodiment, the performance determination device may determine a random performance attribute information in the information group to be processed as the first attribute information, and retain the first attribute information. Alternatively, the performance determination device may determine the performance attribute information corresponding to the attribute name with the highest priority as the first attribute information based on the priority of the pre-set attribute name. For example, if the priority of the "project leader" attribute name is higher than that of the "author" attribute name, the performance attribute information corresponding to the "project leader" attribute name is determined as the first attribute information. Furthermore, the performance attribute information other than the first attribute information in the information group to be processed is deleted.

[0072] In the above scheme, the performance attribute information is first clustered by attribute name, and the performance attribute information with substantially the same semantic attribute names are divided into the same information group. When the multiple attribute values ​​included in the information group are substantially semantically the same, one performance attribute information in the information group is retained, thereby deleting duplicate data and solving the problem of attribute name conflicts.

[0073] In some embodiments of the present disclosure, data cleaning is performed on the performance attribute information to obtain basic evaluation data, including:

[0074] Step b1, in the performance attribute information, the performance attribute information whose sequence digit number of the attribute value is less than the digit threshold is determined as the second attribute information; wherein the digit threshold corresponds to the attribute name.

[0075] The number of sequence digits may be the number of digits of the attribute value in the numbered form. The number of digits threshold may be a standard value of the number of digits corresponding to the attribute name. For example, the attribute name may be a certificate number, and the attribute value may be a specific certificate number value. The second attribute information may be performance attribute information to be completed with an attribute value.

[0076] In this embodiment, the performance determination device can input the attribute value in the form of a number into the attribute value detection model, determine the number of sequence digits of the attribute value through the attribute value detection model, and judge whether the number of sequence digits is less than the digit threshold. If so, the performance determination device determines the performance attribute information as the second attribute information. Among them, the attribute value detection model can be a neural network model, and this embodiment does not limit the attribute value detection model. For example, the attribute value detection model can be an isolation forest algorithm model or a one-class support vector machine (One-Class SVM) algorithm model.

[0077] Step b2, determining the missing bits of the attribute value in the second attribute information, and determining the missing elements corresponding to the missing bits according to the coding structure relationship corresponding to the attribute names in the second attribute information.

[0078] Among them, the vacant position can be the number of digits where data is missing in the attribute value. The coding structure relationship is used to record the coding rules corresponding to the attribute name. The coding structure relationship can record the elements corresponding to different digits in the number, and the information corresponding to the specific value under the element. For example, the element corresponding to the first digit in the number can be the graduation school, the information corresponding to the value 1 in the digit can be school A, the information corresponding to the value 2 can be school B, and the information corresponding to the value 3 can be school C. The coding structure relationship can be understood as the distribution law of the number.

[0079] In this embodiment, there are multiple methods for determining the vacant position, which are not limited in this embodiment. For example, the performance determination device can determine the vacant position of the attribute value in the second attribute information according to the coding structure relationship, or the attribute value detection model can detect the attribute value in the second attribute information to determine the vacant position corresponding to the attribute value. Further, the performance determination device can query the coding structure relationship according to the vacant position to determine the vacant element corresponding to the vacant position.

[0080] Step b3, query and process the performance source data according to the vacancy elements and personnel names to obtain the vacancy information corresponding to the vacancy elements.

[0081] The vacancy information may be information corresponding to the dimension of the vacancy element by the person to be evaluated.

[0082] In this embodiment, the performance determination device can query the vacant element corresponding to the name of the person in the performance source data to obtain the vacant information corresponding to the vacant element. For example, if the vacant element is the graduation school, the graduation school of the person to be evaluated is queried in the performance source data.

[0083] Step b4, completing the attribute value in the second attribute information according to the missing information and the coding structure relationship to obtain basic evaluation data.

[0084] In this embodiment, the performance determination device can query the coding structure relationship based on the vacancy information, obtain the vacancy number corresponding to the vacancy information, and insert the vacancy number into the vacancy position in the attribute value of the second attribute information, and determine the completed second attribute information as the basic evaluation data.

[0085] In the above solution, automatic filling and repair of attribute values ​​of number type is implemented.

[0086] In some embodiments of the present disclosure, data cleaning is performed on the performance attribute information to obtain basic evaluation data, including:

[0087] If the attribute value in the attribute information includes multiple attribute values ​​with a time series relationship, the outlier in the attribute value is determined according to the trend of the attribute value. Taking the output results of the person to be evaluated as an example, if there is a one-year output result that is significantly prominent in the trend of the output results, the attribute value in the performance attribute information is determined as an outlier, and the performance attribute information is further checked later. Optionally, if at least two attribute values ​​deviate from the baseline for determining the attribute value, the two attribute values ​​are not determined as outliers to avoid data being mistakenly deleted. Optionally, if the attribute name corresponding to the outlier is the number of citations, an anomaly detection model (for example, a K-means algorithm model or a DBSCAN clustering algorithm model) is used to identify whether there are extreme values ​​or duplicate data in the performance source data. If so, the performance source data is automatically corrected according to contextual information (for example, the publication date of the paper, research field, etc.).

[0088] This embodiment does not limit the method for determining the abnormal value. An example is described as follows:

[0089] In an optional implementation, a statistical algorithm (e.g., standard score Z-Score, interquartile range) is used to detect outliers in attribute values ​​with a time series relationship. For example, for the number of citations of a paper by a person to be evaluated, its standard score (Z-Score) is calculated: if Z-Score>3, the number of citations of the paper is considered to be abnormal data. Specifically,

[0090]

[0091] Wherein, X is the attribute value currently being processed. In this embodiment, X may be the number of citations of the paper currently being judged, μ is the mean number of citations of the paper, σ is the standard deviation of the number of citations of the paper, and Z is the standard score.

[0092] In another optional implementation, outliers in attribute values ​​with a time series relationship are identified using unsupervised learning algorithm models such as an isolation forest algorithm model and a one-class support vector algorithm model.

[0093] In some embodiments of the present disclosure, data cleaning is performed on the performance attribute information to obtain basic evaluation data, including: data cleaning is performed according to context information of the performance attribute information to obtain basic evaluation data.

[0094] For example, if the change in the attention of the person to be evaluated on the social platform exceeds the change threshold, it is determined whether the activity data of the person to be evaluated on the social platform is consistent with the increase in the attention. If not, the attention is determined as abnormal data. Alternatively, if the change in the capital expenditure of the person to be evaluated exceeds the change threshold, it is determined whether the project stage corresponding to the capital expenditure is a preset stage. If not, the capital expenditure is determined as abnormal data; wherein the preset stage can be a pre-set stage of increased capital expenditure. Thus, the accidental deletion of data due to data fluctuations is avoided.

[0095] Optionally, in some embodiments of the present disclosure, if abnormal values ​​continue to appear in attribute values ​​with a time series relationship, it may be that the data pattern has changed, and an automatic feedback mechanism can be triggered to optimize the deletion rules of the data. For example, when some data in the real-time data stream is inconsistent with the historical pattern, the performance determination device can update the deletion rules through the feedback mechanism. For example, if the publication data of the paper surges and remains, the performance determination device can automatically adjust the deletion rules and not delete the data.

[0096] In some embodiments of the present disclosure, data cleaning is performed on performance attribute information to obtain basic evaluation data, including: if the attribute value in the attribute information includes multiple attribute values ​​with a time series relationship, the missing value is determined according to the trend of the attribute value to be supplemented.

[0097] In this embodiment, the performance determination device can predict missing values ​​based on attribute values ​​with a temporal relationship through regression analysis or a deep learning model (e.g., a long short-term memory network model). For example, for missing values, K persons with similar characteristics to the person to be evaluated can be determined, where K is a positive integer. The mean, mode, or median of the correlation values ​​corresponding to the K persons and the missing value is taken as the filling value corresponding to the missing value.

[0098] Alternatively, the performance determination device can estimate missing values ​​based on multiple attribute values ​​with a temporal relationship through a generative adversarial network (GANs) or an autoencoder. For example, the missing values ​​can be estimated based on the training of a deep learning model to infer the missing values ​​from the existing pattern and the missing values ​​can be estimated based on the Multiple Imputation by Chained Equations (MICE) technique.

[0099] Step 103, identifying the domain of the person to be evaluated according to the paper citation topology map, and obtaining the cross-domain influence value of the person to be evaluated.

[0100] Among them, the cross-domain impact value can represent the degree of influence of the person to be evaluated in multiple academic fields.

[0101] In this embodiment, the performance determination device can determine the fields involved by the person to be evaluated according to the paper citation topology map, and determine the cross-field impact value of the person to be evaluated according to the number of fields.

[0102] Figure 3 A schematic diagram of a process for determining a cross-domain impact value provided by an embodiment of the present disclosure, such as Figure 3 As shown, in some embodiments of the present disclosure, the field of the person to be evaluated is identified according to the paper citation topology map, and the cross-field influence value of the person to be evaluated is obtained, including:

[0103] Step 301, in the paper citation topology diagram, determine the paper at the previous level of the target paper as the citing paper; wherein the target paper is a paper whose author includes the person to be evaluated.

[0104] The citing paper may be a paper that cites the target paper, and in the paper citation topology graph, the citing party of the paper points to the cited party of the paper.

[0105] In this embodiment, the performance determination device can determine the target paper in the paper citation topology map, and determine the papers located one level above the target paper in the paper citation topology map as cited papers, that is, the performance determination device can determine the papers pointing to the target paper in the paper citation topology map as cited papers.

[0106] Step 302: In the paper citation topology diagram, the papers at the next level of the target paper are determined as cited papers.

[0107] The cited paper may be a paper cited by the target paper.

[0108] In this embodiment, the performance determination device may determine the papers located at the next level of the target paper in the paper citation topology map as cited papers, that is, the performance determination device may determine the papers referred to by the target paper in the paper citation topology map as cited papers.

[0109] Step 303, determining the cross-domain impact value according to the number of citing papers in the first field, the number of cited papers in the second field, the preset citation weight, and the preset cited weight.

[0110] The first field number may be the number of fields involved in the cited papers, and the second field number may be the number of fields involved in the cited papers. The preset citation weight may be the weight of the number of cited papers set in advance, and this embodiment does not limit the preset citation weight. The preset cited weight may be the weight of the number of cited papers set in advance, and this embodiment does not limit the preset cited weight.

[0111] In this embodiment, the performance determination device can count the number of scientific research fields corresponding to all cited papers to obtain the first field number, and multiply the first field number by the preset citation weight to obtain the first product. The performance determination device can count the number of scientific research fields corresponding to all cited papers to obtain the second field number, and multiply the second field number by the preset citation weight to obtain the second product. Further, the first product and the second product are added to obtain the cross-field impact value.

[0112] In the above scheme, the number of fields in which the target paper of the person to be evaluated is cited, as well as the number of fields in which the target paper is cited are determined. Based on these two numbers, the cross-field impact value of the person to be evaluated is determined, which represents the cooperation and influence between the person to be evaluated and people in other disciplines.

[0113] It is understandable that by analyzing the citations of a paper, the influence of the paper on the academic community can be judged. The higher the number of citations, the more widely recognized the paper is. In some embodiments of the present disclosure, the basic evaluation data includes the number of citations of the paper, and the evaluation value includes the trend value of the number of citations. The method also includes: real-time dynamic tracking of the growth of the number of citations of the paper, determining the citation trend value, and the citation trend value can characterize the change of the number of citations of the paper.

[0114] In some embodiments of the present disclosure, the performance determination method further includes: identifying personnel relationships of the person to be evaluated based on the paper citation topology map to obtain the personnel influence value of the person to be evaluated.

[0115] In this embodiment, the performance determination device can determine the authors corresponding to each paper in the paper citation topology map, and establish a relationship map between the persons corresponding to the paper citation topology map. The relationship map can represent the connection between the person to be evaluated and other persons. The relationship map can be understood as a cooperation network of persons, and the relationship map can represent the academic influence of the person to be evaluated.

[0116] In some embodiments of the present disclosure, the basic evaluation data includes paper keywords, the evaluation value includes a hotspot impact value, and the performance determination method further includes: obtaining paper keywords of the target paper, and determining the hotspot impact value of the user to be evaluated based on the paper keywords and pre-set hotspot keywords. The hotspot keywords can be determined by keyword extraction and hotspot trend analysis of paper resources, and the hotspot keywords can be updated in real time.

[0117] In this embodiment, keywords are extracted from the target paper in advance to obtain paper keywords. The paper keywords are compared with the hot keywords to obtain text similarity, and the text similarity is multiplied by the influence of the target paper to obtain the hot influence value. Thus, the innovation of the person to be evaluated is quantified.

[0118] Step 104 , performing network impact analysis on the person to be evaluated based on the network interaction information, and determining the interaction impact value of the person to be evaluated.

[0119] Among them, the interactive impact value can represent the degree of influence caused by the person to be evaluated through network interaction.

[0120] In some embodiments of the present disclosure, the performance determination device may analyze the network influence of the person to be evaluated in multiple dimensions such as the amount of interaction and the scope of interaction based on the network interaction information to obtain the interaction influence value of the person to be evaluated.

[0121] Figure 4 A schematic diagram of a process for determining an interactive influence value provided by an embodiment of the present disclosure, such as Figure 4 As shown, in some embodiments of the present disclosure, a network impact analysis is performed on the person to be evaluated based on the network interaction information to determine the interaction impact value of the person to be evaluated, including:

[0122] Step 401, based on the network interaction information, a plurality of pre-set interaction types are counted to obtain a first statistical quantity corresponding to the plurality of interaction types of the person to be evaluated; wherein the interaction types include sharing types and comment types.

[0123] Among them, the network interaction information can be information collected based on different network platforms, and the comprehensive performance of the person to be evaluated on different network platforms can be characterized based on the network interaction information. The interaction type can be the type of network interaction behavior, and this embodiment does not limit the interaction type. The sharing type can be the type of sharing of the content published by the person to be evaluated on the network. The comment type can be the type of comment on the content published by the person to be evaluated on the network. The content published by the person to be evaluated on the network can be the scientific research results or articles of the person to be evaluated. The first statistical quantity can be the number of interaction types, and the first statistical quantity can include the number of likes, the number of shares, the number of comments, etc. It can be understood that the first statistical quantity can be a direct reflection of the attention paid to the person to be evaluated on the network.

[0124] In this embodiment, the performance determination device obtains network interaction information, and performs real-time quantitative statistics on the interaction types to which the network interaction information belongs, to obtain first statistical quantities corresponding to different interaction types.

[0125] Step 402, determining the interaction value according to the first statistical quantity and the second statistical quantity; wherein the second statistical quantity is the quantity of target interactive information, and the target interactive information is network interactive information of the comment type whose character quantity is greater than a preset quantity threshold.

[0126] The number of characters may be the number of characters included in the target interactive information. The preset number threshold may be a preset threshold for determining whether the network interactive information is a long evaluation. The interactive value may be used to characterize the amount of network interaction of the user to be evaluated.

[0127] In this embodiment, the performance determination device can determine the network interaction information of the comment type whose number of characters is greater than a preset number threshold in the network interaction information of the comment type as the target interaction information, and count the number of the target interaction information to obtain the second statistical number. Further, the performance determination device can assign corresponding weights to the first statistical number and the second statistical number according to the type of field in which the person to be evaluated is located, and perform weighted summation of the first statistical number and the second statistical number to obtain the interaction value. Optionally, if the field type is the first type in which in-depth discussion is greater than wide dissemination, the weight of the second statistical number is greater than the weight of the first statistical number; if the field type is the second type in which wide dissemination is greater than in-depth discussion, the weight of the first statistical number is greater than the weight of the second statistical number.

[0128] Optionally, the first statistical quantities corresponding to different network platforms may have corresponding platform weight values, and the platform weight value of the scientific research platform is smaller than the platform weight value of the social platform. Optionally, the first statistical quantities corresponding to different types of users may have corresponding user weights, and the user weights of media accounts and domain authority accounts may be greater than the user weights of ordinary users.

[0129] Step 403: determine the interactive position corresponding to the network interactive information and the interactive area where the interactive position is located.

[0130] The interactive location may be the location of the person who triggers the generation of network interactive information, and the interactive location may be understood as the location of the person who interacts with the person to be evaluated through the network. The interactive area may be the area where the interactive location is located, and this embodiment does not limit the granularity of the interactive area.

[0131] In this embodiment, the performance determination device can obtain the interactive position corresponding to the network interactive information, and determine the interactive area where the interactive position is located.

[0132] Step 404, calculate the distance between the interaction position and the position of the person to be evaluated to obtain the interaction distance, and determine the interaction range value according to the interaction distance and the number of areas; wherein the number of areas is the number of interaction areas.

[0133] The person location may be the location when the person to be evaluated sends the content corresponding to the network interaction information. The interaction distance may be the distance between the two parties of the interaction. The interaction distance may characterize the scope of influence of the content published by the person to be evaluated. The interaction range value may be used to characterize the range of people who interact with the person to be evaluated online. The interaction range value may characterize the distance of propagation and the area covered by the content published by the person to be evaluated.

[0134] In this embodiment, the performance determination device can obtain the personnel position of the evaluator and calculate the distance between the personnel position and the person to be evaluated to obtain the interaction distance. Further, the interaction range value is determined based on the interaction distance and the number of regions. Optionally, the performance evaluation device can also determine the audience range of the person to be evaluated based on the industry background information of the person who conducts network interaction information.

[0135] Step 405: Determine the interaction impact value according to the interaction amount value and the interaction range value.

[0136] Among them, the interactive influence value can represent the influence of the person to be evaluated in the network interaction.

[0137] In this embodiment, after determining the interaction value and the interaction range value, the performance evaluation device may perform a weighted summation of the interaction value and the interaction range value to obtain the interaction impact value.

[0138] In the above scheme, the interaction impact value is comprehensively determined from multiple dimensions such as the number of interactions, interaction quality, interaction scope, and interaction coverage area. The interaction impact value can more comprehensively represent the impact of the network interaction of the user to be evaluated.

[0139] In some embodiments of the present disclosure, the evaluation value includes a sentiment impact value, and the performance determination method further includes: performing sentiment analysis on the comment type network interaction data through a sentiment model to obtain the sentiment impact value of the person to be evaluated.

[0140] In this embodiment, the performance determination device can perform sentiment analysis on the online interactive data of the comment type through the sentiment perception dictionary and sentiment reasoner, pre-trained language model, etc., to obtain analysis results, and calculate and process multiple analysis results to obtain sentiment impact values. This embodiment does not limit the type of the analysis result. For example, the analysis result may include positive and negative; wherein positive represents positive emotions and negative represents negative emotions. The analysis result may be trust, support, etc.

[0141] Optionally, the evaluation value includes a sentiment trend value, and the performance determination method further includes: performing trend analysis on the analysis results of the sentiment analysis according to a time series to obtain a sentiment trend value, and the sentiment trend value can represent the user's emotional changes towards the person to be evaluated.

[0142] Step 105, input the evaluation value into the trained performance measurement model to obtain the performance value of the person to be evaluated; wherein the evaluation value includes a cross-domain impact value and an interactive impact value.

[0143] The performance record model may be a neural network model that determines the performance value based on the evaluation value. The performance value may be a quantitative result of the work of the person to be evaluated.

[0144] In this embodiment, the performance determination device can input the evaluation value and the preset evaluation value weight corresponding to the evaluation value into the performance measurement model, and redistribute the weights of the evaluation values ​​with reference to the input evaluation value weights through the performance measurement model, and perform weighted calculation on the evaluation values ​​according to the redistributed weights to obtain the performance value of the person to be evaluated.

[0145] In some embodiments of the present disclosure, the training process of the performance measurement model includes: obtaining sample data; wherein the sample data corresponds to the sample personnel one-to-one, and the sample data includes multiple evaluation sample values, sample weights corresponding to each evaluation sample value, and performance sample values ​​corresponding to the multiple evaluation sample values; inputting the evaluation sample values ​​and the sample weights into the initial model to obtain the predicted sample values ​​output by the initial model; training the initial model according to the predicted sample values ​​and the performance sample values ​​to obtain the performance measurement model.

[0146] The sample data may be data for training the initial model. The evaluation sample value may be an evaluation value corresponding to a known performance value. The performance sample value may be a known performance value, which may be understood as a standard value of the performance value. The initial model may be a neural network model that has not yet completed training. The predicted sample value may be the result obtained by the initial model predicting the performance value according to the input evaluation sample value and sample weight.

[0147] In this embodiment, sample data is first constructed, specifically, different types of evaluation sample values, sample weights corresponding to each evaluation sample value, and performance sample values ​​corresponding to multiple evaluation sample values ​​are obtained. Furthermore, the evaluation sample values ​​and sample weights are used as features of the input initial model, and the evaluation sample values ​​are used as standard results of the initial model output to train the initial model. Through the initial model, the evaluation sample values ​​can be weighted again with reference to the input sample weights and the stage of the project. For example: if the project is in the initial stage. The weight of the cross-domain impact value can be greater than the weight of the interactive impact value, thereby reflecting the comprehensiveness of the field embodied by the person to be evaluated in the project. In the application stage of the project, the weight of the interactive impact value can be greater than the weight of the cross-domain impact value, thereby reflecting the influence of the results of the person to be evaluated in the project.

[0148] The evaluation sample values ​​are weighted according to the re-divided weights to obtain the predicted sample values. For example, if w1 and w2 correspond to the sample weights of the cross-domain impact value and the sample weights of the interactive impact value, respectively, the initial model will adjust the sample weights and perform a weighted summation of the cross-domain impact value and the interactive impact value according to the adjusted weights to obtain the predicted sample values. Furthermore, the initial model is trained according to the predicted sample values ​​and the performance sample values, and finally a performance measurement model that has been trained is obtained.

[0149] The performance determination scheme provided in the embodiments of the present disclosure includes: obtaining the performance source data of the person to be evaluated; taking the person to be evaluated as the index item, performing structured processing and data cleaning processing on the performance source data to obtain the basic evaluation data corresponding to the person to be evaluated; wherein the basic evaluation data includes a paper citation topology map and network interaction information; performing field identification on the person to be evaluated based on the paper citation topology map to obtain the cross-field influence value of the person to be evaluated; performing network impact analysis on the person to be evaluated based on the network interaction information to determine the interactive influence value of the person to be evaluated; inputting the evaluation value into a trained performance measurement model to obtain the performance value of the person to be evaluated; wherein the evaluation value includes a cross-field influence value and an interactive influence value. By adopting the above technical scheme, the performance source data is preprocessed based on the person to be evaluated, and the paper citation topology map and network interaction information of the person to be evaluated are obtained. The influence value of the person to be evaluated in the cross-field citation dimension of the paper is determined based on the paper citation topology map, and the influence value of the person to be evaluated in the network interaction dimension is determined based on the network interaction information. The performance value of the person to be evaluated is determined based on the influence values ​​of the above multiple dimensions through the model, which realizes the automated dynamic quantitative evaluation of the influence of the person to be evaluated. It takes a short time and includes multiple dimensions that represent influence. The evaluation result is more objective and accurate.

[0150] In some examples of the present disclosure, the formula for calculating the performance value determined according to the performance measurement model may be:

[0151] P=μ′(α∑A+β∑B+γ∑C+δ∑D+ε∑E+θ∑F)+G;

[0152] Wherein, P is the performance value; AF are six different evaluation values; α-θ are the adjusted weights corresponding to the six different evaluation values; μ′ is the correction coefficient determined by the performance measurement model; G is the adjustment coefficient determined by the performance measurement model.

[0153] In some embodiments of the present disclosure, the performance determination method further includes: performing standardization processing on basic evaluation data of different numerical types of different persons to be evaluated. The standardization processing may include normalization processing and standard score standardization processing.

[0154] Specifically, the basic evaluation data of numerical type is determined as numerical evaluation data, and the attribute values ​​of the attribute value evaluation data of the same attribute name of different persons to be evaluated are unified into a unit. The attribute name may include: number of published articles, number of project participation, etc.

[0155] In the normalization process, the maximum and minimum values ​​of the attribute values ​​are determined, and the attribute values ​​currently being processed are normalized according to the attribute values ​​currently being processed and the maximum and minimum values. The formula may be:

[0156]

[0157] Among them, X is the attribute value currently being processed; Xmin and Xmax are the minimum and maximum values ​​of the attribute value respectively; Xnorm is the normalized attribute value, ranging from [0,1].

[0158] For example, suppose that in a scientific research system, the number of articles published by a research team ranges from 10 to 100. If a person to be evaluated has published 30 articles, then his normalized attribute value is:

[0159]

[0160] Thus, the number of published articles by different researchers can be compared on the same scale.

[0161] In the standard score standardization process, the data is converted to a standard normal distribution, which is suitable for standardizing indicators with significant distribution differences. For example, assuming that in the scientific research system, the mean score of the academic influence of researchers is 50, the standard deviation is 10, and a researcher scores 70, then his standard score is (70-50) / 10=20, which means that the influence of the researcher is 2 standard deviations higher than the average level. The standard score allows even if the impact scores of different research fields vary greatly, they can be converted to a unified standard.

[0162] In some embodiments of the present disclosure, the basic evaluation data includes output data, and the evaluation value also includes an output impact value. The output data includes one or more of project data, laboratory data, achievement data, achievement transformation data, and award data. Accordingly, the performance determination method further includes: determining the output impact value according to the regional level and regional grade of the generated data.

[0163] In some embodiments of the present disclosure, the performance determination method further includes: determining a person portrait of the person to be evaluated according to the evaluation value to highlight the characteristics of the person to be evaluated.

[0164] In some embodiments of the present disclosure, the performance determination method also includes: dynamically displaying evaluation values ​​and performance values ​​through a visualization platform, and providing personalized improvement suggestions based on performance values, to assist the person to be evaluated to adjust research directions and resource allocations in a timely manner.

[0165] The performance determination scheme can be implemented based on a cloud platform. The performance evaluation device is designed based on a cloud computing architecture to ensure elastic expansion and efficient storage of data processing capabilities, support cross-regional and cross-institutional scientific research cooperation and performance sharing, and improve management efficiency and synergy effects.

[0166] Figure 5 A schematic diagram of the architecture of a performance determination method provided in an embodiment of the present disclosure is shown in FIG. Figure 5 As shown, the performance determination system can call a processing module, an acquisition module, an input module, and a visualization module. The processing module can determine the evaluation value based on the basic evaluation data, and the processing module can include a first processing module and a second processing module; the acquisition module can capture the performance source data; the input module can dynamically determine the performance value based on the evaluation value, and identify outstanding evaluation values; the visualization module can display the dynamic change trend of the evaluation value and the performance value.

[0167] Figure 6 A schematic diagram of a performance determination method provided by an embodiment of the present disclosure is shown as follows: Figure 6 As shown, in this embodiment, the performance determination device periodically collects performance source data, and dynamically determines basic evaluation data based on the performance source data, and determines the performance value based on the basic evaluation data, so that the person to be evaluated can obtain the performance value and continue to improve. The improvement of the person to be evaluated affects the performance source data, so that the dynamically determined basic evaluation data changes, thereby realizing a closed-loop adjustment of the performance value.

[0168] The performance determination scheme provided by the disclosed embodiment improves the accuracy of evaluation. The application of multi-dimensional data fusion and intelligent models makes the scientific research performance evaluation more comprehensive and objective, effectively avoiding the limitations of single indicator evaluation, and constructing a more comprehensive, dynamic and intelligent performance determination scheme, realizing the transition from static evaluation to dynamic evaluation, and upgrading from single indicator to multi-dimensional comprehensive evaluation. It is timely, with real-time monitoring and dynamic feedback mechanisms, which help to identify problems in a timely manner and take measures. It optimizes resource allocation and provides a scientific basis for the rational allocation of scientific research funds and resources by accurately evaluating the performance of scientific researchers.

[0169] Figure 7 This is a schematic diagram of the structure of a performance determination device provided by an embodiment of the present disclosure. The device can be implemented by software and / or hardware, and the device can also be integrated in an electronic device. Figure 7 As shown, the performance determination device comprises:

[0170] The acquisition module 701 is used to acquire the performance source data of the person to be evaluated;

[0171] The preprocessing module 702 is used to perform structured processing and data cleaning processing on the performance source data with the person to be evaluated as an index item to obtain basic evaluation data corresponding to the person to be evaluated; wherein the basic evaluation data includes a paper citation topology map and network interaction information;

[0172] The first processing module 703 is used to identify the field of the person to be evaluated according to the paper citation topology map to obtain the cross-field influence value of the person to be evaluated;

[0173] The second processing module 704 is used to perform network impact analysis on the person to be evaluated according to the network interaction information to determine the interaction impact value of the person to be evaluated;

[0174] The input module 705 is used to input the evaluation value into the trained performance measurement model to obtain the performance value of the person to be evaluated; wherein the evaluation value includes the cross-domain impact value and the interactive impact value.

[0175] Optionally, the preprocessing module 702 includes:

[0176] An extraction submodule is used to extract names from the performance source data to obtain the names of the personnel to be evaluated;

[0177] A determination submodule, used to determine the performance attribute information corresponding to the personnel name in the performance source data; wherein the performance attribute information includes an attribute name and an attribute value;

[0178] The cleaning submodule is used to perform data cleaning processing on the performance attribute information to obtain the basic evaluation data.

[0179] Optionally, the cleaning submodule is used for:

[0180] Clustering the performance attribute information according to the attribute name to obtain multiple information groups;

[0181] Determine, among the multiple information groups, an information group including multiple attribute values ​​and having a similarity greater than a similarity threshold as an information group to be processed;

[0182] One of the performance attribute information in the information group to be processed is determined as the first attribute information, and the performance attribute information except the first attribute information in the information group to be processed is deleted.

[0183] Optionally, the cleaning submodule is used for:

[0184] In the performance attribute information, the performance attribute information whose sequence digit number of the attribute value is less than a digit threshold is determined as the second attribute information; wherein the digit threshold corresponds to the attribute name;

[0185] Determine a vacant position of an attribute value in the second attribute information, and determine a vacant element corresponding to the vacant position according to a coding structure relationship corresponding to an attribute name in the second attribute information;

[0186] Query the performance source data according to the vacancy element and the personnel name to obtain vacancy information corresponding to the vacancy element;

[0187] The attribute value in the second attribute information is completed according to the missing information and the coding structure relationship to obtain the basic evaluation data.

[0188] Optionally, the first processing module 703 is used to:

[0189] In the paper citation topology diagram, the paper at the previous level of the target paper is determined as the citing paper; wherein the target paper is a paper whose author includes the person to be evaluated;

[0190] In the paper citation topology map, the papers at the next level of the target paper are determined as cited papers;

[0191] The cross-field impact value is determined according to the number of the citing papers in the first field, the number of the cited papers in the second field, the preset citation weight, and the preset cited weight.

[0192] Optionally, the second processing module 704 is used to:

[0193] According to the network interaction information, a plurality of preset interaction types are counted to obtain a first statistical quantity corresponding to the plurality of interaction types of the person to be evaluated; wherein the interaction types include sharing types and comment types;

[0194] Determine the interaction value according to the first statistical quantity and the second statistical quantity; wherein the second statistical quantity is the quantity of target interaction information, and the target interaction information is network interaction information of the comment type whose number of characters is greater than a preset number threshold;

[0195] Determine an interactive position corresponding to the network interactive information and an interactive area where the interactive position is located;

[0196] Calculating the distance between the interaction position and the position of the person to be evaluated to obtain the interaction distance, and determining the interaction range value according to the interaction distance and the number of areas; wherein the number of areas is the number of the interaction areas;

[0197] The interaction impact value is determined according to the interaction amount value and the interaction range value.

[0198] Optionally, the device further comprises a training module for:

[0199] Obtaining sample data; wherein the sample data corresponds to the sample personnel one by one, and the sample data includes a plurality of evaluation sample values, a sample weight corresponding to each of the evaluation sample values, and a performance sample value corresponding to the plurality of evaluation sample values;

[0200] Inputting the evaluation sample value and the sample weight into an initial model to obtain a predicted sample value output by the initial model;

[0201] The initial model is trained according to the predicted sample values ​​and the performance sample values ​​to obtain the performance measurement model.

[0202] The performance determination device provided in the embodiments of the present disclosure can execute the performance determination method provided in any embodiment of the present disclosure, and has the corresponding functional modules and beneficial effects of the execution method.

[0203] Figure 8 The structure diagram of an electronic device provided by the embodiment of the present disclosure is shown in FIG. Figure 8 As shown, the electronic device 800 includes one or more processors 801 and a memory 802 .

[0204] The processor 801 may be a central processing unit (CPU) or other forms of processing units having performance determination capabilities and / or instruction execution capabilities, and may control other components in the electronic device 800 to perform desired functions.

[0205] The memory 802 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 801 may run the program instructions to implement the performance determination method of the embodiment of the present disclosure described above and / or other desired functions. Various contents such as input signals, signal components, noise components, etc. may also be stored in the computer-readable storage medium.

[0206] In one example, the electronic device 800 may further include: an input device 803 and an output device 804, which are interconnected via a bus system and / or other forms of connection mechanisms (not shown). In addition, the input device 803 may also include, for example, a keyboard, a mouse, etc. The output device 804 may output various information to the outside, including determined distance information, direction information, etc. The output device 804 may include, for example, a display, a speaker, a printer, and a communication network and a remote output device connected thereto, etc. Of course, for simplicity, Figure 8 Only some of the components related to the present disclosure in the electronic device 800 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 800 may also include any other appropriate components.

[0207] In addition to the above-mentioned method and device, the embodiment of the present disclosure may also be a computer program product, which includes computer program instructions, and when the computer program instructions are executed by a processor, the processor executes the performance determination method provided by the embodiment of the present disclosure.

[0208] The computer program product may be written in any combination of one or more programming languages ​​to write program code for performing the operations of the disclosed embodiments, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0209] In addition, the embodiment of the present disclosure may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the performance determination method provided by the embodiment of the present disclosure.

[0210] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or 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.

[0211] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the process, method, article or device including the elements. The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A performance determination method, characterized in that: include: Obtain performance source data of the person to be evaluated; Taking the person to be evaluated as an index item, the performance source data is structured and cleaned to obtain basic evaluation data corresponding to the person to be evaluated; wherein the basic evaluation data includes a paper citation topology map and network interaction information; Performing field identification on the person to be evaluated according to the paper citation topology map to obtain the cross-field influence value of the person to be evaluated; Performing a network impact analysis on the person to be evaluated according to the network interaction information to determine the interaction impact value of the person to be evaluated; The evaluation value is input into the trained performance measurement model to obtain the performance value of the person to be evaluated; wherein the evaluation value includes the cross-domain impact value and the interactive impact value.

2. The method according to claim 1, characterized in that Taking the person to be evaluated as an index item, the performance source data is structured and cleaned to obtain basic evaluation data corresponding to the person to be evaluated, including: Extracting names from the performance source data to obtain names of personnel corresponding to the personnel to be evaluated; Determine the performance attribute information corresponding to the personnel name in the performance source data; wherein the performance attribute information includes an attribute name and an attribute value; The performance attribute information is subjected to data cleaning processing to obtain the basic evaluation data.

3. The method according to claim 2, characterized in that The data cleaning process of the performance attribute information to obtain the basic evaluation data includes: Clustering the performance attribute information according to the attribute name to obtain multiple information groups; Determine, among the multiple information groups, an information group including multiple attribute values ​​and having a similarity greater than a similarity threshold as an information group to be processed; One of the performance attribute information in the information group to be processed is determined as the first attribute information, and the performance attribute information except the first attribute information in the information group to be processed is deleted.

4. The method according to claim 2, characterized in that: The data cleaning process of the performance attribute information to obtain the basic evaluation data includes: In the performance attribute information, the performance attribute information whose sequence digits of the attribute value are less than a digit threshold is determined as the second attribute information; wherein the digit threshold corresponds to the attribute name; Determine a vacant position of an attribute value in the second attribute information, and determine a vacant element corresponding to the vacant position according to a coding structure relationship corresponding to an attribute name in the second attribute information; Query and process the performance source data according to the vacancy element and the personnel name to obtain vacancy information corresponding to the vacancy element; The attribute value in the second attribute information is completed according to the missing information and the coding structure relationship to obtain the basic evaluation data.

5. The method according to claim 1, characterized in that: The step of identifying the field of the person to be evaluated according to the paper citation topology map to obtain the cross-field impact value of the person to be evaluated includes: In the paper citation topology diagram, the paper at the previous level of the target paper is determined as the citing paper; wherein the target paper is a paper whose author includes the person to be evaluated; In the paper citation topology map, the papers at the next level of the target paper are determined as cited papers; The cross-field impact value is determined according to the number of the citing papers in the first field, the number of the cited papers in the second field, the preset citation weight, and the preset cited weight.

6. The method according to claim 1, characterized in that The performing of network impact analysis on the person to be evaluated according to the network interaction information to determine the interaction impact value of the person to be evaluated includes: According to the network interaction information, a plurality of preset interaction types are counted to obtain a first statistical quantity corresponding to the plurality of interaction types of the person to be evaluated; wherein the interaction types include sharing types and comment types; Determine the interaction value according to the first statistical quantity and the second statistical quantity; wherein the second statistical quantity is the quantity of target interaction information, and the target interaction information is network interaction information of the comment type whose number of characters is greater than a preset number threshold; Determine an interactive position corresponding to the network interactive information and an interactive area where the interactive position is located; Calculate the distance between the interaction position and the position of the person to be evaluated to obtain the interaction distance, and determine the interaction range value according to the interaction distance and the number of areas; wherein the number of areas is the number of the interaction areas; The interaction impact value is determined according to the interaction amount value and the interaction range value.

7. The method according to claim 1, characterized in that The training process of the performance measurement model includes: Obtaining sample data; wherein the sample data corresponds to the sample personnel one by one, and the sample data includes a plurality of evaluation sample values, a sample weight corresponding to each of the evaluation sample values, and a performance sample value corresponding to the plurality of evaluation sample values; Inputting the evaluation sample value and the sample weight into an initial model to obtain a predicted sample value output by the initial model; The initial model is trained according to the predicted sample values ​​and the performance sample values ​​to obtain the performance measurement model.

8. A performance determination device, characterized in that: include: An acquisition module is used to obtain the performance source data of the person to be evaluated; A preprocessing module is used to perform structured processing and data cleaning processing on the performance source data with the person to be evaluated as an index item to obtain basic evaluation data corresponding to the person to be evaluated; wherein the basic evaluation data includes a paper citation topology map and network interaction information; A first processing module is used to identify the field of the person to be evaluated according to the paper citation topology map to obtain the cross-field influence value of the person to be evaluated; A second processing module is used to perform network impact analysis on the person to be evaluated according to the network interaction information to determine the interaction impact value of the person to be evaluated; An input module is used to input the evaluation value into the trained performance measurement model to obtain the performance value of the person to be evaluated; wherein the evaluation value includes the cross-domain impact value and the interactive impact value.

9. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing instructions executable by the processor; The processor is used to read the executable instructions from the memory and execute the instructions to implement the performance determination method described in any one of claims 1-7 above.

10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and the computer program is used to execute the performance determination method described in any one of claims 1-7.

Images