88 results about "Scoring rule" patented technology

The invention relates to a multi-competency intelligent scoring method and system, and belongs to the technical field of intelligent evaluation and talent evaluation. The method comprises the following steps: acquiring multi-source response data of a target evaluation object, executing evidence sufficiency and consistency analysis for each competency dimension, generating a judgment state identifier, and constructing a structural description; performing responsibility distribution on the evaluation evidence, mapping the evaluation evidence into a corresponding evidence model and generating a priority constraint relationship; performing time sequence consistency and stability analysis on the evidence model, and constructing a time sequence constraint rule to prohibit unconstrained score backtracking; and carrying out resolution processing on the continuous undetermined dimension, generating a scoring result through a preset risk scoring rule and a conservative determination strategy, and synchronously generating a scoring basis path and confidence information. According to the method, standard management and control of multi-source evidences and accurate constraint of the whole scoring process are realized, the scoring accuracy and traceability are effectively improved, and reliable support is provided for multi-scene talent evaluation decision.

The invention discloses an industry instruction fine tuning data set automatic generation method and system, and belongs to the technical field of artificial intelligence, S10: selecting industry standard document data, and analyzing the data into a plurality of paragraph texts as data sources according to chapters; s20, generating different types of questions for the paragraph text by using a large language model; s30, scoring the generated questions according to a scoring rule, and filtering the questions with scores lower than a preset threshold value; s40, the solvability and difficulty of the question are adjusted through a large language model, and the quality of the question is optimized, and S50, a first version answer of the question is generated by using the large language model, and the answer is optimized by using a hierarchical region optimization search algorithm; s60, constructing a sample pool through global and local selection methods, calculating a sample compression ratio and screening a data set; the method has the beneficial effects that through a highly automatic process and an optimization algorithm, the data set generation efficiency and quality are improved, the labeling cost is reduced, and support is provided for rapid deployment of a large language model in industry application.

The invention discloses a user intention identification method and device, computer equipment and a storage medium, belongs to the technical field of artificial intelligence, and is applied to an online system of an insurance service scene. According to the method, feature extraction is performed by using the user data, so that the limitation of screening intentional users singly depending on artificial experience is avoided, and the risk that high-intentional customers are missed in screening is effectively reduced. And secondly, intelligent evaluation is performed on the user intention through a pre-trained intention probability prediction model, and a prediction result is further optimized in combination with a probability calibration model, so that intention identification is more accurate and reliable. Then, according to the invention, quantitative scoring is carried out on the user intention by adopting a probability scoring rule, and the screening standard of the intention user is ensured to be uniform and quantifiable, so that the scientificity and the effect of a touch strategy are improved. Finally, high-intention users can be automatically and intelligently identified, so that an insurance company can more widely reach potential customers, the user experience is enhanced, and the customer conversion rate is improved.

The invention relates to the technical field of unmanned ship hunting, and discloses a multi-USV hunting method based on multi-factor scoring and adaptive task assignment, comprising the following steps: (1) establishing USV mathematical models including a kinematics model and a dynamics model; (2) establishing a multi-USV hunting task model; (3) establishing a multi-factor scoring rule: obtaining a distance factor score, an angle factor score and an energy consumption factor score, and obtaining a comprehensive score between the surrounding USV and the confronting USV according to each factor score; and (4) determining an adaptive task allocation strategy according to the comprehensive score between the hunting USV and the confrontation USV, and realizing multi-USV hunting. According to the method, the multiple confronting USVs are trapped to the designated target area to be trapped, the trapped tasks are distributed to each trapped USV in a self-adaptive mode, it is ensured that each confronting USV can be trapped by the trapped USV at the same time, the trapped USV can be trapped at the same time, trapped efficiency is improved, energy consumption needed for trapped USV is lower, and trapped voyage is shorter.

The invention relates to the technical field of education informatization, in particular to a multidisciplinary-oriented intelligent homework reviewing management platform and method, and the platform comprises the following steps: an image structure subdivision module, a behavior path recognition module, a rule execution monitoring module, an error cause structure labeling module, and a data label filing module. According to the method, stroke path analysis is combined with corner frequency and closed region detection, the handwritten character recognition precision is improved, the misjudgment rate of interdisciplinary symbol hybrid scenes is reduced, the structural expression ability of the problem solving process is enhanced by means of sequential construction of writing paths and logic nodes, and the problem solving efficiency is improved by combining parameter linkage of scoring frequency and scoring fluctuation. According to the method, the dynamic adaptive capacity of the scoring rule is optimized, the sequence comparison of a causal chain and knowledge nodes is adopted, the logic integrity of error cause positioning is improved, label clustering and distribution statistics are used, homework data archiving has the multi-dimensional screening and teaching backtracking capacity, and the collaborative efficiency of a homework reviewing system is promoted.

The invention discloses a claim dialogue intention recognition method and device, equipment and a storage medium, and belongs to the technical field of artificial intelligence. According to the method, the dialogue intention recognition model of the vector embedding layer, the multi-head self-attention layer, the feedforward network layer and the classification layer is introduced, so that the accuracy and robustness of intention recognition in the claim dialogue are effectively improved. Compared with the prior art, the method has the advantages that the text semantics are deeply represented by using the vector embedding layer, so that the understanding ability of the model to different expression modes is enhanced; a multi-head self-attention mechanism can capture key information in the context, and the perception ability of the model to the dialogue context is improved; the feed-forward network layer is used for realizing nonlinear combination of the features and is beneficial to mining implied domain features in the corpora; and the classification layer performs label matching and probability distribution calculation in combination with a preset scoring rule, so that efficient and accurate intention recognition is realized. According to the scheme, the user intention recognition effect is improved, and the requirement for high real-time performance in the actual claim settlement service is met.

The invention belongs to the technical field of artificial intelligence, and relates to an evaluation method, system and device of an intelligent cabin scene multi-modal model and a medium, and the method comprises the steps: 1) a user questions a vehicle-mounted image to obtain questions, and a multi-modal large language model is used to classify each question according to a pre-designed question label system; 2) inputting the vehicle-mounted image and the question into an intelligent cabin scene multi-modal model to generate a question answer; 3) determining evaluation dimensions and importance scores of the dimensions based on the question types; 4) using a chain reasoning mechanism to construct a scoring process and formulating a scoring rule in advance; and 5) inputting the vehicle-mounted image, the question, a reference answer formulated in advance and the generated question answer into a multi-modal large language model, and performing comprehensive scoring according to the evaluation dimension and the importance score of each dimension based on a scoring process and a scoring rule. According to the method, comprehensive, reliable and automatic evaluation of the intelligent cabin scene multi-modal model can be realized.

The invention discloses a heterogeneous workflow task scheduling method and device and a medium. The method comprises the steps of obtaining original data of sub-tasks under multiple feature dimensions according to a preset period; according to a preset scoring rule of each feature dimension, converting the original data into a quantitative score under the corresponding feature dimension; based on the original data, adopting an entropy weight method and an analytic hierarchy process to calculate the weight of the subtask under each feature dimension; calculating a total feature score of the sub-tasks based on the quantitative scores and weights of the sub-tasks under each feature dimension; and obtaining a scheduling scheme of the heterogeneous workflow task through the total feature score of each subtask. According to the method, the timeliness and adaptability of the scheduling decision are improved.

The invention discloses a unit state analysis early warning method applied to a centralized control system, and the method comprises the steps: obtaining original data of a data interface layer, carrying out the data preprocessing and parameter reconstruction of the original data after statistical classification, and calling an equipment three-dimensional model in a model layer to carry out digital twin modeling, the preprocessed data, the reconstructed feature parameters and the digital twinborn model are correspondingly output; multi-dimensional index analysis is carried out on the preprocessed data and the reconstructed characteristic parameters by calling an analysis model and a diagnosis model in the model layer, an independent scoring rule is constructed, and the instantaneous state and the risk trend of the centralized control system are comprehensively evaluated; and by calling an early warning model and an evaluation model in the model layer, dynamically analyzing unit overall risk output early warning information according to the multi-dimensional analysis result, and displaying a dynamic simulation result and an early warning result of a unit state through a digital twin model. The objective degree and standardization problems of centralized control system evaluation are effectively solved.

The invention provides an intelligent reviewing method, and belongs to the technical field of computers. In the method, firstly, a target image is processed to obtain a target answer text corresponding to a student answer, secondly, in response to a first input of a first user, a scoring detailed rule corresponding to a target test question is obtained, and the scoring detailed rule can only comprise a first scoring condition and can also comprise the first scoring condition and a second scoring condition at the same time; the second scoring condition is a logical scoring condition, an integrity scoring condition and / or a relevance scoring condition; and finally, based on the scoring detailed rule, scoring the target answer text to obtain a first score corresponding to the student answer, thereby realizing structuralization, standardization and automation of the subjective question reviewing process. According to the method and the device, through a mode of combining image recognition, rule configuration and semantic scoring, not only is the efficiency of reviewing subjective questions significantly improved, but also the dependence of the scoring result on subjective judgment of teachers is effectively reduced.

The invention discloses an enterprise low-carbon evaluation method and system based on retrieval enhancement generation, and the method comprises the steps: carrying out the format recognition and table semantic reconstruction of an unstructured report through a preprocessing module, carrying out the data cleaning and unit standardization in combination with a carbon accounting standard, and constructing a vector index database; in the retrieval stage, the evaluation indexes are converted into query vectors, and related evidences are accurately positioned through mixed retrieval; constructing an enhanced cue word containing unit standardization logic to guide large model reasoning, and generating a structured intermediate result; an anti-fact consistency verification mechanism is introduced, the generated content and the original context are compared, and a low-confidence result is removed; and finally, carrying out weighted calculation in combination with a scoring rule and evidence confidence to obtain an enterprise low-carbon level score. According to the method, high-precision and traceable automatic low-carbon evaluation is realized by fusing table semantic perception based on physical coordinates, unit normalized reasoning of an embedded industry standard and an anti-fact consistency verification mechanism.

The invention discloses a batch composition scoring system, method and device based on multi-dimensional quantization and a storage medium. The system comprises an input processing module, a scoring module and a scoring module, wherein the input processing module extracts basic information of each composition from a batch composition picture based on multiple threads; the large model module extracts composition text feature information from the corresponding composition pictures according to the basic information of each composition based on the message queue, analyzes the composition text feature information according to a preset special constraint rule, a penalty factor rule and a basic score rule, and outputs a multi-dimensional score of each composition picture; and the multi-dimensional scoring engine performs multi-dimensional quantitative scoring on each composition picture through a multi-dimensional quantitative scoring formula according to the multi-dimensional scores. According to the method, the multi-dimensional quantification rule is fused to the large model, the objectivity of traditional rule quantification is reserved, a dimension self-defining mechanism is introduced, a user is allowed to expand the scoring dimension and adjust the quantification standard or the weight proportion according to the requirement, and fair and accurate automatic batch scoring is achieved.

The embodiment of the invention relates to a document score generation method and system based on a large model, and the method comprises the steps: generating a score dimension list and a score constraint condition according to a user-defined score statement, and defining a score logic according to the score dimension list and the score constraint condition; constructing a structured scoring rule, and inputting the structured scoring rule into the plurality of basic large models; analyzing the submitted document by utilizing the basic large model to obtain a plurality of semantic unit vectors, performing similarity matching on the plurality of semantic unit vectors and each scoring dimension in the structured scoring rule to screen out a target semantic unit vector, and according to the target semantic unit vector and the structured scoring rule, obtaining the target semantic unit vector; generating a preliminary scoring result of each scoring dimension; and inputting the preliminary scoring result into a high-level large model, and performing weighted aggregation on the preliminary scoring result of each scoring dimension generated by each basic large model by using the high-level large model so as to generate a document score with higher accuracy.

The present invention discloses a new urban construction integration project evaluation method and system based on multiple dimensions, which relates to the field of urban construction evaluation and multidimensional data modeling and analysis technology, including collecting the original value and score of the same indicator in multiple new urban construction integration projects, and constructing an indicator scoring point set. The score values ​​in the indicator scoring point set are normalized, and the influence of outliers is eliminated. Multi-class scoring function models are fitted and the goodness of fit is calculated, and the standard scoring function form is determined according to the goodness of fit. A scoring function template is established based on the standard scoring function form and function parameters, and stored in a scoring function template library. The method of the present invention realizes a scientific, unified, and reusable quantitative evaluation mechanism for new urban construction integration projects, effectively overcoming the problems of strong subjectivity of scoring rules, difficulty in migration, and inconsistent evaluation standards in existing methods, and improving the objectivity, flexibility and intelligence of the evaluation system.

The invention discloses a structure calculation index quantification-based technical standard intelligent scoring system and method, and the system comprises a construction module which is used for obtaining a design file related to a building, constructing a building structure model based on the design file, and constructing a construction model based on the building structure model; calculating the constructed initial model of the building structure to obtain a calculation index of the building structure; the first calculation module is used for acquiring preset scoring rule information corresponding to the building structure and performing scoring calculation on the calculation indexes of the building structure according to the preset scoring rule information to obtain scoring data corresponding to each structure calculation index; the distribution module is used for distributing corresponding weights for the calculation indexes of the building structure according to the building type in the design file; and the second calculation module is used for calculating the score data of each structure calculation index according to the corresponding weight to obtain the total score data of the building structure.

The present invention relates to the technical field of duty efficiency, and specifically, to a comprehensive evaluation method for duty efficiency based on artificial intelligence. It includes the following steps: establishing specific goals for efficiency evaluation and determining key indicators for evaluating duty efficiency; collecting duty data related to the evaluation according to the determined indicators; using the Extreme Gradient Boosting model to analyze the duty data and evaluating the duty efficiency based on the duty scoring rules. The design of the present invention can effectively identify the key factors affecting duty efficiency by using advanced machine learning algorithms such as the Extreme Gradient Boosting model and combining the feature selection process, so as to provide more accurate and fair evaluation results. This not only helps to more truly reflect the work performance of duty personnel, but also promotes the transparency and scientificity of performance management. Considering the characteristics of duty data changing with time and environment, by introducing a dynamic adjustment mechanism, the model can better adapt to the complex situations in actual work.

The invention relates to the field of artificial intelligence, and particularly provides a 3D model scoring method and system based on a neural network model, a terminal and a medium, and the method comprises the steps: employing a pre-trained neural network model to generate standard fusion features for a 3D model uploaded by a teacher user, and employing a local geometric feature extraction model to extract local geometric features, extracting global topological relation characteristics by using a global topological relation modeling model, and performing weighted fusion on the local geometric characteristics and the global topological relation characteristics by using an attention mechanism to obtain fusion characteristics; using a pre-trained neural network model to generate a to-be-scored fusion feature for the 3D model uploaded by the student user; comparing the fusion feature to be scored with a standard fusion feature, and obtaining a 3D model score based on a pre-configured scoring rule according to a comparison result; and feeding back the 3D model score of the student user to the teacher user. According to the invention, two-channel feature extraction of CNN and GNN is combined, so that the scoring efficiency and accuracy are improved.

The invention provides a virtual practical training scoring method and system for the whole process of ship power operation, and belongs to the technical field of virtual practical training evaluation and assessment. Comprising the following steps: S1, dividing a virtual practical training whole process of ship power device operation into different links, and enabling a student to reserve a corresponding virtual practical training after completing a preposed theoretical course of one link; s2, respectively constructing scoring content based on a student virtual practical training operation process for each link, and drawing up a specific scoring rule; s3, performing virtual practical training on the successfully reserved trainees in sequence, obtaining virtual practical training scores according to a scoring rule, and sending different suggestions to the trainees according to the score range of the virtual practical training scores; and S4, extracting a link of error-prone operation of the current semester, constructing a multi-dimensional screening parameter, searching associated knowledge, and reconstructing the content of the preposed theoretical course, thereby reducing faults during fault identification and fault repair operation errors after fault identification.

The present invention belongs to the technical field of edge computing, and specifically relates to an adaptive risk-resistant Kubernetes cluster scheduling method based on a penalty factor. The present invention resists sudden high-load computing tasks by reserving a certain amount of computing resources and adaptively changing the reserved computing resource pool according to the state of the cluster; the present invention improves the preference algorithm in the default scheduling algorithm of Kubernetes, introduces the mean and standard deviation to evaluate the cluster resource state, and adds corresponding penalty factors to balance the resource states in different dimensions and optimize the scoring rules; at the same time, an influence weight is introduced as a supplement to the scoring rules. When the resource states and scoring scores of two Worker nodes are all the same, the historical scores obtained by this node are incorporated into the scoring rules to preferentially select a more suitable Worker node for this load task, effectively ensuring the load balance of the cluster.

This invention belongs to the field of artificial intelligence technology and discloses a multi-tool collaborative LLM intelligent agent analysis method and system for high-tech enterprise certification. The method includes identifying the assessment intent for high-tech enterprise certification, identifying data-gathering and inference-type tools, and constructing a binary classification directed dependency graph; asynchronously and in parallel calling tools for business registration, intellectual property, tax and finance, judicial risk, and R&D expense aggregation according to predecessor dependencies; performing format normalization, field mapping, subject alignment, conflict retention, deduplication, dimension aggregation, and scoring caliber mapping on the returned data; performing cross-dimensional cross-validation before comprehensive inference, and inputting the unified intermediate dataset into the LLM to perform a single comprehensive inference. This scheme can reduce serial waiting and invalid inference calls, and reduce the impact of subject confusion, caliber conflicts, and misuse of scoring rules on the certification conclusion.

The present disclosure describes risk control techniques. The techniques comprise calculating risk scores and determining a risk degree, wherein the calculating risk scores includes performing calculation processing on bury point data of a user according to a predetermined risk scoring rule to acquire the risk scores; and the determining a risk degree includes receiving a service request from the user, determining a corresponding risk assessment rule and a corresponding predetermined threshold based on the service request, determining an assessment score by searching for a risk score based on the corresponding risk assessment rule from the risk scores, and determining a risk assessment result by comparing the assessment score with the corresponding predetermined threshold. The procedures of calculating risk scores and determining a risk degree are performed separately, such that the high-speed response of a risk control service is ensured.

The invention provides a geographic space transformation evaluation method and device, equipment and a storage medium. The method comprises the steps that current to-be-evaluated geographic space information is acquired; extracting an evaluation factor for evaluation from the geographic space information; scoring the evaluation factor according to a pre-formulated scoring rule to obtain a scoring result; and obtaining an evaluation result of the geographic space transformation based on the scoring result. According to the method, the evaluation factors of the geographic space are scored through the pre-formulated scoring rule to finally obtain the quantified scoring result, the current geographic space transformation potential is evaluated based on the quantified evaluation result, automatic evaluation is achieved through the quantified evaluation mode, the method does not need to completely depend on manual evaluation, and the evaluation efficiency is improved. Therefore, the efficiency of geographic space transformation evaluation is improved.

The invention provides a multi-modal training method and device based on intelligent quality inspection, electronic equipment and a storage medium. The method comprises the following steps: acquiring an original dialogue record of an employee to be subjected to quality inspection and a user, and a clue conversion rate of the employee to be subjected to quality inspection; performing text conversion and cleaning on the original dialogue recording to obtain recording text data; performing quality inspection scoring on the recorded text data according to a preset scoring rule to obtain a multi-dimensional score, the multi-dimensional score being used for representing the invitation ability of the employee to be subjected to quality inspection; and under the condition that the clue conversion rate and the multi-dimensional score meet quality inspection conditions, generating quality inspection information data corresponding to the staff to be subjected to quality inspection. Through a preset scoring rule, multi-dimensional intelligent scoring is performed on the call data by using a large model, the comprehensive ability of an inviter is comprehensively evaluated, and the scoring rule can be freely and flexibly adjusted, so that a specific rule can be set according to requirements; the actual combat ability of the employees is further enhanced through the training module, and the training effect and the invitation conversion rate are improved.

The embodiment of the invention provides a quantifiable fall risk determination method, electronic equipment, a storage medium and a product, and relates to the technical field of data processing, and the method comprises the steps: obtaining a preliminary screening result of a fall risk of a target object, and obtaining a body function evaluation parameter, collected by timing equipment, of the target object; based on a preset quantitative scoring rule, calculating a body function scoring result of the target object; obtaining maximum angle range data, limb function quantitative data and gravity center displacement quantitative data of a specific body part of the target object in a preset motion test action; based on a preset weight coefficient and a quantitative fusion model, obtaining a quantifiable fall risk value of the target object; and determining the fall risk level of the target object according to a preset risk level division rule and the quantifiable fall risk value of the target object. The embodiment of the invention is used for providing a technical scheme capable of observably and accurately quantifying the falling risk of the old people.

The application provides an enterprise credit intelligent analysis method and system fusing a knowledge graph and a dialogue engine, and relates to the technical field of enterprise credit risk analysis.The application firstly acquires multi-source credit records of a target enterprise and associates source information fields to construct an enterprise credit knowledge graph; analyzes a credit query request to obtain a credit conclusion unit and a graph query condition; searches a candidate fact set and forms an evidence chain data package; groups according to a preset conflict rule and calculates a confidence score based on a preset scoring rule to determine a main conclusion and an alternative conclusion and generate a conflict explanation, which is output to a dialogue engine to generate a dialogue reply.The application associates the credit conclusion with the corresponding evidence chain, introduces a conflict grouping and confidence sorting mode, and improves the structured expression ability and conclusion reliability of the enterprise credit analysis result.

The invention relates to the technical field of data management, and discloses an enterprise finance and taxation data management method and system based on big data, and the method comprises the steps: building an enterprise finance and taxation data middle platform, obtaining data, carrying out the preprocessing of the obtained data, and numbering the data to form a finance data set, a taxation data set, a business data set, and a market data set. Calculating a financial health predicted value Cjk, a tax compliance predicted value Shg and a business stability predicted value Ywd by combining an algorithm calculation formula, constructing a risk scoring rule, and performing risk assessment and enterprise risk level assessment according to the financial health predicted value Cjk, the tax compliance predicted value Shg and the business stability predicted value Ywd. According to the method and the system, the risk assessment and enterprise risk level assessment results are output, comprehensive and targeted assessment is carried out on the enterprise risk from different dimensions, the assessment efficiency is high, the assessment effect is accurate, and accurate monitoring and dynamic early warning of enterprise finance and tax data are realized.

The application provides an industrial mold structure parameter optimization method and system based on multi-source data, relates to the field of industrial mold design, and has the technical scheme as follows: original parameters are extracted from multi-source mold data and converted into parameter records to ensure the consistency of data sources; a local object set is obtained through extension and a correlation between structure objects is established to determine a parameter allowable range and lay a foundation for subsequent optimization; when a parting boundary is adjusted, the alignment deviation of a guide part, the fitting length of a mold locking fitting boundary, and the connection state of an exhaust passage can be calculated according to the correlation to realize linkage analysis; a candidate scheme is generated by adjusting parameters within the parameter allowable range under the condition of meeting a preset hard constraint condition, a target scheme is determined through a preset scoring rule, and intelligent optimization is realized; finally, the landing point of the target scheme in multi-source mold data is checked reversely and written back synchronously to ensure the consistency of data and the executability of the scheme.

The invention discloses a normal distribution-based standard quantitative score optimization system, method and device and a storage medium, and the system comprises the steps: when a large model module receives batch composition pictures transmitted by an input processing module, carrying out the multi-dimensional feature analysis of each composition picture, and outputting the multi-dimensional score of each composition picture; a multi-dimensional scoring engine calculates a multi-dimensional quantitative initial total score of each composition picture according to the multi-dimensional scores; and the score optimization module performs composition ranking according to a preset ranking rule according to the multi-dimensional quantized initial total score, and optimizes the multi-dimensional quantized initial total score of each composition picture according to a composition ranking result based on normal distribution. According to the method, through a coupling mechanism of AI ranking generation and normal distribution scoring, the problem of scoring stiffness caused by excessive dependence on a fixed scoring rule in a traditional system is solved, and fair and flexible automatic scoring conforming to an education evaluation rule is realized.

The invention relates to a universal finance and accounting simulation experiment teaching method, which is established on the basis of the existing enterprise-level finance software system, is based on the application of finance and accounting examination in college teaching, and enables teachers to rapidly copy basic accounts for finance and accounting students in the actual teaching process of finance and accounting examination. Understanding of students on a finance and accounting review software system and knowledge points is understood, scoring is carried out according to rules by extracting operation data of the students in a finance system and submitted finance data, and scoring points and scoring rules can be prefabricated in advance by professional teachers; and the learning understanding of the financial and accounting students on each practical training course of financial and accounting examination can be well and comprehensively evaluated. According to the invention, a multi-dimensional scoring system is constructed, the experiment result of the student is evaluated, multiple aspects such as operation standardization, data entry accuracy and experiment report integrity in the experiment process are also included, and the comprehensive ability of the student is comprehensively reflected.

The elevator operation risk scoring method based on time sequence anomaly detection and correlation analysis comprises the steps that multi-source data of the Internet of Things, the Internet and the like of an elevator are obtained, a plurality of initial risk indexes are constructed, and good and bad samples are divided by calculating risk initial values; calculating the correlation between each index and the good and bad attributes of the sample, and adopting chi-square test and V value quantification and screening out effective risk indexes; determining weights and scores of the indexes in multiple dimensions such as identity features and operation capability according to the correlation, forming a scoring rule, and calculating the score of each dimension; and carrying out weighted summation on the dimension scores to obtain a total risk score, and dividing the elevators into different risk grades according to the total elevator total score distribution. According to the elevator risk assessment method, objective, dynamic and efficient assessment of elevator risks is achieved through a data-driven full-process automatic model, and a scientific basis is provided for precise supervision and preventive maintenance.