Professional technology maturity evaluation method based on artificial intelligence

Abstract

The invention discloses an artificial intelligence-based professional technology maturity evaluation method, which comprises the following steps of: constructing a dynamic knowledge base by crawling technical literatures, patents and industry reports, and performing cross-technical field semantic embedding in combination with an input technical text to generate semantic representation; extracting multi-scale features based on the dynamic knowledge base and semantic representation, and generating a comprehensive feature set; constructing a technical dependency graph by using the comprehensive feature set and a dynamic knowledge base, planning a reasoning path through reinforcement learning, evaluating the innovativeness, feasibility and commercial value of a professional technology, and generating an evaluation score; calculating a comprehensive score by adopting a cross-technical field weight of self-adaptive threshold adjustment, and mapping the comprehensive score into a maturity level; the evaluation result is optimized by integrating user feedback through online learning; according to the method, the problems of static knowledge base lagging, cross-domain term ambiguity and inference path insufficiency are solved, and the real-time performance, precision and specialty of evaluation are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence and natural language processing, and particularly relates to a method for evaluating the maturity of professional technologies based on artificial intelligence. Background Art

[0002] In recent years, the rapid development of artificial intelligence technology has promoted the innovation of technology maturity evaluation methods. Traditional technology maturity evaluations, such as the Technology Readiness Level (TRL) framework, are widely used in fields such as aerospace, energy, and national defense. They mainly evaluate the maturity stage of a technology from concept to commercialization through expert assessment. With the progress of big data and natural language processing technologies, artificial intelligence-based automated evaluation methods have gradually emerged. These automated evaluation methods mainly use pre-trained language models to analyze some literature, patents, and technical reports, extract key technical features, and perform reasoning in combination with a knowledge base to achieve quantitative evaluation of technologies. They are significantly superior to traditional manual methods in terms of efficiency and coverage, and are widely used in fields such as technology challenge announcements, R & D management, and patent examination.

[0003] However, there are still some deficiencies in the current artificial intelligence-based automated evaluation methods. First, existing methods mostly rely on static knowledge bases and are difficult to integrate the latest technical literature and patent data, resulting in evaluation results lagging behind the technological frontier. Second, in the evaluation of cross-domain technologies, existing methods have insufficient handling of semantic ambiguities of multidisciplinary terms and are difficult to accurately capture the characteristics of emerging technologies, thus affecting the accuracy and universality of the evaluation. Third, the reasoning logic of existing methods lacks dynamic analysis of the technology evolution path and is difficult to comprehensively reflect the maturity differences of technical solutions. These deficiencies directly limit the professionalism and reference value of the evaluation results.

[0004] CN115600893A discloses an intelligent evaluation method and system for the navigation safety management of large shipping hubs. Its solution mainly constructs an evaluation index system for the current situation of navigation safety management. By introducing a safety management maturity model, weights are assigned to the established evaluation indicators based on membership analysis and reliability and validity tests, and the correlation degree between each indicator and the current situation level of the navigation hub safety management is weighed. An expert evaluation module including self-evaluation, peer evaluation, and professional evaluation is constructed, and the optimal ratio of self-evaluation, peer evaluation, and professional evaluation is calculated. Then, quantitative calculations are performed on the determined evaluation index system, index weights, and evaluation index levels. Finally, visualization tools such as radar charts and bar charts are used to display the evaluation results, and improvement suggestions and directions are matched according to the evaluation results. In this solution, although the grade division of the maturity evaluation is mentioned, the relationship with the dynamic knowledge base is not involved, and the problem of evaluation results lagging behind the technological frontier cannot be solved.

[0005] CN117993375A discloses a method and system for cross - field application of patent technologies. Its solution mainly extracts technical elements based on patent documents by using text mining and deep - learning methods to generate a summary of technical features, evaluates domain adaptability by using statistical analysis and pattern - recognition methods, constructs a virtual prototype by using 3D modeling and computational fluid dynamics methods, optimizes the technical solution by applying simulation software and iterative design methods, and conducts cross - field application analysis by using knowledge graphs and association - rule mining. This solution provides in - depth analysis for market demand and technical adaptability through text analysis and simulation software, but it lacks dynamic analysis of the technical evolution path and is difficult to comprehensively reflect the maturity differences of technical solutions. Summary of the Invention

[0006] The purpose of this part is to outline some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Simplifications or omissions may be made in this part, as well as in the abstract and title of the specification of this application, to avoid obscuring the purpose of this part, the abstract, and the title. However, such simplifications or omissions shall not be used to limit the scope of the present invention.

[0007] In view of the above - mentioned existing problems, the present invention is proposed. Therefore, the present invention provides an artificial - intelligence - based professional - technology maturity assessment method to solve the problems raised in the background technology.

[0008] To solve the above - mentioned technical problems, the present invention provides the following technical solution: An artificial - intelligence - based professional - technology maturity assessment method, including: Obtain technical literature, patents, and industry reports to construct a dynamic knowledge base, perform cross - technical - field semantic embedding by inputting technical texts in combination with the dynamic knowledge base, and generate semantic representations; Extract multi - scale features from the input technical texts according to the dynamic knowledge base and semantic representations to generate a comprehensive feature set; Construct a technology - dependence graph by using the comprehensive feature set and the dynamic knowledge base, generate an evaluation score for professional technology, calculate a comprehensive score based on the evaluation score by using cross - technical - field weight calculation with adaptive threshold adjustment, map the comprehensive score to a professional - technology maturity level, and generate a professional - maturity assessment result; Optimize the evaluation result by incorporating user feedback through online learning to improve the professionalism of the professional - maturity assessment result.

[0009] As a preferred solution of the artificial - intelligence - based professional - technology maturity assessment method of the present invention, among them: Obtaining technical literature, patents, and industry reports to construct a dynamic knowledge base includes: Use a crawler tool to obtain technical literature, patents, and industry reports from open databases to construct a dynamic knowledge base; By extracting the keywords of the input technical text and each piece of crawled data and applying time weighting, the technical content associated with the keywords of the input technical text is screened out; Merge the screened technical content with the knowledge base to update the knowledge base.

[0010] As a preferred solution of the method for evaluating the maturity of professional technology based on artificial intelligence according to the present invention, wherein: through the input text content and the dynamic knowledge base, semantic embedding across technical fields is performed to generate semantic representations, including: Adopt a Transformer model for multi-task learning to generate semantic representations adapted to cross-technical fields.

[0011] As a preferred solution of the method for evaluating the maturity of professional technology based on artificial intelligence according to the present invention, wherein: according to the dynamic knowledge base and semantic representations, multi-scale features are extracted from the input technical text to generate a comprehensive feature set, including: Perform sentence splitting, word segmentation, and dependency syntactic analysis on the input technical text to generate a syntactic structure representation; Based on the syntactic structure representation, the global technical goal and sub-technical goals are respectively extracted to generate a feature vector representing the overall function of the technology and a modular sub-technical feature set; Fuse the feature vector of the overall function of the technology and the sub-technical feature set to generate a comprehensive feature set.

[0012] As a preferred solution of the method for evaluating the maturity of professional technology based on artificial intelligence according to the present invention, wherein: using the comprehensive feature set and the dynamic knowledge base to construct a technology dependency graph to generate an evaluation score for professional technology, including: Represent the nodes in the technology dependency graph as sub-technologies, and the edges as the dependency relationships between sub-technologies; Plan the inference path of the technology dependency graph through reinforcement learning, and evaluate the innovation, feasibility, and commercial value of professional technology to generate an evaluation score for professional technology.

[0013] As a preferred solution of the method for evaluating the maturity of professional technology based on artificial intelligence according to the present invention, wherein: evaluating the innovation, feasibility, and commercial value of professional technology, including: Evaluate the innovation of professional technology by comparing the semantic similarity between sub-technologies and the technical content in the dynamic knowledge base; Evaluate the feasibility of professional technology by matching the currently available resources in the dynamic knowledge base with the required resources in the sub-technical feature set and considering the adaptability of the sub-technical environment; Evaluate the commercial value of professional technology by estimating the matching degree of the potential market size and application scenarios.

[0014] As a preferred solution of the artificial intelligence-based professional technology maturity assessment method of the present invention, wherein: based on the evaluation score, a comprehensive score is calculated using cross-technical field weights with adaptive threshold adjustment, and the comprehensive score is mapped to a professional technology maturity level to generate a professional maturity assessment result, including: Based on the evaluation score, cross-technical field weights are determined through reinforcement learning, and a comprehensive score is calculated; Adaptive threshold adjustment is performed on the cross-technical field weights using the grading accuracy based on historical technology evaluation data; The comprehensively scored adjusted by the threshold is mapped to a professional technology maturity level to obtain a professional maturity assessment result.

[0015] As a preferred solution of the artificial intelligence-based professional technology maturity assessment method of the present invention, wherein: the professional technology maturity levels include the start-up stage, the bubble stage, the trough stage, the climbing stage, and the maturity stage.

[0016] As a preferred solution of the artificial intelligence-based professional technology maturity assessment method of the present invention, wherein: the evaluation result is optimized by integrating user feedback through online learning, including: Online learning adjusts the cross-technical field weights based on the difference between user feedback and the professional maturity assessment result.

[0017] Compared with the prior art, the beneficial effects of the invention are: 1. By crawling technical literature, patents, and industry reports to construct a dynamic knowledge base and updating the content of the knowledge base with time weighting, this method can timely reflect the latest trends in technology, overcome the deficiency of the static knowledge base leading to lagging evaluation results, and ensure the real-time nature of the evaluation results; 2. The Transformer model of multi-task learning is used for cross-domain semantic embedding, combined with multi-scale feature extraction, to generate a global feature vector and a sub-technology feature set, effectively solving the problems of cross-disciplinary technical term ambiguity and insufficient feature capture, and providing reliable support for emerging technology evaluation; 3. By using reinforcement learning to plan the reasoning path of the technology dependence graph, realizing the dynamic evaluation of innovation, feasibility, and commercial value, and integrating user feedback optimization through an online learning model, not only improves the professionalism and reasoning depth of the evaluation results, but also can dynamically adjust the weights according to the user feedback results, enhancing the reference value of the evaluation results and the user experience. Description of the Drawings

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings. Among them: Figure 1 It is the overall flowchart of the professional technology maturity assessment method based on artificial intelligence according to an embodiment of the present invention. Specific embodiments

[0019] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0020] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0021] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0022] The present invention is described in detail in conjunction with schematic diagrams. When detailing the embodiments of the present invention, for the convenience of explanation, the cross-sectional views showing the device structure will be enlarged locally in a non-general proportion, and the schematic diagrams are only examples and should not limit the scope of protection of the present invention herein. In addition, in actual production, three-dimensional spatial dimensions including length, width, and depth should be included.

[0023] At the same time, in the description of the present invention, it should be noted that the orientation or positional relationships indicated by terms such as "upper, lower, inner, and outer" are based on the orientation or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, the terms "first, second, or third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0024] Unless otherwise clearly specified and defined in the present invention, the terms "installation, connection, and coupling" shall be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may also be a mechanical connection, an electrical connection, or a direct connection, or may be indirectly connected through an intermediate medium, or may be the communication inside two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0025] Embodiment 1 Referring to Figure 1 , which is the first embodiment of the present invention. This embodiment provides a method for evaluating the maturity of professional technologies based on artificial intelligence, including: S1. Obtain technical literature, patents, and industry reports to construct a dynamic knowledge base, and perform cross-technical field semantic embedding through the input technical text combined with the dynamic knowledge base to generate semantic representations; Furthermore, use a crawler tool to obtain technical literature, patents, and industry reports from open databases to construct a dynamic knowledge base; Specifically, through the Scrapy framework combined with Selenium, crawl open (free) databases (such as the Open Data Platform of the Chinese Academy of Sciences, CNIPA, and the National Bureau of Statistics of China); It should be explained that Selenium is a third-party library of Python, which operates the browser through the externally provided interfaces to enable the browser to complete automated operations; Specifically, the time range for crawling data needs to be set before crawling data; preferentially crawl data in the past 1 month (based on the current date), and gradually expand the months to ensure the timeliness of the crawled data; for example, based on the current date of April 27, 2025, crawl the data in the past month of this date, that is, from March 27, 2025 to April 27, 2025. Then, as the number of months iterates, assuming the current date comes to May 27, 2025, then the data crawled in the past month of this date becomes from April 27 to May 27, 2025; It should be noted that in practical applications, due to the update speed of the data itself and the large amount of crawled data, by gradually crawling data month by month, not only can the timeliness of the data be ensured, but also sufficient time for data storage can be ensured; Furthermore, use natural language processing tools (TF-IDF) to extract the keywords of the input technical text and the keywords of the crawled data at the same time, assign time weights to each crawled data (where the data closer to the current date has a higher weight, and the data farther from the current date has a lower weight), and use time weighting to screen out the technical content associated with the keywords of the input technical text; Specifically, based on the difference between the publication date of the crawled data and the current date, an exponential decay function is used to assign a time weight to each piece of crawled data:

[0026] Among them, represents the time weight of the crawled data ; represents the difference (in days) between the publication time of the crawled data and the current date; Specifically, the keywords of the input technical text and the crawled data are converted into semantic representations, and their correlation is calculated through cosine similarity:

[0027] Among them, represents the correlation between the technical text and the keywords of the crawled data, K represents the set of keywords of the input technical text, represents the set of keywords of the crawled data; represents the semantic vector of the keyword k of the technical text; represents the keyword of the crawled data ; represents the cosine similarity of the two keyword vectors; respectively represent the sizes of the sets of keywords of the technical text and the crawled data; Specifically, combining the correlation between the technical text and the keywords of the crawled data with the time weight of the crawled data, calculate the comprehensive score of each piece of crawled data :

[0028] It should be noted that sorting the comprehensive scores of each piece of crawled data in descending order can filter out the crawled data with similar days and that conform to the keywords of the technical text, so as to obtain the technical content associated with the keywords of the input technical text; Furthermore, merge the filtered technical content with the knowledge base to update the knowledge base; It should be noted that the merge and update operations are aimed at optimizing the query performance of the dynamic knowledge base and removing duplicates (eliminating redundant or low-correlation content); for example, if the knowledge base already contains a piece of technical content about "deep learning", then the newly crawled similar technical content will update the meta-information (such as publication date, citation times) of the existing record, rather than adding it repeatedly; Further, use the Transformer model of multi-task learning to generate semantic representations adapted to cross-technical fields; Specifically, a pre-trained Transformer model is used, and the model tasks are set as follows: Task 1, identify technical terms; Task 2, determine the technical field to which the input text belongs; Task 3, semantic embedding generation to optimize term representation; Specifically, the Transformer model first receives the input technical text through the input layer, uses a tokenizer (such as WordPiece of BERT) to split the text into sub-word units, and defines special Token markers ([CLS] and [SEP], where [CLS] is used to aggregate the semantic information of the entire input technical text, and [SEP] is used to mark the end of the input technical text or separate different text segments) and converts them into a Token sequence (composed of multiple Tokens); then maps each Token to an ID in the vocabulary (Token ID), adds positional encoding to each Token to capture word order information; converts the Token IDs in the input layer into embedding vectors, and captures the context relationship between Tokens by feeding them into the Transformer encoder to generate context-related hidden representations; based on the hidden representations generated by the Transformer encoder in the input layer, term recognition and domain classification are performed simultaneously; Specifically, term recognition is represented by a conditional random field (CRF) as:

[0029] where x represents the input technical text, and y represents the output target label sequence; represents the parameters of the Transformer encoder; W and U represent the transformation matrix and the emission matrix respectively; and represent the i-th Token label and the (i - 1)-th Token label in the label sequence respectively; represents the context-related hidden representation generated by the Transformer encoder; Y represents an enumeration; n is the total number of Token labels; represents all possible target label sequences; It should be noted that the conditional random field provides the Token range of the term for to ensure that the semantic vector is only calculated for the term Token; Specifically, domain classification is represented by a fully connected layer and Softmax in the output layer as:

[0030] where, represents the probability distribution of the technical field categories, represents the hidden representation of the [CLS] Token; C represents the total number of technical fields; It is represented as an index of technical fields and is used to traverse all technical fields; It is represented as the bias term of the fully connected layer; It is represented as the weight matrix of the fully connected layer and is used to map to technical fields; It should be noted that through the fully connected layer and Softmax in the output layer, the quality can be optimized to make adapt to specific technical fields; Specifically, the Transformer model for multi-task learning generates semantic representations as follows:

[0031] Among them, It is represented as the semantic vector of term W, and Avg represents the average pooling operation, that is, taking the average of the hidden representations of multiple Tokens of the term; It is represented as the Token index corresponding to the term; It should be noted that through the cooperation of the conditional random field and the fully connected layer and Softmax in the output layer, semantic representations adapted across technical fields are jointly generated, ensuring that the semantic representations are adapted to different technical fields and solving the problem of term ambiguity across technical fields (for example, in computer networks, the processing pressure on a server or network can be called load, such as network load, while in mechanical engineering, the weight or external force borne by a structure can also be called load, such as constant torque load); S2. Extract multi-scale features from the input technical text according to the dynamic knowledge base and semantic representations to generate a comprehensive feature set; Furthermore, the input technical text is segmented into sentences, words are segmented, and dependency syntactic analysis is performed to generate a syntactic structure representation; Specifically, by calling the spaCy library in Python, the input technical text is segmented into multiple independent sentences according to punctuation marks (such as full stops, commas) and grammar rules; for example, for "Federated learning is a distributed algorithm that protects data security through differential privacy and is applicable to multi-party collaboration scenarios", it is segmented into sentence 1: "Federated learning is a distributed algorithm", sentence 2: "It protects data security through differential privacy", and sentence 3: "It is applicable to multi-party collaboration scenarios"; Specifically, by calling the Jieba library in Python, the sentences are split into words in combination with the technical field dictionary; Specifically, by calling the DependencyParser of the spaCy library in Python, the syntactic structure is analyzed to generate a dependency tree to reveal the syntactic dependencies between words (such as subject-predicate-object, modification relationships); Specifically, the syntactic structure is represented as:

[0032] Among them, DepTree represents a dependency tree, and s represents a sentence; Furthermore, according to the syntactic structure representation, the global technical objective and sub-technical objectives are extracted respectively to generate a feature vector representing the overall function of the technology and a modular sub-technical feature set; Specifically, select the main clause from the syntactic structure representation (usually the first sentence or the sentence with the most dependencies), and calculate the centrality score of the sentence based on the dependency relationship weights between words:

[0033] Among them, represents the dependency relationship weight, represents the i-th sentence; represents the centrality score of each sentence; It should be noted that the sentence with the highest centrality score is the main clause; Specifically, by traversing the dependency tree, find the root node (predicate or copula), subject, and object / complement; if the root node is a verb, extract the verb + object; if the root node is a copula, extract the subject + complement; obtain the global objective function It is expressed as: or

[0034] Among them, represents the object / complement, represents the subject, represents the root node; Specifically, substitute into the Transformer model for multi-task learning to obtain:

[0035] Among them, represents the feature vector of the overall function of the technology; Specifically, by combining the dependency tree with the technical domain dictionary, check whether there are subordinate clauses or adverbial clauses in the sentence, and calculate the technical relevance between the subordinate clause and the main clause :

[0036] Among them, Dictionary represents the technical domain dictionary, F is an indicator function, which is 1 when both the subordinate clause and the main clause are in the technical domain dictionary, otherwise 0; e represents the subordinate clause or adverbial clause; Specifically, if , then call the Transformer model for multi-task learning to obtain:

[0037] Among them, represents a modular sub-technical feature set; Furthermore, fuse the feature vector of the overall technical function and the sub-technical feature set to generate a comprehensive feature set; It should be noted that by generating a comprehensive feature set, syntactic relationships and relevant terms are considered, ensuring that the extracted technical text features not only conform to grammatical logic but also cover the technical field, solving the problem of insufficient capture of input technical text features by traditional methods; S3. Use the comprehensive feature set and the dynamic knowledge base to construct a technical dependency graph, generate an evaluation score for professional technology, based on the evaluation score, calculate a comprehensive score using cross-technical field weights with adaptive threshold adjustment, map the comprehensive score to the professional technology maturity level, and generate a professional maturity evaluation result; Further, input the comprehensive feature set and call the dynamic knowledge base to construct a technical dependency graph, represent the nodes in the technical dependency graph as sub-technologies, and represent the edges as the dependency relationships between sub-technologies; It should be noted that sub-technologies are composed of sub-technical feature sets, and the dependency relationships between sub-technologies are the dependency relationships between each sub-technology in the sub-technical feature set; Specifically, by traversing the technical dependency graph, obtain the set of all inference paths (passing through each edge and each node to form an inference path) in the technical dependency graph; Furthermore, plan the inference paths of the technical dependency graph through reinforcement learning, evaluate the innovation, feasibility, and commercial value of professional technology, and generate an evaluation score for professional technology; Specifically, reinforcement learning uses a deep Q-network to plan the inference paths of the technical dependency graph. Set the deep Q-network as a 3-layer fully connected network with a hidden layer dimension of 256; the learning rate is 1e-3; the exploration rate is initially 0.9 and linearly decays to 0.1; use the current inference node, comprehensive feature set, and evaluation dimensions (innovation, feasibility, commercial value) as the state of the deep Q-network, use the evaluation dimension (innovation, feasibility, commercial value) for selecting the next inference path as the action of the deep Q-network, and use the evaluation dimension (innovation, feasibility, commercial value) as the reward; Specifically, evaluate the innovation Innov of professional technology by comparing the semantic similarity between the sub-technology and the technical content in the dynamic knowledge base:

[0038] Among them, represents the novelty of the sub-technology, which is obtained by comparing and evaluating with the dynamic knowledge base; Represents the technical similarity; Represents the dynamic knowledge base; Represents a single sub-technical feature in the sub-technical feature set; Represents a single technical feature in the dynamic knowledge base (referring to the technology itself, not sub-technologies); It should be noted that Innov greater than or equal to 7 indicates high innovation, 4 - 6 indicates improvement of existing technologies, and less than 4 indicates combination of existing technologies; Specifically, by matching the currently available resources in the dynamic knowledge base with the required resources in the sub-technical feature set and considering the adaptability of the sub-technical environment, the feasibility Feas of the professional technology is evaluated:

[0039] Among them, Represents the available resources of the technology (such as open-source algorithms), obtained by matching the dynamic database; Represents the resources required by the technology (such as computing power), obtained through the sub-technical feature set; Represents the environmental adaptation coefficient (0 - 1), such as policy compliance; It should be noted that Feas greater than or equal to 8 indicates high feasibility, 5 - 7 indicates feasibility under restricted conditions (i.e., it can be implemented, but certain conditions need to be met), and less than 5 indicates feasibility under risky conditions (i.e., it may be implemented and there are certain risks); Specifically, by estimating the matching degree between the potential market size and the application scenario, the commercial value Comm of the professional technology is evaluated:

[0040] Among them, Represents the potential market size, obtained from the dynamic knowledge base; Represents the application scenario matching degree, obtained by scenario-embedded computing; It should be noted that Comm greater than or equal to 7 indicates high commercial value, 4 - 6 indicates moderate commercial value, and less than 4 indicates low commercial value; Furthermore, based on the evaluation scores, the cross-technical field weights are determined through reinforcement learning, and the comprehensive score is calculated; Specifically, the comprehensive score Maturity is expressed as:

[0041] Among them, 、 、 Represent the weights of Innov, Feas, and Comm respectively, and their initial weights are 0.4, 0.3, and 0.3 respectively, which are determined through the rewards in reinforcement learning:

[0042]

[0043] Among them, represents an increase or decrease in actions; Furthermore, by adopting the hierarchical accuracy based on historical technology evaluation data, the cross-technology field weights are adaptively adjusted by threshold; Specifically, whenever the weights are adjusted, normalization is required:

[0044] Specifically, based on the hierarchical accuracy of historical technology evaluation data, subtract the two action changes to ensure the stability of the reward R:

[0045] Among them, H_Accuracy represents the hierarchical accuracy of historical technology evaluation data; Furthermore, map the comprehensive score after threshold adjustment to the professional technology maturity level to obtain the professional maturity evaluation result; Specifically, the mapped professional technology maturity levels are respectively: Initial stage: 0 - 2; indicating that the technical concept is initially formed and lacks practical verification; Foam stage: 2 - 4; indicating that the technology receives high attention but its practicality needs to be proven; Trough stage: 4 - 6; indicating that the technology is not yet perfect and further research and development are required; Climbing stage: 6 - 8; indicating that the technology is initially verified and has commercial potential; Maturity stage: 8 - 10; indicating that the technology is stable and can be applied on a large scale; It should be noted that through the adaptive threshold adjustment and mapping the comprehensive score to the professional technology maturity level, not only can the maturity of the technical solution be effectively identified, but also whether the technical solution has reference value can be judged; S4. Optimize the evaluation result by integrating user feedback through online learning to improve the professionalism of the professional maturity evaluation result; Furthermore, through the online learning model, establish a loss function of the mean square error between the user score and the model prediction, and adjust the cross-technology field weights; Specifically, the loss function is expressed as:

[0046] Among them, represents the previously obtained comprehensive score, represents the comprehensive score obtained through user feedback; M represents the number of user feedbacks; L represents the loss value; Specifically, adjust the weights across different technical fields:

[0047] Specifically, when the loss value is 0, do not adjust the weights across different technical fields; Denoted as an action, Denoted as the descending gradient of the loss value; Denoted as the new adjusted weights across different technical fields, Denoted as the weights across different technical fields that currently need to be adjusted; It should be noted that optimizing and dynamically adjusting the weights through user feedback enhances the reference value of the evaluation results and the user experience.

[0048] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript.

[0049] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0050] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured product including an instruction device that implements the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0051] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are executed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions for implementing the steps specified in one process or a plurality of processes and / or blocks Figure 1 one process or a plurality of processes and / or blocks Figure 1 in one block or a plurality of blocks.

[0052] Although the preferred embodiments of the present application have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic creative concept. Therefore, the appended claims are intended to be construed to cover the preferred embodiments as well as all changes and modifications falling within the scope of the present application.

[0053] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these modifications and variations.

Claims

1. An artificial intelligence-based professional technology maturity assessment method, characterized in that Including: Obtain technical literature, patents, and industry reports to build a dynamic knowledge base, perform semantic embedding across technical fields by inputting technical text in combination with the dynamic knowledge base, and generate semantic representations. Extract multi-scale features from the input technical text according to the dynamic knowledge base and semantic representations, and generate a comprehensive feature set. Construct a technology dependency graph using the comprehensive feature set and dynamic knowledge base, generate an evaluation score for professional technology, based on the evaluation score, calculate a comprehensive score using cross-technical field weights with adaptive threshold adjustment, map the comprehensive score to a professional technology maturity level, and generate a professional maturity evaluation result. Optimize the evaluation result by incorporating user feedback through online learning to improve the professionalism of the professional maturity evaluation result.

2. The method for evaluating the maturity of professional technology based on artificial intelligence according to claim 1, wherein Obtain technical literature, patents, and industry reports to build a dynamic knowledge base, including: Use a crawler tool to obtain technical literature, patents, and industry reports from open databases and build a dynamic knowledge base. Extract keywords from the input technical text and each crawled data item and apply time weighting to filter out technical content associated with the keywords in the input technical text. Merge the filtered technical content with the knowledge base to update the knowledge base.

3. The method for evaluating the maturity of professional technology based on artificial intelligence according to claim 2, characterized in that Perform semantic embedding across technical fields by inputting text content in combination with the dynamic knowledge base to generate semantic representations, including: Adopt a Transformer model for multi-task learning to generate semantic representations adapted to cross-technical fields.

4. The method for evaluating the maturity of professional technologies based on artificial intelligence according to claim 2 or 3, characterized in that, Extract multi-scale features from the input technical text according to the dynamic knowledge base and semantic representations to generate a comprehensive feature set, including: Perform sentence splitting, word segmentation, and dependency syntax analysis on the input technical text to generate a syntactic structure representation. According to the syntactic structure representation, extract the global technical goal and sub-technical goals respectively, and generate a feature vector representing the overall function of the technology and a modular sub-technical feature set. Fuse the feature vector representing the overall function of the technology and the sub-technical feature set to generate a comprehensive feature set.

5. The method for evaluating the maturity of professional technology based on artificial intelligence according to claim 4, characterized in that Construct a technology dependency graph using the comprehensive feature set and dynamic knowledge base to generate an evaluation score for professional technology, including: Represent the nodes in the technology dependency graph as sub-technologies and the edges as the dependency relationships between sub-technologies. Plan the inference path of the technology dependency graph through reinforcement learning, evaluate the innovation, feasibility, and commercial value of professional technology, and generate an evaluation score for professional technology.

6. The method for evaluating the maturity of professional technology based on artificial intelligence according to claim 5, wherein Evaluate the innovation, feasibility, and commercial value of professional technology, including: Evaluate the innovation of professional technology by comparing the semantic similarity between sub-technologies and technical content in the dynamic knowledge base. Evaluate the feasibility of professional technology by matching the currently available resources in the dynamic knowledge base with the required resources in the sub-technical feature set and considering the adaptability of the sub-technical environment. Evaluate the commercial value of professional technology by estimating the matching degree of the potential market size and application scenarios.

7. The method for evaluating the maturity of professional technology based on artificial intelligence according to claim 5, wherein, Based on the evaluation score, calculate a comprehensive score using cross-technical field weights with adaptive threshold adjustment, map the comprehensive score to a professional technology maturity level, and generate a professional maturity evaluation result, including: Based on the evaluation score, determine cross-technical field weights through reinforcement learning and calculate a comprehensive score. Adopt the hierarchical accuracy based on historical technology evaluation data to adaptively adjust the threshold of the cross - technology field weight; Map the comprehensive score after threshold adjustment to the professional technology maturity level to obtain the professional maturity evaluation result.

8. The method for evaluating the maturity of professional technology based on artificial intelligence according to claim 7, characterized in that, The professional technology maturity level includes the starting stage, the bubble stage, the trough stage, the climbing stage and the mature stage.

9. The method for evaluating the maturity of professional technology based on artificial intelligence according to claim 7, characterized in that Optimize the evaluation result by integrating user feedback through online learning, including: online learning adjusts the cross - technology field weight based on the difference between user feedback and the professional maturity evaluation result.

Images