Enterprise technology demand insight method and system based on large model

Abstract

The invention belongs to the technical field of artificial intelligence, and particularly relates to an enterprise technology demand insight method and system based on a large model. According to the method, an intelligent insight framework is constructed and comprises four core modules, namely a multi-dimensional data intelligent acquisition module, an enterprise technical capability portrait module, a technical gap intelligent identification module and a large-model-driven technical demand intelligent reasoning and generation module. The data acquisition module acquires multi-source data of patents, papers, products and the like. And the technical capability portrait module identifies the core technical field and capability level of the enterprise by using an achievement transformation large model. And the technical gap identification module encodes technical data into a high-dimensional vector through semantic space mapping, and identifies the technical gap by adopting cosine similarity and a dynamic threshold mechanism. The demand reasoning generation module adopts a causal reasoning and technical decomposition tree method to decompose technical gaps into hierarchical demands. According to the method, a fusion result transformation large model semantic understanding and an intelligent gap recognition algorithm are adopted, and scientific support is provided for enterprise technology strategy decision making.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, specifically to a method and system for gaining insights into enterprise technology needs based on a large model. Background Technology

[0002] With the deepening development of the digital economy and technological innovation, enterprise technological innovation has become a key driver for enhancing core competitiveness and achieving sustainable development. Accurately identifying and understanding enterprise technology needs is crucial for formulating technology development strategies, optimizing resource allocation, and improving innovation efficiency. However, due to the complexity of the enterprise technology environment and the diversity of technology needs, traditional technology needs analysis methods face numerous challenges.

[0003] Currently, understanding enterprise technology needs faces three core technical challenges: effective integration of multi-source heterogeneous data, deep semantic understanding and reasoning of complex factors, and continuous tracking of dynamically evolving needs. Existing technical solutions to these challenges mainly include: statistical analysis methods based on patent metrics, which identify technological trends through patent citation analysis and technology lifecycle analysis, but lack a deep understanding of individual enterprise characteristics; text mining methods based on natural language processing, which extract demand elements using named entity recognition and relation extraction techniques, but are limited in cross-domain generalization performance due to limitations in training data quality; manual survey methods based on expert knowledge, which, while providing high accuracy, suffer from strong subjectivity, high costs, and low efficiency; and knowledge graph methods based on ontology modeling, which support semantic understanding by constructing domain knowledge graphs, but often lag behind actual needs when facing rapidly developing emerging technology fields. These methods all have significant shortcomings in terms of data integration depth, semantic reasoning capabilities, and dynamic adaptability.

[0004] In recent years, large language models, represented by ChatGPT and GPT-4, based on the Transformer architecture, have demonstrated revolutionary capabilities in semantic understanding, knowledge reasoning, and content generation. Through technologies such as Prompt Engineering, RAG retrieval enhancement generation, and Chain-of-Thought reasoning, they have provided novel technical pathways for understanding enterprise technology needs. Especially in specialized fields such as technology transfer, large models trained specifically for these areas possess a deeper understanding of concepts like technological innovation, industrial application, and commercial value assessment, providing a strong technical foundation for understanding enterprise technology needs. However, the inventors have found that existing technology needs analysis solutions based on general-purpose large models have significant shortcomings in the following aspects: First, they lack specialized application methods for understanding enterprise technology needs, failing to fully leverage the advantages of large models in semantic understanding and reasoning; second, they lack an effective mechanism for integrating multi-dimensional enterprise data, making it difficult to construct a comprehensive and accurate enterprise technology profile; and third, they lack a systematic reasoning framework from the current state of enterprise technology to future needs, failing to achieve accurate identification of technology gaps and intelligent generation of needs. Therefore, there is an urgent need to develop new methods for understanding enterprise technology needs based on the large-scale technology transfer model, and to fully utilize the professional advantages of the large-scale technology transfer model in terms of technology understanding, transfer logic, and value assessment, so as to solve the shortcomings of traditional methods in terms of in-depth analysis, intelligent reasoning, and automation. Summary of the Invention

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] This invention fully leverages the powerful capabilities of large language models in semantic understanding, knowledge reasoning, and content generation. By constructing an intelligent collection and preprocessing mechanism for multi-dimensional enterprise data, and utilizing semantic space mapping technology based on large models to accurately profile enterprise technological capabilities, it identifies gaps in enterprise technological development through a semantic distance analysis-based technology gap identification algorithm. Finally, a demand reasoning generation mechanism driven by large models outputs personalized insights into technological needs. This method effectively solves the problems of insufficient semantic understanding, inadequate reasoning and analysis capabilities, and low personalization in enterprise technology demand analysis, providing intelligent technical support for enterprise technological innovation decisions.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] The first aspect of this invention provides a method for gaining insights into enterprise technology needs based on a large model, comprising:

[0009] Multi-dimensional intelligent data acquisition plays a core role in the acquisition and integration of enterprise technical information in this invention. Through intelligent acquisition strategies and large-model-driven quality control, it achieves comprehensive collection and high-quality preprocessing of enterprise technology-related data. The system can automatically identify enterprise industry characteristics, formulate personalized data acquisition plans, and establish a panoramic view of enterprise technology through multi-source data fusion.

[0010] In this invention, the enterprise's technological capability profile plays a central role in comprehensively assessing its current technological status. Through the semantic analysis capabilities of a large-scale model, it automatically identifies the enterprise's technological strengths and competitive weaknesses, constructing a multi-dimensional map of its technological strength. The system can accurately assess the enterprise's development level in different technological fields, identify technological evolution trends, and precisely position the enterprise's technological competitive advantage through comparative analysis with industry benchmarks. The specific implementation process includes:

[0011] (1) Identification of core technological capabilities: Utilize large-scale model analysis of enterprise patents, papers, and product data to design prompt templates such as "Identify the enterprise's core technology areas and technical expertise based on the following data." Assess the enterprise's capability level in each technology area through indicators such as patent quantity, quality, and citation frequency, and determine the level of technology maturity by combining the technology life cycle theory.

[0012] (2) Assessment of Technological Competitive Position: A comparative analysis is conducted with major competitors in terms of patent portfolio, technical indicators, and market share. Time series analysis is used to track the company's technological evolution path and identify key nodes in technological development. Indicators such as R&D investment intensity, the growth trend of technical personnel, and patent application density are analyzed to quantify the company's technological innovation activity.

[0013] Intelligent identification of technological gaps is the core innovative technology of this invention. This module utilizes the semantic understanding capabilities of a large-scale technology transfer model to map the current state of a company's technology with cutting-edge industry technologies into a unified high-dimensional semantic space. By calculating semantic distance, it achieves precise quantitative identification of technological gaps. The system can automatically identify gaps in a company's technological development, assess the degree of impact of these gaps, and prioritize them, providing a scientific basis for generating technological needs. Semantic space mapping and gap identification include the following key elements:

[0014] (1) Construction of the technology semantic space: Enterprise technology data T_enterprise and cutting-edge technology data T_frontier are encoded into high-dimensional semantic vectors through a large-scale technology transfer model. Enterprise technology vector V_enterprise=Encoder(T_enterprise)∈R^d, cutting-edge technology vector V_frontier=Encoder(T_frontier)∈R^d, where d is 1024 or 1536 dimensions. Normalized by L2 regularization: V'_enterprise=V_enterprise / ||V_enterprise||_2.

[0015] (2) Quantitative Calculation of Technology Gap: Semantic distance is calculated using cosine similarity: Distance[i,j]=1-cosine_similarity(V'_enterprise[i],V'_frontier[j]). A threshold is dynamically set based on the enterprise's technological strength: threshold_gap=base_threshold×(1+enterprise_level×industry_factor). Technology fields exceeding the threshold are identified.

[0016] (3) Prioritization of gaps: The technological gaps are quantitatively assessed by combining dimensions such as market share impact, revenue impact, and competitive position impact. A two-dimensional matrix of impact degree and urgency is constructed to prioritize the technological gaps, thereby determining key directions for subsequent demand generation.

[0017] By using this semantic space-based method for identifying technological gaps, the system ensures both the accuracy and objectivity of gap identification and achieves a deep understanding of complex technological relationships. This allows for the accurate identification of key gaps in enterprise technology development, laying a solid foundation for the precise generation of technological needs.

[0018] In this invention, a large-scale model-driven technology requirement reasoning and generation module plays a central role in transforming technology gaps into concrete needs. Through the logical reasoning and knowledge integration capabilities of the large-scale model, this module converts abstract technology gaps into actionable technology requirement solutions. The system can automatically analyze the root causes and impact chains of technology gaps, generate targeted technology acquisition recommendations, and hierarchically plan requirements according to urgency and strategic value. The final output is a complete technology requirement report including implementation paths and risk assessments, providing precise guidance for enterprise technology decision-making.

[0019] A second aspect of the present invention provides an enterprise technology demand insight system based on a large model, comprising:

[0020] The multi-dimensional intelligent data acquisition module achieves comprehensive collection and high-quality processing of enterprise technical data by constructing personalized data acquisition strategies and a large-model-driven quality control mechanism. This module uses Scrapy distributed crawling technology and API calls to acquire multi-source heterogeneous data, utilizes the Jaccard similarity algorithm for data deduplication, and achieves data standardization and fusion through JSON Schema format.

[0021] The Enterprise Technology Capability Profile module is responsible for constructing a comprehensive assessment of the enterprise's current technological status and analyzing its competitive position. This module utilizes the semantic understanding capabilities of a large-scale model to identify the enterprise's core technology areas, tracks the trajectory of technological evolution through time-series analysis, and generates a multi-dimensional enterprise technology capability map by combining industry comparative analysis.

[0022] The intelligent technology gap identification module comprises a semantic encoding unit, a distance calculation unit, and a gap ranking unit. The semantic encoding unit uses a large-scale technology transfer model to map enterprise technologies and cutting-edge technologies into high-dimensional vectors; the distance calculation unit identifies technology gaps through cosine similarity calculation and a dynamic threshold setting mechanism; and the gap ranking unit prioritizes technologies based on an impact-urgency matrix, generating a list of key technology gaps to focus on.

[0023] The large-model-driven intelligent reasoning and generation module for technology requirements achieves intelligent transformation from technology gaps to specific needs by constructing a chain-like reasoning mechanism and a multi-level requirement generation algorithm. This module uses causal reasoning logic to analyze the impact chain of gaps, employs a technology decomposition tree method to decompose requirements, and uses multi-attribute decision analysis to assess and obtain pathways, ultimately generating a technology requirement insight report containing visual charts.

[0024] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps in the above-described enterprise technology requirements insight method based on a large model.

[0025] A fourth aspect of the present invention provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the above-described enterprise technology demand insight method based on a large model.

[0026] The beneficial effects of this invention are:

[0027] (1) This invention breaks through the limitations of traditional enterprise technology analysis that relies on manual data collection and processing by adopting an intelligent data acquisition strategy and a large model-driven quality control mechanism. This method utilizes multi-source data fusion and semantic understanding capabilities to achieve automated collection and high-quality preprocessing of multi-dimensional technical data such as enterprise patents, papers, products, and finances. It fully leverages the advantages of large models in data understanding and quality assessment, and significantly improves data processing efficiency and accuracy.

[0028] (2) This invention innovatively introduces semantic space mapping technology, which maps the current state of enterprise technology with the cutting-edge technology in the industry to a unified high-dimensional semantic space, thereby achieving accurate quantitative identification of technology gaps. The system adopts a cosine similarity calculation and dynamic threshold setting mechanism to overcome the problems of strong subjectivity and difficulty in quantification in traditional technology comparison analysis, and provides an objective and quantifiable analysis method for identifying enterprise technology gaps.

[0029] (3) This invention designs a large-model-driven chain reasoning mechanism, overcoming the limitations of traditional technical requirement analysis, which is simple in logic and lacks depth. By combining causal reasoning logic and technology decomposition tree method with multi-attribute decision analysis and hierarchical requirement generation, it realizes the intelligent transformation from technical gaps to specific requirements. This method not only ensures the logical integrity of requirement analysis, but also achieves a deep understanding and reasoning of complex technical relationships.

[0030] (4) This invention has good versatility and practicality. On the one hand, based on the semantic understanding capability of the large-scale achievement transformation model, the system can adapt to the needs of enterprises in different industries and technical fields; on the other hand, through the generation of visual reports and multi-level demand planning, it provides scientific data support and implementation guidance for enterprise technology development decisions, significantly improving the pertinence and effectiveness of enterprise technology innovation.

[0031] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0032] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0033] Figure 1 This is an overall flowchart of the enterprise technology demand insight method based on a large model according to an embodiment of the present invention;

[0034] Figure 2 This is a flowchart of the multi-dimensional intelligent data acquisition method according to an embodiment of the present invention;

[0035] Figure 3This is a flowchart illustrating the enterprise's technological capabilities according to an embodiment of the present invention;

[0036] Figure 4 This is a flowchart illustrating the intelligent identification of technological gaps in an embodiment of the present invention;

[0037] Figure 5 This is a flowchart of the intelligent reasoning and generation of technical requirements driven by the large model in this embodiment of the invention; Detailed Implementation

[0038] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0039] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0040] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0041] S101: Multi-dimensional intelligent data collection. By constructing personalized collection strategies and a large-model-driven quality control mechanism, this step achieves comprehensive collection and high-quality preprocessing of enterprise technical data. It utilizes Scrapy distributed crawling technology and API interfaces to acquire multi-source data such as patents, papers, and products, laying a data foundation for subsequent technical analysis.

[0042] S102: Enterprise Technological Capability Profile. Utilizing the semantic understanding capabilities of a large-scale technology commercialization model, this profile identifies the core technological areas and capability levels of enterprises. By tracking technological evolution through time-series analysis and combining it with industry comparative analysis, it generates a multi-dimensional enterprise technological strength map, providing an accurate technological baseline for identifying technological gaps.

[0043] S103: Intelligent identification of technological gaps, including three sub-steps: First, using a large-scale technology transfer model, enterprise technology data and cutting-edge technology data are encoded into high-dimensional semantic vectors to construct a unified technology semantic space; then, semantic distance is calculated through cosine similarity, and a dynamic threshold setting mechanism is used to identify technological gaps; finally, a two-dimensional matrix is ​​constructed by combining the degree of impact and urgency to prioritize the gaps.

[0044] S104: Intelligent reasoning and generation of technology requirements driven by a large model. It uses causal reasoning logic to analyze the impact chain of technology gaps and employs the technology decomposition tree method to break down complex requirements into core technologies, supporting technologies, and other levels. Through multi-attribute decision analysis, it evaluates different acquisition paths, generates hierarchical technology requirements, and outputs a complete insight report including visual charts.

[0045] Reference Figure 2 A flowchart of the multi-dimensional data intelligent acquisition process is provided, detailing the construction of multi-dimensional data intelligent acquisition.

[0046] Step 201: Enterprise Feature Identification and Data Collection Strategy Configuration. A rule-based enterprise classification algorithm is employed to automatically identify the enterprise's technological focus by analyzing key information such as industry codes and main business descriptions in business registration information. For example, for enterprises in the "Intelligent Manufacturing" category, the system focuses on collecting patent technology data, product technical specifications, and production process documents; for "Internet Technology" enterprises, it focuses on acquiring software copyrights, open-source project code, and technical blogs, ensuring the targeted and comprehensive nature of data collection.

[0047] Step 202: Multi-dimensional Data Source Identification and Interface Configuration. Establish a data source system covering the entire enterprise technology landscape, including patent databases (State Intellectual Property Office, WIPO), academic paper databases (CNKI, Wanfang), corporate website, recruitment platforms, financial reports, product catalogs, etc. For structured data, use direct API calls; for unstructured data, use Scrapy distributed crawler technology, configure a User-Agent rotation mechanism and a proxy IP pool, and set the request frequency to 8-15 times per second to avoid anti-crawler restrictions.

[0048] Step 203: Intelligent Data Acquisition Execution and Monitoring. Initiate multi-threaded concurrent acquisition tasks to achieve efficient acquisition of multi-dimensional enterprise data. This includes the following sub-steps:

[0049] Step 203-1: Keyword Strategy Generation and Data Collection Task Configuration. Generate a keyword library based on the company name, core products, and main technical fields. For example, for "Huawei Technologies Co., Ltd.", generated keywords include combinations such as "Huawei + 5G", "Huawei + chip design", and "Huawei + communication technology" to ensure the relevance and comprehensiveness of the collected data.

[0050] Step 203-2: Concurrent Data Acquisition and Status Monitoring. Start multi-threaded acquisition tasks and monitor the acquisition progress, success rate, and exceptions of each data source in real time. Set up an acquisition task scheduling mechanism to automatically retry failed acquisition tasks, ensuring the stability and integrity of data acquisition.

[0051] Step 204: Data Quality Control and Cleaning. Intelligent quality control and standardization processing are performed on the collected raw data. This includes the following sub-steps:

[0052] Step 204-1: Data Quality Assessment and Filtering. A large-scale data transformation model is used to assess data quality across three dimensions: accuracy, completeness, and timeliness. A quality score threshold of 0.7 is set; data below this threshold is marked for manual review. A prompt template is designed to allow the large-scale model to assess the relevance of the data to the company's technological field and automatically filter out irrelevant information.

[0053] Step 204-2: Duplicate Data Detection and Deduplication. The Jaccard similarity algorithm is used to detect nearly duplicate data, with a similarity threshold of 0.85. For data records with similarity exceeding the threshold, the edit distance algorithm is used for precise judgment, retaining versions with more complete information.

[0054] Step 204-3: Entity Recognition and Relationship Construction. Key entities such as technical terms, personnel names, and organization names are extracted using named entity recognition technology to establish a company-specific technical dictionary and entity relationship graph, laying the foundation for data fusion.

[0055] Step 205: Standardization, Fusion, and Storage of Heterogeneous Data. Heterogeneous data from different data sources is converted into a unified JSON Schema format. An entity-based data fusion method is used, anchoring data around core entities such as company name, patent applicants, and key technical personnel. Ultimately, a comprehensive enterprise technology data warehouse is constructed, encompassing dimensions such as basic company information, technical capability data, innovation achievement data, and talent team data, providing a high-quality data foundation for subsequent technical capability profiling analysis.

[0056] Reference Figure 3 The flowchart illustrates the steps involved in creating a profile of an enterprise's technological capabilities, providing a detailed explanation of how to do so.

[0057] Step 301: Core Technology Capability Identification. Utilizing large-scale model analysis of enterprise patents, papers, products, and other technical data, the system automatically identifies the technological fields the enterprise is involved in. For example, it identifies technology tags such as artificial intelligence, industrial internet, and predictive maintenance from the "Intelligent Manufacturing Equipment Fault Prediction System," and establishes a distribution map of the enterprise's technological fields.

[0058] Step 302: Technical Capability Level Assessment. A technical capability assessment model is constructed based on indicators such as patent quantity, quality, and citation frequency. The assessment formula is designed as: C_score=α×P_count+β×P_quality+γ×P_citation, where α, β, and γ are weighting coefficients, set to 0.3, 0.4, and 0.3 respectively, to ensure the importance of patent quality and influence.

[0059] Step 303: Time-Series Analysis of Technological Development Trends. By analyzing the trends in technological data from 2019 to 2024, the direction of enterprise technological development is predicted. This includes the following analytical dimensions:

[0060] Step 303-1: Assessment of Technological Innovation Activity. An activity assessment system was constructed, comprising six indicators including R&D investment intensity, patent application frequency, and the speed of new technology adoption. The average annual growth rate of patent applications was 16.2%, and the R&D investment as a percentage of revenue increased from 4.1% to 6.8%, indicating a continuous rise in innovation activity.

[0061] Activity level calculation formula: Activity_Level(t)=Σ(wi×Normalized_Indicator(i,t)) where w1-w6 are the weights of each indicator: 0.25, 0.20, 0.18, 0.15, 0.12, 0.10.

[0062] Step 304: Comparative Analysis of Industry Competitive Position. Select 4-5 major competitors for technological comparison. Construct a competitiveness assessment matrix based on three dimensions: number of patents, technological advancement, and market performance. Generate a technology competitiveness radar chart to visually display the company's relative position in each technology field.

[0063] The competitive position assessment adopts a standardized scoring mechanism: Competitive Position = 0.4 × Patent Strength + 0.35 × Tech Advancement + 0.25 × Market Performance

[0064] Step 305: Comprehensive Generation of Enterprise Technology Profile. Integrating results from technology capability identification, development trend analysis, and competitive position assessment, a comprehensive enterprise technology profile is generated, including the following core elements:

[0065] Technology Advantage Distribution Map: Identifies the capability level and development potential in core technology areas.

[0066] Technology Development Trend Curve: Predicting the Development Trajectory of Various Technology Fields Over the Next 3 Years

[0067] Competitive Position Quadrant Chart: Positioning a company within the "Technological Strength - Market Influence" coordinate system

[0068] Technology Maturity Radar Chart: Displays the distribution of various technologies across their lifecycle stages.

[0069] The final output is a standardized enterprise technology baseline profile, which includes key information such as identification of areas of technological advantage, capability level assessment, and development trend prediction, providing an accurate benchmark for subsequent technology gap analysis.

[0070] Reference Figure 4 The flowchart of the intelligent identification steps for technology gaps provides a detailed explanation of the construction of semantic space mapping and technology gap identification.

[0071] Step 401: Technical Data Preprocessing and Cleaning. Standardize the collected enterprise technical data and cutting-edge industry technical data, extracting core technology descriptions. For example, extract the core technologies "machine vision + defect detection + industrial application" from the enterprise patent "An industrial defect detection device and method based on machine vision," ensuring the quality and consistency of the input data.

[0072] Step 402: Large-Scale Model Semantic Encoding Configuration. Based on the encoding capabilities of the large-scale technology transfer model, a semantic encoding strategy specifically designed for technology gap analysis is implemented. Model parameters are configured to achieve optimal technical semantic representation, including setting the encoding dimension to 1536 dimensions to ensure the integrity of semantic information, and employing an average pooling strategy to handle long text inputs.

[0073] Step 403: Semantic Space Construction and Vector Standardization. This involves semantically encoding technical data using a large-scale technology transfer model, specifically including the following sub-steps:

[0074] Step 403-1: Technology Vector Encoding. Input the enterprise technology data T_enterprise and the cutting-edge technology data T_frontier into the large model to obtain semantic vector representations: V_e = Encoder(T_enterprise), V_f = Encoder(T_frontier), where the encoder adopts a large-scale technology transfer model based on the Transformer architecture.

[0075] Step 403-2: Vector normalization. Perform L2 regularization on the obtained semantic vector: V'_e = V_e / ||V_e|| 2 ,V'_f=V_f / ||V_f|| 2 This ensures the accuracy and comparability of subsequent similarity calculations.

[0076] Step 404: Semantic Distance Calculation and Threshold Setting. The cosine similarity algorithm is used to calculate the semantic distance between the enterprise's technology and cutting-edge technologies. A dynamic threshold mechanism is set to adapt to the different technological foundation levels of different enterprises. The threshold calculation formula is: θ = θ_base × (1 + α × L_enterprise + β × F_industry), where θ_base is the base threshold of 0.3, L_enterprise is the enterprise's technological capability level coefficient, and F_industry is the industry's technology intensity factor.

[0077] Step 405: Identification and Classification of Technological Gaps. Based on the calculated semantic distance and set thresholds, identify the technological gaps in enterprise development. Classify the identified gaps according to technological fields. For example, in intelligent manufacturing enterprises, identify specific technological gaps such as "edge computing technology" and "digital twin modeling," providing clear technological directions for subsequent needs analysis.

[0078] Step 406: Gap Priority Assessment and Ranking. Construct a technology gap impact-urgency assessment matrix, comprehensively considering multiple dimensions such as market impact, competition impact, and income impact. Use the analytic hierarchy process (AHP) to determine the weights of each dimension, generate gap importance scores, prioritize the technology gaps according to their scores, and output a list of key technology needs.

[0079] Reference Figure 5 The flowchart of intelligent reasoning and generation steps for technology requirements driven by large models provides a detailed explanation of the construction of intelligent reasoning and generation for technology requirements driven by large models.

[0080] Step 501: Analysis of the Impact Chain of Technological Gap. A mechanism for transmitting the impact of technological gaps is constructed using "if-then" causal reasoning logic. For example, "If edge computing capabilities are lacking, real-time data processing of equipment cannot be achieved, thus affecting production efficiency optimization and quality control accuracy," establishing a complete logical chain between the gap and its business impact.

[0081] Step 502: Technology Requirements Decomposition and Hierarchical Construction. The complex technology requirements are decomposed into multiple levels using the technology decomposition tree method, specifically including the following sub-steps:

[0082] Step 502-1: Core Technology Identification. Identify the key core technologies for achieving the technical objectives. For example, break down the "intelligent quality inspection system" into core technology components such as machine vision, deep learning algorithms, and data processing platforms.

[0083] Step 502-2: Supporting Technology Review. Analyze the supporting and complementary technologies required for the implementation of the core technologies, establish a technology dependency map, and ensure the completeness and feasibility of the technology requirements.

[0084] Step 503: Technology Acquisition Path Assessment. A multi-attribute decision analysis method is used to assess the feasibility of different technology acquisition methods. Assessment dimensions include technological feasibility, cost-effectiveness, timeframe, and risk level. The optimal technology acquisition strategy is determined through weighted scoring.

[0085] Step 504: Layered Requirements Generation and Planning. Based on the urgency and strategic importance of the technology gap, generate three levels of technology requirements: urgent requirements (6-12 months), important requirements (1-3 years), and strategic requirements (3-5 years). Each requirement includes structured elements such as requirement description, key technologies, implementation path, expected results, and investment budget.

[0086] Step 505: Demand Feasibility Verification and Optimization. Using a techno-economic feasibility analysis framework, and combining financial indicators such as Net Present Value (NPV) and Internal Rate of Return (IRR), the economic rationale for the demand is verified. Expert opinions are collected using the Delphi method to optimize and adjust the demand plan, ensuring its scientific validity and operability.

[0087] Step 506: Generation of Technology Requirements Report. Integrate the requirements analysis results to generate a complete technology requirements insight report, including visualizations such as a technology gap radar chart, a requirements priority matrix, an implementation path Gantt chart, and a return on investment analysis. This provides comprehensive data support and implementation guidance for enterprise technology decisions.

[0088] Example 2

[0089] This embodiment provides an enterprise technology requirements insight system based on a large model, which includes:

[0090] The multi-dimensional data intelligent acquisition module is used to intelligently collect and preprocess enterprise technology-related data to generate standardized datasets. This module drives multi-source data acquisition by constructing a distributed acquisition architecture and quality control mechanism (Q = {q1, q2, ..., q}), transforming raw data D into a structured technical dataset T = {t1, t2, ..., t}.

[0091] The enterprise technology capability profiling module is used to construct a multi-dimensional technology capability analysis mechanism and generate an enterprise technology strength map. This module identifies and evaluates technology fields and capability levels through a designed capability assessment function C(T), continuously updating the technology profile P to improve the system's accuracy in understanding the enterprise's current technological status.

[0092] The intelligent technology gap identification module includes the following functional units: a semantic encoding unit that uses the large-scale technology transfer model g to obtain the enterprise technology vector V(ti) and the cutting-edge technology vector V(f). j The gap calculation unit is based on semantic distance gap(ti,f). j The gap is identified using a combination of a comprehensive priority score (gi) and a dynamic threshold δ; the priority ranking unit sorts the requirements by a comprehensive priority score (gi) to generate a Top-N technical requirement list R.

[0093] The large-model-driven intelligent reasoning and generation module for technology requirements is used for in-depth analysis of technology gaps and generation of implementation recommendations. This module integrates a causal reasoning engine, a technology decomposition tree builder, and a visualization report generator, outputting a complete technology requirement insight report with decision support.

[0094] Example 3

[0095] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the enterprise technology requirements insight method based on a large model as described above.

[0096] Example 4

[0097] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the enterprise technology requirements insight method based on a large model as described above.

[0098] Those skilled in the art will recognize that embodiments of the present invention can take the form of a method, system, or computer program product. Therefore, the invention can be implemented in hardware, software, or a combination of both. Furthermore, the invention can take the form of a computer program product on a computer-usable storage medium (including but not limited to disk storage and optical storage), containing one or more computer-usable program codes.

[0099] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A technology for gaining insights into enterprise technology needs based on a large model, characterized in that, This includes multi-dimensional intelligent data collection, enterprise technology capability profiling, intelligent identification of technology gaps, and intelligent reasoning and generation of technology requirements driven by a large model. Specifically, multi-dimensional intelligent data collection uses personalized collection strategies to acquire comprehensive enterprise technology data; enterprise technology capability profiling is used to build an enterprise technology strength assessment system based on a large-scale technology transfer model; intelligent identification of technology gaps includes semantic encoding units, distance calculation units, and gap ranking units, achieving unified semantic space mapping and objective quantification of technology gaps through the large-scale technology transfer model; and intelligent reasoning and generation of technology requirements driven by a large model generates hierarchical technology requirements based on causal reasoning logic.

2. A multi-dimensional intelligent data acquisition process, characterized in that, The system employs a rule-based enterprise classification algorithm to identify key technological concerns of enterprises; constructs a multi-dimensional data source system covering patent databases, academic paper repositories, and enterprise websites; utilizes Scrapy distributed crawling technology and API interfaces to concurrently collect data from multiple sources; employs a large-scale technology transfer model to assess data quality and sets quality scoring thresholds for filtering; and builds a panoramic enterprise technology data warehouse through entity-based data fusion methods.

3. A process for constructing a profile of an enterprise's technological capabilities, characterized in that, By using a large-scale technology transfer model, we can analyze enterprise patents, papers, and product data to identify core technology areas; construct a multi-dimensional quantitative evaluation model of technological capabilities, including four dimensions: patent index, innovation index, market index, and team index; track the enterprise's technological evolution trajectory and identify key development nodes through time series analysis; and generate a competitive position evaluation matrix by conducting a technological comparative analysis with major competitors.

4. The enterprise technology capability profile construction process according to claim 3, characterized in that, The design formula for evaluating technological capabilities is C_score=α×P_count+β×P_quality+γ×P_citation, where α, β, and γ are weighting coefficients, set to 0.3, 0.4, and 0.3 respectively; the activity level calculation formula Activity_Level(t)=Σ(wi×Normalized_Indicator(i,t)) is constructed to evaluate the enterprise's technological innovation activity.

5. A process for intelligent identification of technological gaps, characterized in that, The enterprise technology data T_enterprise and the cutting-edge technology data T_frontier are encoded into high-dimensional semantic vectors V_enterprise and V_frontier through a large-scale technology transfer model. The semantic vectors are then subjected to L2 regularization. The semantic distance Distance[i,j] = 1 - cosine_similarity(V'_enterprise[i], V'_frontier[j]) is calculated using cosine similarity. A dynamic threshold mechanism threshold_gap = base_threshold × (1 + enterprise_level × industry_factor) is designed to adapt to different enterprise technology bases.

6. The intelligent identification process for technological gaps according to claim 5, characterized in that, Based on the concept of quantitative identification of technological gaps in a unified semantic space, this approach achieves objective comparability assessment by mapping heterogeneous technology descriptions to the same high-dimensional semantic coordinate system, overcoming the limitations of traditional technology comparison analysis, which is characterized by strong subjectivity and difficulty in quantification. A two-dimensional matrix of impact degree and urgency is constructed to prioritize the identified technological gaps.

7. A large-model-driven intelligent reasoning and generation process for technical requirements, characterized in that, The impact chain of technological gaps is analyzed using "if-then" causal reasoning logic; complex technological requirements are decomposed into core technologies, supporting technologies, and other levels using the technology decomposition tree method; the feasibility of different technology acquisition paths is evaluated through multi-attribute decision analysis; and technological requirements planning is generated at three levels—urgent requirements, important requirements, and strategic requirements—based on urgency and strategic importance.

8. The intelligent reasoning and generation process for technical requirements according to claim 7, characterized in that, Based on the concept of demand generation through chain reasoning of large models, a complete transmission mechanism of technology gap and business impact is established through causal reasoning logic. Combined with technology decomposition tree, it realizes intelligent transformation from abstract gap to specific and operable demand, overcoming the problems of simple logic and insufficient depth in traditional demand analysis.

9. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the steps in the large-model-based enterprise technology requirements insight method as described in any one of claims 1-8.

10. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the steps of the enterprise technology requirements insight method based on a large model as described in any one of claims 1-8.

Images