Innovation method and device and computer readable medium

Abstract

The invention discloses an innovation method, relates to the technical field of artificial intelligence innovation methods, and aims to quantify various innovation methods into computable spatial points by constructing an evolution-action-system three-dimensional innovation space, deeply associate and automatically construct an innovation knowledge graph, and select by knowledge extraction, completion and multi-source fusion dynamic support methods. And based on an evolution-effect-system sequence dynamic optimization method, in combination with multi-dimensional quantitative indexes and interpretability comparison, objective evaluation or sorting evaluation of the scheme is realized, the problems of fragmentation, weak knowledge support and experience dependence on implementation of a traditional innovation method are effectively solved, the universality, normalization and intelligent level of the innovation method are remarkably improved, and the innovation method is suitable for popularization and application. The purpose of effectively completing the innovation process is achieved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence innovation methods, specifically to an innovative method, device, and computer-readable medium. Background Technology

[0002] Classical or modern innovation methods are primarily based on TRIZ, introducing systematic tools, innovation methods, ideality calculation formulas, and conceptual solution evaluation systems. Through element recombination, contradiction analysis, and causal chain analysis, a systematic transformation path from market demand to technological solution innovation is established, strengthening the integration of technological solutions with specific needs and effectively reducing R&D cycles and innovation thresholds. Based on user needs analysis and TRIZ, a feedback mechanism optimizes the recommendation algorithm, improving the accuracy of innovative solution matching. Product parameters are automatically associated with 39 TRIZ engineering parameters, and the matching path is optimized by combining user behavior analysis. Cross-domain technological solution transfer is achieved; however, the keyword system relies on the quality of manual annotation and requires TRIZ experts for contradiction transformation.

[0003] The method for solving industrial product problems based on TRIZ physical contradictions automates problem processing through a five-step process, including screening key problem points, analyzing characteristic parameters, identifying and separating characteristics, and intelligently recommending TRIZ product principles and cases.

[0004] The TRIZ-based product design method uses a technical approach of innovation point extraction, keyword simplification, and cross-database retrieval and comparison. It combines the TRIZ contradiction matrix to generate new solutions, and then uses multi-dimensional evaluation and screening through extension learning to finally output the optimal technical solution, which significantly improves novelty and solution generation efficiency.

[0005] Based on an industry solution case library, a visual causal chain modeling tool was developed, which has a high rate of key defect identification and high speed of problem tracing, but modeling complex systems takes a long time.

[0006] In summary, these innovative methodologies share the common characteristic of deeply integrating the core tools of TRIZ theory with modern computational technology. They lower the innovation threshold through structured processes, enhance the systematic solution capabilities for complex engineering problems, and form an innovation support system covering the entire R&D process. However, these methods are often difficult to understand, cover a limited range of innovative ideas or models, heavily rely on expert experience, and lack computer-aided innovation tools. Currently, they also suffer from poor domain generalization, a limited number of universally applicable solutions to problems, and a lack of potential for widespread adoption and intelligent data mining capabilities. Therefore, more effective and feasible universal innovation methods are still needed. Summary of the Invention

[0007] The purpose of this invention is to provide an innovative method, device, and computer-readable medium that addresses the problems of existing innovative methods being overly abstract, lacking relevant interpretable information, having sparse auxiliary application knowledge, exhibiting diverse expressions across various innovative schools of thought, and lacking usage standards and norms. Furthermore, related innovative methods suffer from semantic gaps in application, are difficult to generalize, heavily rely on experience or expert systems for problem-solving, lack universally accepted and feasible steps, and lack general applicability.

[0008] To achieve the above objectives, a novel innovation method is designed, providing a general representation and application framework. This gives the innovation process advantages such as domain universality, standardized use, feasibility of implementation, and interpretability of results. The specific technical solution is as follows:

[0009] An innovative approach includes constructing a unified multi-dimensional representation space, using spatial points to represent specific innovative methods, using them in conjunction with innovative knowledge, and completing the innovation process according to implementation strategies.

[0010] Optionally, the multi-dimensional aspect includes at least three dimensions: evolution, function, and system.

[0011] Optionally, the representation using spatial points specifically involves mapping various innovative methods to spatial points in the multi-dimensional representation space and describing their dimensional tendencies through quantitative indicators.

[0012] Optionally, the use of innovation knowledge in conjunction with innovation knowledge specifically refers to the selection and application of specific innovation methods in the space through the constructed innovation knowledge graph;

[0013] The innovative knowledge includes the concepts, instances, functions, relationships, and attributes involved in various specific innovative methods, including but not limited to the screening, integration, and processing of world, industry, and domain data, which are processed and output using knowledge graph construction methods and associated with specific innovative methods in the space.

[0014] The aforementioned use of related knowledge means that, in the process of applying specific innovative methods, innovative knowledge must be used to assist in the implementation of subsequent steps.

[0015] The steps for constructing the innovative knowledge graph include, but are not limited to, the following:

[0016] Preprocessing involves format conversion, Chinese word segmentation, knowledge extraction, and reasoning of relevant texts associated with innovative knowledge.

[0017] Knowledge extraction involves named entity recognition, keyword extraction to obtain a keyword list, and knowledge combination extraction to obtain an innovative knowledge list.

[0018] Relationship prediction: Predicting the relationships between named entities;

[0019] Knowledge completion: Fill in the missing items and potential relationships in knowledge triples, and output the completed knowledge combination;

[0020] Knowledge linking and fusion involves format conversion of knowledge from different sources, entity integration, disambiguation, alignment, association, and relationship annotation, and outputting a linked and fused knowledge graph.

[0021] The system reasons, mines, and updates knowledge graphs, receives linked and merged knowledge graphs, other background data and materials, allows users to select, algorithms reason and mine new knowledge combinations, updates the knowledge graph, and outputs innovative knowledge.

[0022] Optionally, the innovation process according to the implementation strategy includes:

[0023] The implementation strategy follows the spatial dimension selection order, optimizing and utilizing innovative methods.

[0024] The preferred approach is that if an innovation method in one dimension is not feasible, an innovation method in the next dimension is selected, and the selection is carried out through innovation scheme evaluation.

[0025] Optionally, the evolutionary dimension reflects objective laws and trends, conforms to the development of time, space and resources, and realizes macro-prediction and trend guidance for the future development of the technology system;

[0026] The aforementioned operational dimension, guided by functional operation, focuses on the operation and change processes of objects, functions, attributes, components, and structures to achieve innovation;

[0027] The system dimension, starting from the integrity of the technical system, focuses on the internal and external components, causal relationships, and contradictions of the system, and revolves around the collaborative innovation of the internal and external technical system.

[0028] Optionally, the evolutionary dimension can be further subdivided into evolutionary sub-dimensions;

[0029] The evolutionary sub-dimensions include, but are not limited to, value, completeness, simplification, coordination, controllability, dynamism, intelligence, transmissibility, heterogeneity, life cycle, biomimicry, and sustainability.

[0030] The action dimension further includes action sub-dimensions;

[0031] The sub-dimensions of action include, but are not limited to, sub-dimensions of function change, function separation, function measurement, function introduction, function retention, function combination, and function movement;

[0032] The system dimension further includes system sub-dimensions;

[0033] The system sub-dimensions include, but are not limited to, supersystem, system, microsystem, component, composition, element, attribute, and relation sub-dimensions.

[0034] Optionally, the evaluation of the innovative solutions is achieved by constructing an objective evaluation benchmark system for the innovative solutions based on multi-dimensional quantitative indicators, invention level stratification, and interpretability comparison mechanism.

[0035] To achieve the above objectives, an apparatus for an innovative method has also been designed, comprising a processor and a memory, the memory storing a computer program, the processor calling the computer program to execute any step of the aforementioned innovative method.

[0036] A computer-readable storage medium is also designed, wherein a computer program is stored in the computer program, which is invoked by a processor to execute any step of the aforementioned innovative method.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0038] I. This invention establishes a unified multi-dimensional representation space of "evolution-action-system" to map various innovative methods into quantifiable spatial points, thus solving the problems of diverse expressions, limited coverage of innovative capabilities, and large semantic gaps in traditional methods.

[0039] This framework standardizes and calculables the innovation process, enhancing the universality and scalability of innovation methods.

[0040] Second, this invention deeply links innovative methods with a structured and rich innovative knowledge graph. This graph is constructed through automated knowledge extraction, relationship prediction, completion, fusion and reasoning, covering knowledge in many aspects such as concepts, instances, functions and attributes. It not only provides knowledge support for the selection and application of innovative methods, but also solves the problem of knowledge sparsity in existing methods, and improves the interpretability and intelligence of the innovation process, making the decision-making basis more sufficient.

[0041] Third, this invention proposes a dimensional sequential implementation strategy, such as "evolution → function → system", which can automatically switch when there are few or no innovative methods in a certain dimension, reducing the blindness of method selection and improving innovation efficiency.

[0042] Meanwhile, an objective evaluation benchmark system based on multi-dimensional quantitative indicators, invention level stratification, and interpretability comparison was established. Combined with dynamic threshold optimization, it can automatically sort and screen innovative solutions to ensure the quality, reliability, and innovativeness of the final output solution. The screening results also help to optimize the above-mentioned innovative methods.

[0043] Fourth, this invention further refines the three main dimensions into a rich system of sub-dimensions, which can support more accurate matching of innovative methods and more in-depth problem analysis.

[0044] In summary, by constructing a multi-dimensional spatial framework, deeply integrating knowledge graphs, optimizing implementation strategies and evaluation mechanisms, and supplementing them with a sub-dimensional system, this invention ultimately achieves the universality, standardization, intelligence, and efficiency of the innovation process, significantly lowering the innovation threshold and improving the systematic problem-solving capabilities and the quality of innovative solutions. Attached Figure Description

[0045] Figure 1 This is a flowchart of the innovative method.

[0046] Figure 2 To unify the representation of multi-dimensional spatial graphs;

[0047] Figure 3 A dimensional information map of the innovation paradigm space;

[0048] Figure 4 A summary of innovative methods for defining the main dimensions and a graph of the dimensional tendency quantification results;

[0049] Figure 5 A summary of innovative methods and a quantitative result diagram of dimensional tendencies for the main functional dimension;

[0050] Figure 6 A summary of innovative methods and a quantitative result diagram of dimensional tendencies for the main system dimension;

[0051] Figure 7 A diagram summarizing the three dimensions of the innovation paradigm space and the corresponding innovation knowledge and specific innovation paradigms (methods);

[0052] Figure 8 A diagram illustrating the three levels of knowledge related to the concept of materials;

[0053] Figure 9 A diagram showing the multi-dimensional evaluation results of the innovative solution;

[0054] Figure 10 Example diagram of an indicator-based evaluation system for innovative solutions. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] In this embodiment, an innovative method is provided, which includes constructing a unified multi-dimensional representation space, using spatial points to represent specific innovative methods, using them in association with innovative knowledge, and completing the innovation process according to the implementation strategy.

[0057] The multidimensional approach includes at least three dimensions: evolution, function, and system. By stripping away highly subjective thinking and incorporating specific innovative methods (which can be called innovative paradigms) into the paradigm space, all innovative methods can be formed into computable rules, axioms, or concepts.

[0058] To encompass all innovation methods, a new concept, innovation paradigm, is defined to represent various innovation methods, as follows:

[0059] The multi-dimensional representation space, based on the specific implementation characteristics of the innovation paradigm (method), is classified into three dimensions in the paradigm space: evolution, system, and function, which can formally represent all innovation methods.

[0060] The spatial point refers to a concrete result in which various innovative methods are manifested throughout the entire innovation paradigm space.

[0061] The representation uses spatial points, where each innovative method is a point in the space. Specifically, various innovative methods are mapped to spatial points in the multi-dimensional representation space, and different dimensional tendencies are assigned to them. These dimensional tendencies are described using quantitative indicators to present a more intuitive expression in the space. For example,... Figure 1 As shown, the data is presented according to the dimensions of evolution, function, and system, corresponding to the perspectives of time, space, and resources. The location can be determined by the orientation quantification values ​​of different dimensions, thus allowing it to be distributed to a specific location in space.

[0062] It is worth noting that, Figure 2 For specific implementation examples, any innovative methods not listed can be included following this approach.

[0063] like Figure 3 As shown, various innovative paradigms (methods) can be categorized into main dimension categories based on their implementation characteristics. This can be designed into three main dimensions: main evolution, main function, and main system. Various innovative paradigms (methods) can be summarized around the characteristics of the main dimensions, thus getting closer to the specific dimension axis of space. Innovative methods that are not listed can be included according to this idea.

[0064] The evolutionary dimension primarily relies on objective laws. Evolution is a holistic view, derived from a comprehensive dataset. It represents a prediction of the future guided by objective laws and possesses predictability. For example, value engineering stems from multiple requirements regarding trends, structures, and functions, derived from objective laws such as time cycles, spatial changes, evolutionary trends, and invention paradigms. The dominant evolutionary paradigm integrates objective environments, changes, laws, and trends. Therefore, formulas, axioms, and laws all fall under the category of dominant evolutionary paradigms, which can be continuously refined. Dominant evolution primarily originates from various evolutionary capabilities, including relevant data. It is something independent of human will, an objective law that will emerge as long as the system exists, such as evolutionary trends, scientific effects, scientific transformations, axioms, formulas, and laws.

[0065] The aforementioned dimension of action mainly reflects functional and relational characteristics. The action comes from changes in functions, attributes, structures, behaviors, etc., inside and outside the system. It focuses on the local. All kinds of processes and methods, whether inherent or externally influenced, can be categorized into the dimension of action for local operation-oriented innovation processes.

[0066] The system dimension mainly reflects the characteristics of system composition. The system is based on the spatiotemporal changes and causal relationships of the system, grasps the main contradictions (internal causes) and secondary contradictions (external causes) in the development and changes of the system, and emphasizes the results of changes. Related innovative methods can be classified into the system dimension.

[0067] In a complete technological system, including components such as power, transmission, control, and execution, corresponding innovative knowledge is provided. This knowledge can be correlated with innovative paradigms, such as... Figure 3 As shown.

[0068] The spatial points mentioned here may vary in their spatial distribution and thus exhibit different paradigmatic dimensional tendencies. Therefore, while assigning a primary dimension label, it is also necessary to provide a qualitative or quantitative assessment of each dimension, such as using evolutionary degree, influence degree, or systematic degree as a measure. This will be discussed in detail below:

[0069] This paper conducts an in-depth analysis of various innovative methods, and provides a detailed discussion of the mapping between different known innovative methods, thus offering explanations of relevant dimensions, such as... Figure 4 As shown, the main evolutionary paradigm is based on evolutionary description and modification. A specific implementation can be to include evolutionary trends, scientific effects, etc. in the main evolutionary dimension. That is, the value of the evolutionary dimension is much greater than that of the other two dimensions, and it is basically displayed along the evolutionary dimension.

[0070] like Figure 5 As shown, under the main action dimension, the related paradigms define the related functions, operations, etc. These paradigms (methods) mainly reflect the action characteristics, so the magnitude of the action dimension is much larger than the other two dimensions, including but not limited to the paradigms or methods listed in the figure.

[0071] like Figure 6 As shown, for the main system paradigm, innovative knowledge can also be used to connect it. Other paradigms are integrated through the quantification of evolution degree / function degree / system degree, and thus mapped onto the paradigm space. Shallow reasoning methods such as STC method and nine-screen method can also be used to improve the main system dimension.

[0072] Furthermore, each paradigm dimension can be further refined into multiple sub-dimensions based on actual usage needs, thereby enabling more granular optimization of implementation strategies for innovative paradigms (methods).

[0073] The evolutionary dimension further includes evolutionary sub-dimensions;

[0074] The aforementioned evolutionary sub-dimensions, based on objective laws and macro trends, integrate global factors such as time cycles, spatial changes, and resource distribution to achieve scientific prediction and regular guidance of the future development of technological systems. The evolutionary sub-dimensions include, but are not limited to, value, completeness, simplification, supersystem, coordination, controllability, dynamism, intelligence, transmissibility, heterogeneity, life cycle, biomimicry, and sustainability.

[0075] The action dimension further includes action sub-dimensions;

[0076] The aforementioned sub-dimensional focus on the local functions and operational orientation of a technical system, emphasizing innovation through specific functional adjustments, attribute optimizations, or structural reorganizations. Therefore, this sub-dimensional emphasizes the process of action rather than the result, such as addition, subtraction, stabilization, measurement, and assistance, specifically based on "generation, change, combination, separation, movement, measurement, and maintenance." Consequently, related innovation methods, primarily action-oriented, can all be categorized under the primary action-dimensional paradigm and reflected in the implementation of specific innovation paradigms (methods).

[0077] The system dimension further includes system sub-dimensions;

[0078] The system sub-dimensional analysis starts from the overall nature of the technical system, analyzes the internal and external components, causal relationships, and contradictions of the system, and solves the collaborative optimization problem of complex systems. The technical system achieves a better result through innovation. The relevant innovation methods themselves emphasize the spatiotemporal changes of the system, the supersystems inside and outside the system, and the microsystems, as well as the driving effect of changes inside and outside the system. Therefore, the innovation methods are mainly presented from the system dimension rather than the process of action, which is called the main system paradigm. The system sub-dimensional analysis includes, but is not limited to, composition, components, parameters, relationships, and attributes.

[0079] Specific implementation examples Figure 7 As shown, taking the evolutionary sub-dimension as an example, innovation paradigms (methods) can be further summarized within it, thus facilitating subsequent specific implementation.

[0080] The aforementioned use in association with innovative knowledge specifically involves selecting and applying specific innovative methods in the space through the constructed innovative knowledge graph, which can be used based on natural language;

[0081] The innovative knowledge mentioned includes concepts, functions, instances, relationships, and attributes related to various specific innovative methods, including but not limited to the screening, integration, and processing of world, industry, and domain data. This knowledge can be summarized into several categories: units, structures, functions, attributes, parameters, and values. It is processed and output using knowledge graph construction methods and then associated with specific innovative methods in the knowledge space.

[0082] The aforementioned association with innovative knowledge refers to the establishment of semantic associations by innovative paradigms (methods) based on their own characteristics, such as semantic information reflecting three aspects: evolution, system, and function, using the representation results of innovative knowledge.

[0083] The innovative knowledge can be represented by characters, vectors, or a combination of both.

[0084] The aforementioned use of related knowledge means that, in the process of applying specific innovative methods, innovative knowledge must be used to assist in the implementation of subsequent steps.

[0085] Innovative knowledge can be defined as: knowledge associated with the application of innovative paradigms during the innovation process, such as... Figure 7 As shown.

[0086] The knowledge graph construction steps include, but are not limited to, preprocessing, knowledge extraction, relation prediction, and knowledge completion, as well as knowledge linking, fusion, reasoning, and mining.

[0087] The preprocessing involves format conversion, Chinese word segmentation, and reasoning of the text, resulting in a more comprehensive knowledge graph with the ability to supplement new knowledge, which can be used for preprocessing in multiple fields.

[0088] The knowledge extraction process involves named entity recognition, keyword extraction to obtain a keyword list, and pairwise combination to obtain a named entity list, thereby enabling automatic and rapid extraction of entity relationships from the knowledge graph.

[0089] Optionally, entity recognition models are used to extract entities, and the knowledge graph is constructed through steps such as preparing the corpus dataset, designing the knowledge graph pattern layer, and designing the knowledge graph data layer. This achieves automatic construction, effectively improving the graph quality and increasing the efficiency and accuracy of entity extraction. It assists enterprise designers in rapid design, avoiding resource waste and reducing repetitive work. Using a natural language processing (NLP) database as the core extension of the knowledge graph, for example, using the Chinese language model ERNIE as a natural language data translation model, and using graph neural network technology to construct a natural language data model for interpretation and multi-dimensional analysis, it can quickly process diverse natural language data.

[0090] The relationship prediction is a prediction of the relationship between two entities.

[0091] Optionally, referencing the triple set in the world knowledge graph, a vertical domain entity recognition model is used to identify whether the first and second entities to be identified are domain entities. If both are domain entities, a domain triple set is constructed. This allows for the automatic identification of vertical domain knowledge from a general knowledge graph, and the automatic construction of a high-quality knowledge graph that covers various types of knowledge.

[0092] Optionally, entities can be categorized into primary and secondary entities, data table schemas can be defined, and Cypher statements can be generated. Visualization tools can be provided for knowledge modeling, data annotation, and text augmentation, constructing a text dataset using augmented large-sample data. A trained named entity recognition model can be used to extract entities from the target text data, enabling the automatic construction of vertical domain knowledge graphs.

[0093] The knowledge completion process involves performing entity recognition, entity extraction, entity deduplication, entity filtering, or triplet processing on the extracted text data to achieve automatic expansion of the knowledge graph.

[0094] Optionally, links from the encyclopedia page's references can be retrieved, added to a queue to determine the training set, and the document topic distribution can be calculated using the LDA model for clustering. Credibility labeling can then be performed to construct a more comprehensive and higher-quality knowledge graph.

[0095] Optionally, acquire industry datasets, train them using rule-based knowledge and neural network models to obtain rule knowledge graphs or models, and classify and categorize the industry data. Clearly display industry data that meets project requirements and perform quantitative analysis.

[0096] Optionally, a domain knowledge graph can be constructed using specialized domain knowledge, and Chinese text features can be extracted using an incrementally pre-trained BERT model. Furthermore, stacked convolutional neural networks and student reordering networks can be combined for sequence labeling learning, enabling high-precision knowledge graph completion and prediction of relationships between entities, achieving accurate, efficient, and automated knowledge mining, maintenance, and updating.

[0097] The linking and fusion processes utilize big data technology to perform format conversion, entity tripartite integration, disambiguation, processing, random pairing, and relation labeling on knowledge. The input graph is used to build a model, and weighted multiple relation labels and associations are obtained to automatically construct the knowledge graph. This reduces knowledge blind spots, simplifies data query time, improves data utilization value and work efficiency, and provides high-quality structured data.

[0098] Optionally, representation learning algorithms can be used to transform the knowledge in the classification domain library into triples for parallel network processing. By leveraging probabilistic knowledge in the knowledge graph, topic classification and historical conversation information can be extracted simultaneously to determine the user's true intent, thereby improving the query efficiency of human-computer interaction information.

[0099] Optionally, knowledge base data in vertical fields can be acquired to generate a question set, which can then be used to implement the functions of an intelligent question-and-answer system and improve the efficiency of knowledge integration.

[0100] Optionally, a graph structure can be constructed using superpixels, and multimodal data can be aligned and unified through graph topological constraints. The modal data can be fused using self-attention and mutual attention mechanisms to obtain semantic information between target entities and overcome the problem of multi-source information mapping.

[0101] The reasoning and mining processes, targeting large amounts of text data, generate a first recommended entity, at least one second recommended entity, and a correlation with at least one recommended entity. After the user selects an entity, a confirmation message is added to the knowledge graph. This can display unstructured data and assist in explaining various embodiments of the present invention.

[0102] In addition, innovative knowledge also includes knowledge related to general or domain-specific problems or solutions, as well as knowledge descriptions, ontology, instances, assertions, constraints, etc., that can be linked to the explanatory space of innovative paradigms.

[0103] Optionally, through literature review and data retrieval, design concepts, attributes, and constraints to organize and express them, and store them in a knowledge graph—constructing a high-quality knowledge graph with low manual cost and improving the efficiency of evidence analysis.

[0104] Optionally, a functional base model (a specific type of innovative knowledge) can be used as an example, such as... Figure 8 As shown, the primary flow is the material, and the secondary flow includes solids, liquids, gases, loose substances, particles, etc. It can be further linked to other examples to initially complete the construction of an innovative knowledge concept.

[0105] Further develop innovative knowledge indexing.

[0106] The innovation knowledge index establishes a mapping relationship between computer systems and innovation paradigms, facilitating association, retrieval, and use.

[0107] The implementation strategy follows a certain order in selecting and using innovative paradigms (methods).

[0108] The preferred order is to consider and select specific innovation paradigms sequentially based on evolution, function, and system dimensions, using the sequence of "evolution first, then function, then system," repeating cyclically. First, consider whether evolutionary trend paradigms can achieve innovation. Evolution is often the first step, playing a driving role, and can generally be used in conjunction with function and system, identifying technological gaps. It guides subsequent, more in-depth innovation paradigms and can even pinpoint the problem. Next, consider whether functional decomposition or the introduction of paradigms (methods) can achieve innovation. Function is often used in conjunction with system, emphasizing the process of function. Finally, consider innovation based on relevant innovation paradigms at the supersystem level, system level, component or component level, which often demonstrate the effect of function.

[0109] The order can optionally follow the sequence of "first action, then system, then evolution," which is related to people's specific experience in execution.

[0110] The aforementioned order can optionally follow the sequence of "system first, then function, then evolution," which is also a common approach to selecting an innovation paradigm.

[0111] Furthermore, the above implementation strategy preferably establishes index relationships or other associations between innovation paradigms, innovation knowledge, and solution libraries during various innovation processes. This enables solution library retrieval, recommendation, generation, and evaluation screening, further reducing the blindness of paradigm screening and achieving the goal of efficiently realizing the innovation process.

[0112] Furthermore, the above implementation strategy can optionally utilize only the innovation paradigm to establish index relationships or other associations between the paradigm and the solution library for solution library retrieval and recommendation. Combining the evaluation results of the retrieved or generated technical solution library with implementation conditions, the most suitable paradigm library can also be obtained. Furthermore, the most suitable solution library can be selected optimally.

[0113] Furthermore, it allows for the integration and generation of various technical solutions, thus bridging the last mile.

[0114] The solution library includes, but is not limited to, various books, solution encyclopedias, journal articles, patents, standards and specifications, technical documents, operation manuals, industry solutions, etc.

[0115] The selection and use of the innovation paradigm (method) or innovation knowledge are achieved through reasoning.

[0116] Alternatively, since machines struggle to achieve intuition and focus primarily on reasoning, while humans excel in intuitive abilities but have limitations in shallow and deep reasoning, it is necessary to input innovative knowledge, algorithms, laws, formulas, and innovative paradigms into the machine. It is also necessary to consider using reasoning models or algorithms to connect different innovative paradigm spatial dimensions and related innovative knowledge, thereby providing solutions.

[0117] The reasoning model or algorithm includes, but is not limited to, various deterministic or uncertain reasoning strategies, such as Bayesian decision-making, operations research, rule-based reasoning, inductive reasoning, and causal reasoning.

[0118] Optionally, it can be implemented based on natural language interaction.

[0119] By processing and analyzing the input natural language, and based on semantic understanding, associations are established with innovative knowledge or paradigms. Combining internal and external constraints with recommendation algorithms or models, reasonable innovative paradigms (methods) or innovative knowledge are recommended to the input. This stage also includes functions such as entity linking and concept matching.

[0120] The completion of the innovation process refers to the execution process from user input to operational decision-making, and then to the provision of an evaluated innovative solution.

[0121] The evaluated innovative solutions need to meet certain innovation evaluation criteria.

[0122] Evaluation benchmarks can also be designed with multi-dimensional indicators, quantitative values, and explanations to establish an innovation capability evaluation system, such as problem-solving effectiveness, solution feasibility, problem coverage, problem relevance, target relevance, prediction effect, prediction coverage, etc., and provide relevant discrete values ​​and continuous ranges.

[0123] Specific embodiments are as follows: Figure 9 As shown: The final generated solution can be further evaluated. The output results are analyzed according to the evaluation criteria, scored from 1 to 5 points. The evaluation criteria for each indicator are as follows:

[0124] Accuracy: Accurately determine the meaning of the output text and provide the correct relation type.

[0125] Clarity: Able to understand the intended meaning of the input text and provide corresponding feedback.

[0126] Interpretability: The output results can be explained in detail and with reasonable justification.

[0127] Faithfulness: Consistent with the semantic representation of the target output. (Optional)

[0128] Holistic: It can comprehensively analyze and compare the input and output content from a holistic perspective.

[0129] The results can be combined with those generated manually or by a large model, and a detailed response example can be provided: The evaluation results of the above solution are {{“Accuracy”: (4.0 / 5.0), “Clarity”: (3.0 / 5.0), “Explainability”: (3.0 / 5.0), “Fidelity”: (3.0 / 5.0), “Completeness”: '...'}.

[0130] Optionally, the scheme can be divided into an indicator system for refined quantitative evaluation, such as... Figure 10 The indicators shown include: parameter optimization degree, contradiction improvement degree, system improvement degree, principle innovation degree, technological revolution, and explanation, including the interpretation of the solution.

[0131] Optionally, each indicator can be scored from 1 to 10.

[0132] Optionally, each indicator can be assigned the same or different weights, and a weighting strategy can be designed based on the actual situation or user preferences, thereby providing a more refined evaluation result for the selection of the solution.

[0133] Based on the evaluation results of innovative solutions to problems using different innovation paradigms (methods), the best innovative solution is selected, and the innovation process is completed.

[0134] Optionally, an ideal quantification evaluation model can be developed. For example, a high ideality requirement means a lower cost budget and a greater number of effects. This is also one of the criteria for selecting the best technical solution. By combining the dynamic selection threshold with the requirements, various innovative solutions can be evaluated and ranked, and the innovation process can be completed after the best solution is selected.

[0135] The explanations provided for the innovation evaluation results can utilize methods such as large models, human interpretation, and intelligent agent comparison to conduct comprehensive evaluation and analysis, ultimately selecting the best innovative solution and completing the innovation process.

[0136] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An innovative method, characterized in that, This includes constructing a unified multi-dimensional representation space, using spatial points to represent specific innovation methods, linking them with innovation knowledge, and completing the innovation process according to implementation strategies.

2. The innovative method according to claim 1, characterized in that: The multidimensional approach includes at least three dimensions: evolution, function, and system.

3. The innovative method according to claim 2, characterized in that: The use of spatial points specifically involves mapping various innovative methods to spatial points in the multi-dimensional representation space and describing their dimensional tendencies through quantitative indicators.

4. The innovative method according to claim 3, characterized in that: The aforementioned use in association with innovative knowledge specifically refers to the selection and application of specific innovative methods in the space through the construction of an innovative knowledge graph; The innovative knowledge includes the concepts, instances, functions, relationships, and attributes involved in various specific innovative methods, including but not limited to the screening, integration, and processing of world, industry, and domain data, which are processed and output using knowledge graph construction methods and associated with specific innovative methods in the space. The aforementioned use of related knowledge means that, in the process of applying specific innovative methods, innovative knowledge must be used to assist in the implementation of subsequent steps. The steps for constructing the innovative knowledge graph include, but are not limited to, the following: Preprocessing involves format conversion, Chinese word segmentation, knowledge extraction, and reasoning of relevant texts associated with innovative knowledge. Knowledge extraction involves named entity recognition, keyword extraction to obtain a keyword list, and knowledge combination extraction to obtain an innovative knowledge list. Relationship prediction: Predicting the relationships between named entities; Knowledge completion: Fill in the missing items and potential relationships in knowledge triples, and output the completed knowledge combination; Knowledge linking and fusion involves format conversion of knowledge from different sources, entity integration, disambiguation, alignment, association, and relationship annotation, and outputting a linked and fused knowledge graph. The system reasons, mines, and updates knowledge graphs, receives linked and merged knowledge graphs, other background data and materials, allows users to select, algorithms reason and mine new knowledge combinations, updates the knowledge graph, and outputs innovative knowledge.

5. The innovative method according to claim 4, characterized in that: The innovation process, as described in the implementation strategy, includes: The implementation strategy follows the spatial dimension selection order, optimizing and utilizing innovative methods. The preferred approach is that if an innovation method in one dimension is not feasible, an innovation method in the next dimension is selected, and the selection is carried out through innovation scheme evaluation.

6. The innovative method according to any one of claims 1-5, characterized in that: The aforementioned evolutionary dimension reflects objective laws and trends, conforms to the development of time, space and resources, and realizes macro-prediction and trend guidance for the future development of technological systems. The aforementioned operational dimension, guided by functional operation, focuses on the operation and change processes of objects, functions, attributes, components, and structures to achieve innovation; The system dimension, starting from the integrity of the technical system, focuses on the internal and external components, causal relationships, and contradictions of the system, and revolves around the collaborative innovation of the internal and external technical system.

7. The innovative method according to claim 6, characterized in that: The evolutionary dimension can be further subdivided into evolutionary sub-dimensions; The evolutionary sub-dimensions include, but are not limited to, value, completeness, simplification, coordination, controllability, dynamism, intelligence, transmissibility, heterogeneity, life cycle, biomimicry, and sustainability. The action dimension further includes action sub-dimensions; The sub-dimensions of action include, but are not limited to, sub-dimensions of function change, function separation, function measurement, function introduction, function retention, function combination, and function movement; The system dimension further includes system sub-dimensions; The system sub-dimensions include, but are not limited to, supersystem, system, microsystem, component, composition, element, attribute, and relation sub-dimensions.

8. The innovative method according to claim 5, characterized in that: The evaluation of innovative solutions is achieved by constructing an objective evaluation benchmark system for innovative solutions based on multi-dimensional quantitative indicators, invention level stratification, and interpretability comparison mechanism.

9. An innovative method of equipment, characterized in that, It includes a processor and a memory, the memory storing a computer program, the processor calling the computer program to perform the steps of the innovative method according to any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that is invoked by a processor to perform the steps of the innovative method according to any one of claims 1-8.

Images