Multi-party participated patent semantic arbitration mechanism and system

Abstract

The invention provides a multi-party participated patent semantic arbitration mechanism and a multi-party participated patent semantic arbitration system, and solves the problems of difficult proof, long period, high cost and opaque judgment of intellectual property disputes in AIGC and complex research and development. The system receives the technical scheme evidence and converts the technical scheme evidence into a DIKWP semantic map (data, information, knowledge, wisdom and intention); hash is stored in a block chain to ensure that evidences cannot be tampered and can be traced; the contribution or infringement degree is quantified through a data-intention algorithm; performing multi-party semantic game decision with human experts in combination with an AI appraisal engine; and a traceable judgment book is generated, and the smart contract automatically executes right attribution, compensation or distribution. Intelligent, fair, efficient and interpretable arbitration is realized, and key technical support is provided for intellectual property management in the AI era.

Description

Technical Field

[0001] This invention belongs to the field of intelligent intellectual property dispute resolution technology. More specifically, it relates to a method and system for handling patent ownership and infringement disputes using computer technology, artificial intelligence, semantic modeling, and distributed ledger technology. This invention is particularly applicable to resolving intellectual property disputes arising from emerging scenarios such as AI-generated content, multi-entity collaborative R&D, and complex technology licensing and transfer. It aims to improve the fairness and efficiency of patent dispute resolution by constructing an objective, efficient, transparent, and automatically enforceable arbitration framework. The core technical fields involved in this invention include: knowledge graph and semantic web technologies, particularly the extended DIKWP (Data, Information, Knowledge, Wisdom, Intent) model; blockchain technology, especially its applications in data storage, tamper-proofing, and smart contracts; and the application of artificial intelligence in legal technology, such as natural language processing, machine learning, and automated reasoning. Background Technology

[0002] The rapid development of information technology, especially the explosive growth of generative artificial intelligence (AIGC) technology based on deep learning in recent years, is profoundly reshaping all aspects of content creation, technological research and development, and even the socio-economic landscape. From ChatGPT's text generation to MidJourney's image creation, and various code generation and scientific discovery models, AIGC is producing potentially commercially valuable and innovative content at an unprecedented scale and speed. However, this technological revolution also poses unprecedented challenges to the existing intellectual property legal system. The traditional intellectual property legal framework is proving inadequate in addressing core issues raised by AIGC, such as "originality determination," "application of infringement standards," and "delineation of liability."

[0003] First, regarding the determination of originality and ownership of rights, the emergence of AIGC blurs the line between human creation and machine generation. In an invention generated by AI, is the "author" the developer who provided the training data, the engineer who designed the algorithm, the user who input the prompt, or the AI ​​model itself? There is no unified standard in judicial practice across countries. For example, in one case, the Beijing Internet Court recognized for the first time that an AI image generated by a user through carefully designed prompts and parameter adjustments possessed "originality" and granted authorship rights to that user. However, the U.S. Copyright Office rejected a request for authorship of the AI ​​system itself in another case, citing the lack of a "human author." This uncertainty is even more complex in the patent field because patents require higher levels of inventiveness and utility. In R&D projects involving multi-party collaboration—for example, one party provides the core algorithm, another provides proprietary datasets, and yet another trains and optimizes the model, ultimately resulting in a new technical solution generated by the AI ​​system—how to fairly and accurately define the contributions of each party and allocate patent rights becomes an extremely thorny problem.

[0004] Secondly, regarding infringement determination and liability tracing, the "black box" nature of AIGC presents significant challenges to proving infringement. Rights holders find it difficult to prove whether an AI model used their copyrighted or patented work without authorization during its training. Even if AI-generated content is substantially similar to prior works, the complexity and unexplainable nature of the generation process make tracing the specific steps of infringement and determining the liable party extremely difficult. The traditional "platform-user" binary liability structure has evolved into a more complex "developer-operator-user" ternary structure in the AIGC era, exponentially increasing the difficulty of liability apportionment.

[0005] Besides the new challenges brought by AIGC, traditional patent dispute resolution mechanisms also have many inherent pain points. First, the difficulty and high cost of evidence collection. Patent litigation or arbitration typically requires a significant investment of time and money. Parties need to collect and organize massive amounts of technical documents, R&D records, email correspondence, etc., as evidence, a cumbersome and costly process. For individual inventors or small and medium-sized enterprises, the high cost of rights protection often deters them from pursuing this path.

[0006] Second, the trial period is long. Patent cases are highly technical, and judges or arbitrators need to spend a lot of time understanding the technology involved, which leads to a generally long trial period. A patent dispute may take several years from filing a lawsuit to a final judgment, which is unbearable for an industry with rapid technological iteration.

[0007] Third, professional barriers and subjective influence. The determination of patent infringement, especially the determination of "equivalent infringement," relies heavily on the judgment of technical experts and is inherently subjective. Different experts or judges may have different understandings of the same technical feature, leading to uncertainty in the ruling.

[0008] Fourth, cross-border rights protection and enforcement are difficult. In the context of globalized R&D and market competition, intellectual property disputes often involve multiple countries or regions. Differences in legal systems across different jurisdictions, cross-border acquisition of evidence, and cross-border enforcement of judgments all face numerous obstacles, greatly increasing the complexity and cost of rights protection.

[0009] To address these challenges, academia and industry have begun exploring the application of emerging technologies to improve intellectual property management and protection. Among these, blockchain technology, due to its decentralized, immutable, and traceable characteristics, is considered to have enormous potential in intellectual property rights confirmation, evidence preservation, and transactions. By recording the hash value and timestamp of a work or invention on the blockchain, strong evidence of originality can be provided. However, most existing blockchain applications remain at the level of "evidence preservation," that is, recording the fact that certain data existed at a certain point in time. While this approach solves the problem of evidence solidification, it cannot delve into the semantic level of technical content for understanding and comparison. It can prove "what" was submitted, but it cannot answer "what the submitted content means" or "whether it is essentially the same as or similar to another technical solution." Therefore, simple blockchain evidence preservation has very limited effectiveness in resolving complex issues of patent ownership division and infringement determination.

[0010] On the other hand, knowledge graph technology, as a semantic network, can represent entities and their relationships in a structured way and has been applied to the field of patent retrieval and analysis. By constructing a patent knowledge graph, the development path of technology, the citation relationships between patents, and the composition of technical features can be clearly displayed. However, traditional knowledge graphs are mainly used for static analysis of existing patent documents, lacking the ability to handle dynamic and adversarial multi-party disputes, and also struggling to model and evaluate higher-level semantics such as the "intelligence" and "intention" of an invention.

[0011] In conclusion, neither traditional legal frameworks nor existing single-technology solutions are sufficient to effectively address the increasingly complex patent disputes of the AI ​​era. There is an urgent need for an innovative mechanism that integrates the objectivity of evidence, the depth of analysis, the fairness of adjudication, and the automation of enforcement. This mechanism must not only solve the problems of evidence preservation and traceability but also delve into the semantic core of the technical content, conduct a refined and interpretable quantitative assessment of each party's contributions, and ultimately achieve reliable and efficient enforcement of the adjudication results. Summary of the Invention

[0012] This invention aims to overcome the shortcomings of existing technologies and provide a multi-party patent semantic arbitration mechanism and system to solve one or more of the following technical problems:

[0013] Traditional patent dispute resolution mechanisms suffer from difficulties in evidence collection, high authentication costs, and susceptibility to tampering or denial, resulting in inefficient arbitration processes and a lack of public credibility.

[0014] In innovative activities involving multiple parties, such as AI-generated content (AIGC) and collaborative research and development, the contributions of each party are difficult to clearly define and quantify, leading to frequent disputes over the division of patent ownership and difficulties in making fair judgments.

[0015] Patent infringement determination, especially the determination of equivalent infringement, is highly subjective and lacks objective, repeatable, and verifiable quantitative analysis tools, resulting in high uncertainty in the ruling.

[0016] The existing arbitration or litigation process is not transparent, and the reasoning for the ruling is often based on legal provisions and subjective arguments, lacking an intuitive and explainable presentation of technical logic and the chain of evidence, making it difficult for the parties to understand and accept the ruling.

[0017] Once an arbitration award is made, its enforcement relies on manual follow-up or judicial enforcement, which is time-consuming and costly. In particular, it faces difficulties in enforcement in cross-border disputes, and cannot guarantee that the rights and interests of the rights holders will be realized in a timely manner.

[0018] There is a lack of a unified technical framework that can integrate semantic understanding, credible evidence storage, intelligent analysis, and automated execution to meet the requirements of high efficiency, high transparency, and high credibility in intellectual property governance in the AI ​​era.

[0019] To address the aforementioned technical problems, this invention provides a multi-party patent semantic arbitration mechanism, which is implemented through a specially designed system.

[0020] The present invention provides a multi-party patent semantic arbitration method, characterized by comprising the following core steps:

[0021] Semantic Evidence Acquisition and DIKWP Graphing: The system provides an interface allowing multiple parties involved in a dispute (such as patent claimants, accused infringers, AI platform providers, and original data providers) to submit their respective technical solution evidence. This evidence can take various forms, including patent documents, R&D notes, code, experimental data, and design drawings. Upon receiving the evidence, the system's semantic parsing module does not simply store it as a file. Instead, it uses artificial intelligence technologies such as Natural Language Processing (NLP) and knowledge extraction to deeply analyze the unstructured or semi-structured evidence content and map it onto a structured DIKWP semantic graph.

[0022] The DIKWP model is an extension of the traditional DIKW (Data-Information-Knowledge-Wisdom) pyramid model, innovatively introducing "Purpose" as the top-level core element. This model deconstructs a technological solution into five interrelated semantic levels:

[0023] Data layer: The most raw, unprocessed symbols or facts, such as raw experimental data, sensor readings, and samples in the training dataset.

[0024] Information layer: Data that has been processed and organized, and given context and meaning, such as statistical charts, technical performance parameters, and feature extraction results.

[0025] Knowledge layer: The rules, patterns, or causal relationships extracted from information, representing "how to do it", such as the core algorithm principle, chemical reaction equation, and circuit structure diagram.

[0026] Wisdom layer: The comprehensive application, judgment, and decision-making of knowledge, representing "why it is better to do it this way", such as algorithm optimization strategies, process improvement plans, and trade-off decisions to solve specific constraints.

[0027] The Purpose layer: This layer drives the ultimate goal or fundamental problem to be solved in the entire technical solution, representing "why do it," such as "improving image recognition accuracy" or "reducing device energy consumption."

[0028] Through this multi-layer semantic modeling, the system not only preserves the "form" of the technical solution, but also captures its "spirit," that is, the complete logical chain from the original data to the final invention intent2.

[0029] Blockchain-based Semantic Evidence Preservation: After generating the DIKWP semantic graph for each participant, the system calculates the structured data of the graph and the hash value (digital fingerprint) of the original evidence file. These hash values, along with metadata such as submission time and participant identities, are packaged into a transaction and broadcast to the blockchain distributed ledger for preservation. The decentralized nature, consensus mechanism, and chain structure of blockchain technology ensure that once the evidence record is on the chain, it cannot be tampered with or deleted by any single party, possessing extremely high credibility. This forms an undeniable, precisely timestamped multi-party semantic evidence library, fundamentally solving the problems of difficulty and easy dispute in traditional evidence preservation.

[0030] Quantitative analysis based on semantic responsibility path: This is the core analysis step of this invention. The system incorporates an innovative "data → intent semantic responsibility path" identification algorithm. This algorithm performs in-depth comparisons between the DIKWP maps submitted by each party. Its working principle is as follows:

[0031] Path Tracing: The algorithm traces backwards through the graphs of each participant, starting with the disputed technical goal (i.e., the intent layer node), to find all complete paths leading to that intent that connect the nodes at the levels of wisdom, knowledge, information, and data. This set of paths vividly depicts the complete innovation narrative of the participant: "What wisdom did they use, what knowledge did they rely on, what information did they process, and from which raw data did this information originate, in order to achieve a certain goal?"

[0032] Path Comparison and Contribution Assessment: The algorithm performs graph matching and difference analysis on the semantic paths of different participants. By comparing the path structure, the overlap of key nodes, and the uniqueness of the paths, the system can quantitatively assess:

[0033] In ownership disputes, the proportion of each party's contribution to the final invention at different semantic levels is crucial. For example, Party A might have contributed key knowledge layer nodes (core algorithms), while Party B might have contributed a large number of data layer nodes (training data). Based on this, the algorithm can provide an objective contribution allocation suggestion.

[0034] In infringement disputes, the degree of similarity between the semantic path of the accused infringing scheme and the semantic path protected by the patent claims is crucial. If the two paths highly overlap in key knowledge, intelligent nodes, and ultimate intent, the likelihood of infringement is high. This provides a data-driven quantitative basis for traditional "full coverage doctrine" and "equivalence doctrine."

[0035] Human-Machine Collaborative Multi-Party Semantic Game Evaluation: In order to balance the objectivity of analysis with the fairness and legality of the ruling, this invention designs a multi-party game evaluation model that integrates an AI evaluation engine and human arbitration experts.

[0036] AI Evaluation Engine: Based on the semantic path analysis results above, the AI ​​engine will automatically calculate a series of key indicators, such as semantic consistency, innovation path difference and purpose convergence, and generate an analysis report that includes preliminary ruling suggestions and detailed data support.

[0037] Human expert oversight: Human arbitration experts (who can be judges, patent examiners, or technical experts) review AI analysis reports, visualized semantic graphs, and path comparison results through a dedicated interactive interface. Experts can utilize their legal knowledge and industry experience to confirm, correct, or reject AI analyses, especially when dealing with legally specific concepts such as "prior art defense" and "estoppel doctrine."

[0038] Interactive Game Theory: This model supports a dynamic, courtroom-like "game" process. Either party can object to the opposing party's evidence graph or AI analysis and submit new evidence. The system receives new evidence in real time, re-analyzes semantics, uploads it to the blockchain, and performs path analysis, dynamically updating the AI ​​review report. This interactive adversarial and negotiation process fully guarantees the legitimacy of the procedure and the participation rights of all parties, ensuring that the final ruling is a consensus reached based on full information exchange and game theory.

[0039] Traceable award generation and automatic execution of smart contracts: After the arbitration process is completed, the system will generate a unique "traceable semantic link award".

[0040] Semantic Link Ruling: This ruling not only includes traditional textual conclusions but also visually presents a complete, highlighted logical reasoning chain from core evidence (key nodes in the DIKWP graph) to the final ruling. For example, the chain is clearly marked: "Because the knowledge layer node 'X' of the accused solution is semantically equivalent to the knowledge node 'Y' of the plaintiff's patent, and the intent node 'Z' they lead to is completely identical, infringement is determined." Each node can be clicked to trace its original evidence record on the blockchain, making the entire ruling process transparent, explainable, and verifiable.

[0041] Automatic execution of smart contracts: Once a ruling is confirmed by experts and recorded on the blockchain, it triggers a pre-deployed smart contract on the blockchain, automatically executing the corresponding legal consequences. For example:

[0042] Confirmation and enforcement of rights: If the invention is determined to be a joint invention, the smart contract will automatically modify the on-chain digital patent certificate and register each party as a joint rights holder according to their contribution ratio.

[0043] Indemnity / Licensing Fee Enforcement: If infringement is determined, the smart contract will automatically transfer the corresponding amount of compensation from the infringer's pre-deposited security deposit account to the rights holder's account.

[0044] Revenue distribution execution: For jointly owned patents, subsequent licensing revenues can be automatically and in real-time distributed to each rights holder according to the determined equity ratio through smart contracts.

[0045] This code-based automatic execution mechanism eliminates the risk of human delays or breaches of contract, greatly improving the efficiency and reliability of judgment enforcement, and is particularly revolutionary in solving the problem of enforcement difficulties in transnational disputes.

[0046] Corresponding to the method of this invention, this invention also provides a patented semantic arbitration system with multi-party participation. This system includes corresponding modules configured to execute the steps of the above-described method, such as participating party terminals, a semantic parsing module, a blockchain evidence storage module, a semantic analysis engine, a multi-party semantic game evaluation module (including an AI evaluation engine and an expert evaluation subsystem), an award generation module, and a smart contract execution module. These modules work together to form a complete, end-to-end intelligent arbitration platform.

[0047] Compared with existing technologies, this invention brings the following significant benefits by deeply integrating the DIKWP semantic model, blockchain technology, and human-machine collaborative intelligence:

[0048] Absolute reliability and comprehensiveness of evidence preservation: By parsing the technical solution into a multi-layered semantic graph (DIKWP) and storing it on the blockchain, this invention not only solidifies the original evidence but, more importantly, solidifies the deep semantic information behind the evidence, including the knowledge context, intellectual achievements, and ultimate intent of the invention. This multi-layered and tamper-proof evidence system provides a comprehensive and credible data foundation for arbitration, effectively preventing evidence tampering and subsequent denial, and ensuring the fairness of arbitration.

[0049] Objectivity and Precision in Contribution and Infringement Determination: The "Data → Intent" semantic responsibility path identification algorithm transforms the traditional reliance on expert subjective judgment in contribution classification and infringement comparison into a calculable and quantifiable data analysis process. By tracing and comparing innovation paths, it can accurately identify the key contribution nodes and innovation paths of each party in the invention, making patent ownership and infringement determination more scientific and accurate, and greatly reducing the uncertainty caused by subjective assumptions.

[0050] The arbitration process is highly efficient and cost-effective: From evidence submission, analysis, and comparison to the generation of preliminary ruling recommendations, most of the work is automated, significantly shortening the time required for traditional arbitration. The automatic execution mechanism of smart contracts further minimizes the time and manpower costs of the ruling execution process. The automation and intelligence of the entire process significantly reduces the overall cycle and cost of patent dispute resolution, making it easier for innovators to pursue their rights.

[0051] Transparency and interpretability of the ruling: This invention's unique "traceable semantic link ruling" transforms the "black box" ruling process into a completely transparent and logically clear presentation of reasoning. Parties involved can intuitively see how the ruling is derived step-by-step based on on-chain evidence, greatly enhancing their understanding and trust in the ruling, and improving the credibility of the arbitration institution.

[0052] Automation and enforceability of judgments: The introduction of smart contracts is a major innovation of this invention. It transforms legal judgments into machine-executable code, enabling immediate, automatic, and cross-regional enforcement of the judgment results. This fundamentally solves the chronic problem of "difficulty in enforcement" of traditional legal documents, ensuring that the legitimate rights and interests of the rights holders can be quickly realized, and upholding the dignity and effectiveness of the law.

[0053] Adaptability to Emerging Technological Challenges: The framework of this invention is specifically designed to address emerging scenarios such as AIGC and complex collaborations. Through refined tracing of semantic contributions, it effectively resolves cutting-edge legal challenges such as unclear ownership of AIGC inventions and the difficulty in assigning contributions from multiple stakeholders. This makes this invention not only an improvement on existing arbitration mechanisms but also a forward-looking and scalable infrastructure for intellectual property governance in the future AI era.

[0054] In summary, this invention provides a revolutionary solution to patent disputes. Through technological innovation, it systematically addresses the core pain points of traditional arbitration in terms of fairness, efficiency, transparency, and enforceability, and has extremely high practical application value and broad prospects for promotion. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the system architecture of a multi-party patent semantic arbitration platform according to an embodiment of the present invention.

[0056] Figure 2 This is a schematic diagram illustrating the process of multiple parties submitting semantic evidence and generating a DIKWP semantic graph and blockchain storage according to an embodiment of the present invention.

[0057] Figure 3 This is a flowchart illustrating a semantic responsibility path recognition algorithm according to an embodiment of the present invention.

[0058] Figure 4 This is a schematic diagram of the architecture of a multi-party semantic game evaluation model according to an embodiment of the present invention.

[0059] Figure 5 This is a schematic diagram illustrating the process of generating semantic links for adjudication results and executing smart contracts according to an embodiment of the present invention. Detailed Implementation

[0060] 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.

[0061] Example 1: Core System Architecture and Workflow

[0062] Reference Figure 1 This invention provides a multi-party participating patent semantic arbitration system 100, which constructs an end-to-end intelligent dispute resolution platform. The system 100 can be physically deployed in a cloud computing environment and consists of a series of collaborative software modules.

[0063] System architecture (refer to) Figure 1 )

[0064] The core architecture of System 100 includes:

[0065] Access and Presentation Layer 110: This layer is the portal for the system to interact with external users and data.

[0066] Multi-Party Participation Terminal 111: Provides a secure web or client interface for all parties involved in a dispute (such as plaintiffs, defendants, and third-party stakeholders). Users can submit case information, upload technical evidence (such as patent specifications, experimental records, source code, design drawings, etc.) through this terminal, and view case progress in real time, receive notifications, and participate in the negotiation and evaluation process.

[0067] Semantic parsing module 112: This is the key module for transforming raw evidence into structured data that machines can understand. It incorporates an advanced Natural Language Processing (NLP) engine (e.g., models based on Transformer architectures such as BERT), a Computer Vision (CV) module, and knowledge extraction algorithms. When a technical document is received, this module performs entity recognition (e.g., identifying technical terms and component names), relation extraction (e.g., identifying connections and causal relationships between components), and intent recognition. Subsequently, based on the DIKWP model theory, it precisely maps the extracted semantic elements to data, information, knowledge, wisdom, and intent. Figure 5 The process involves several levels, ultimately generating a semantically rich DIKWP graph.

[0068] Trusted Evidence Layer 120: This layer is responsible for ensuring the integrity and immutability of all evidence and process records.

[0069] Blockchain Evidence Storage Module 121: This module interacts with an underlying blockchain network (e.g., a consortium blockchain like Hyperledger Fabric to balance efficiency and privacy; or a public blockchain like Ethereum to achieve maximum decentralization and transparency). For each generated DIKWP graph and its original evidence file, this module calculates its SHA-256 hash value and submits this hash value, along with the participants' digital signatures, timestamps, and other information, as the transaction payload to the blockchain distributed ledger 122. Each record on ledger 122 is cryptographically linked to the previous record, forming an immutable chain of evidence.

[0070] Analysis and Decision Layer 130: This layer is the "brain" of the system, responsible for in-depth analysis of evidence and assisting in the generation of decisions.

[0071] Semantic Analysis Engine 131: This engine is the execution unit of the core algorithm, mainly including the semantic responsibility path identification algorithm 131a and the AI ​​evaluation model 131b. It securely retrieves the DIKWP graph stored by all parties from the blockchain ledger 122 to perform path tracing, comparison, and quantitative evaluation tasks.

[0072] Multi-party Semantic Game Evaluation Module 132: This is a collaborative module that tightly couples AI's computing power with human expertise. It includes an expert evaluation subsystem 132a, providing human arbitrators with a powerful dashboard for visually reviewing all data and analysis results and entering their professional opinions. Simultaneously, it continuously interacts with the AI ​​evaluation model 131b in the semantic analysis engine 131, forming a dynamic, human-machine collaborative deliberation loop.

[0073] Execution and Application Layer 140: This layer is responsible for translating the final decision into actual, automated actions.

[0074] Award generation module 141: After the arbitration is completed, this module automatically generates a structured electronic award containing a traceable semantic link based on the final decision output by the review module 132.

[0075] Smart Contract Execution Module 142: This module manages a series of smart contracts 143 that are pre-written and deployed on the blockchain ledger 122. Once the ruling is generated and finally confirmed (e.g., through the arbitrator's digital signature), the module parses the ruling content and calls the corresponding smart contracts (such as rights confirmation contracts, compensation contracts, profit-sharing contracts, etc.) to trigger their automatic execution on the chain.

[0076] External system interface 144: In order to achieve real-world linkage, the smart contract execution module 142 can also use mechanisms such as oracles to conduct secure data interaction with external patent registration systems, bank payment gateways, supply chain management platforms, etc., thereby mapping the results of on-chain operations to the off-chain world.

[0077] Workflow Details

[0078] The following will combine Figures 2 to 5 This paper details a complete workflow of the present invention. Consider the following scenario: Company A (the plaintiff) claims to own a patent for a "smart energy-saving algorithm" and accuses Company B (the defendant) of infringing its patent rights with a new product.

[0079] Step 1: Evidence Submission and Semantic Preservation (Refer to...) Figure 2 )

[0080] The process begins at step 201, where Company A and Company B log in to the arbitration platform through their respective participant terminals 111.

[0081] In step 202, each party submitted technical evidence to support its claims. Company A submitted the specification and claims of its authorized patent, as well as early R&D notes and algorithm prototype code. Company B submitted its product's technical white paper, pseudocode of the core algorithm, and performance test report.

[0082] In step 203, this evidence is sent to semantic parsing module 112. Taking Company A's patent as an example, module 112 performs the following parsing:

[0083] Data layer (D): Extract raw data such as "test environment temperature: 25℃, equipment load: 50%" from the experimental records.

[0084] Information layer (I): Extract performance indicators such as "under the above conditions, energy consumption is reduced by 15%" from the instruction manual.

[0085] Knowledge Layer (K): Extracts its core technical solution, namely a scheduling algorithm based on a "predictive control model", forming the nodes and edges of the knowledge graph.

[0086] The Wisdom Layer (W) identifies the inventive point, namely the optimization strategy of "dynamically correcting the prediction model by introducing weather forecast data".

[0087] Intent layer (P): It clarifies that the ultimate problem it aims to solve is "reducing the overall energy consumption of the data center under dynamic load".

[0088] Company B's evidence was also processed in the same way to generate its own DIKWP semantic graph.

[0089] In step 204, the blockchain evidence storage module 121 calculates the hash value for the DIKWP graph and original file generated by parties A and B.

[0090] In step 205, these hash values ​​are packaged into a blockchain transaction and written to the blockchain distributed ledger 122. Upon completion, the evidence from both A and B acquires a unique, immutable on-chain identity and timestamp, forming a trusted "multi-layered semantic ledger." The evidence preparation phase is now complete.

[0091] Step 2: Semantic Responsibility Path Identification and Contribution Assessment (refer to...) Figure 3 )

[0092] Arbitration enters the analysis phase; the semantic analysis engine 131 is activated and begins executing the semantic responsibility path identification algorithm, such as... Figure 3 As shown.

[0093] In step 301, the engine securely retrieves the DIKWP graphs of both A and B from the blockchain ledger 122.

[0094] In step 302, the graphs are normalized. For example, synonyms such as "energy consumption optimization" in graph A and "power saving control" in graph B are normalized to ensure the accuracy of the comparison.

[0095] In step 303, the focus of the analysis is determined, namely the shared intent node P: "Reducing data center energy consumption".

[0096] In step 304, the algorithm starts from the intention node P and performs reverse path search in the graphs of A and B respectively.

[0097] In the graph of A, find the path PathA: DA→IA→KA (predictive control)→WA (weather data correction)→P.

[0098] In the graph of B, find the path PathB: DB→IB→KB (fuzzy logic control)→WB (real-time load feedback)→P.

[0099] In step 305, a deep comparison is performed on the two paths. The AI ​​evaluation model 131b calculates the following key indicators, the design of which deeply integrates the principles of infringement determination in legal practice:

[0100]

[0101] In step 306, the engine synthesizes the above indicators to generate a quantitative analysis result. For example, the result might show: a low semantic consistency score (because KA and KB have different core algorithms), but a moderate innovation path difference score (because both dynamically adjust their control strategies based on external information), and an extremely high goal convergence score. This result will be used in the next step of game theory evaluation.

[0102] Step 3: Multi-party semantic game evaluation and adjudication (refer to...) Figure 4 )

[0103] The analysis results are sent to the multi-party semantic game evaluation module 132, such as... Figure 4 As shown.

[0104] In step 401, the AI ​​evaluation engine 131b automatically generates a preliminary ruling recommendation report based on the quantitative analysis results and submits it to the human arbitration expert panel. The report may state: "Option B does not constitute literal infringement, but its 'real-time feedback' strategy at the intelligence layer is equivalent to Option A's 'weather correction' strategy in achieving the function of dynamic adjustment, and there is a high probability that it constitutes equivalent infringement."

[0105] In step 402, the arbitration panel reviews the report through the interactive interface of the expert review subsystem 132a. The interface will display the DIKWP graphs and semantic paths of both parties A and B side by side in a visual manner (step 403), and highlight the similar and different nodes identified by AI.

[0106] In step 404, the expert panel conducts a review and interaction. The experts may discover an early internal email among the evidence submitted by Company B, indicating that it was already researching "real-time load feedback" technology before Company A's patent was published. The experts can then issue a challenge to Company B through the system, requesting it to submit more detailed R&D records as supplementary evidence (step 405).

[0107] In step 406, Company B submitted new evidence via the terminal. The system immediately parsed the new evidence, uploaded it to the blockchain, and triggered the semantic analysis engine 131 to run again (step 407).

[0108] In step 408, the AI ​​review engine updates its analysis report. The new report may show that Company B's technology path has an independent source and evolution process, thereby significantly reducing the possibility of equivalent infringement.

[0109] After several rounds of dynamic game-playing cycles involving "evidence submission - AI analysis - expert review - interactive questioning," in step 409, the expert panel, combining the final analysis results of the AI ​​with their own legal expertise, reached a consensus and made a final ruling: Company B did not constitute infringement. The experts input this final decision through subsystem 132a and confirmed it with a digital signature.

[0110] Step 4: Generation and Automatic Execution of the Ruling Result (Refer to...) Figure 5 )

[0111] Once the final ruling is confirmed, the system enters its final stage.

[0112] In step 501, the award generation module 141 is activated. Based on the final decision of the expert panel, it generates an electronic award containing a traceable semantic link.

[0113] In step 502, the ruling is presented in two forms: first, a traditional PDF legal document; and second, an interactive semantic reasoning chain 510 generated on a web interface. This chain clearly demonstrates: "Evidence B1 (Company B's R&D records, on-chain address 0x...) proves its independent R&D path (semantic path PathB) → AI analysis results (high path difference) → expert opinion (adopting AI analysis, combined with evidence B1, confirming independent development) → final ruling (no infringement)." Users can click on any link in the chain to trace back to the original evidence or analysis data stored on the blockchain (step 503).

[0114] In step 504, the smart contract execution module 142 parses the "no infringement" ruling.

[0115] In step 505, the module automatically invokes and triggers the preset "Case Closure" smart contract 143. The operations performed by this contract may include: releasing the deposits paid by both parties at the start of arbitration, marking the case status as "Judgmented - No Infringement" on the blockchain, and automatically sending a case closure notice to both parties. If the judgment results in infringement and requires compensation, the "Compensation Payment" contract will be triggered, automatically transferring funds from Party B's deposit account to Party A. The entire process is completed instantly without manual intervention.

[0116] Through the complete closed loop of the above four steps, the system and mechanism proposed in this invention transform a complex patent infringement dispute into a highly automated, data-driven, transparent, reliable, and automatically executed process, greatly improving the quality and efficiency of intellectual property dispute resolution.

[0117] Example 2: Application of the DIKWP Model in Ownership Disputes

[0118] The mechanism of this invention is also applicable to resolving complex patent ownership disputes, especially in scenarios involving AIGC or multi-person collaboration.

[0119] Imagine this scenario: Three researchers (Zhang San, Li Si, and Wang Wu) collaborate on a new material formulation using an AI platform. Zhang San provides the fundamental theoretical knowledge and preliminary chemical molecular structure hypotheses (contributing to the knowledge layer K); Li Si collects and organizes a large experimental database for training the AI ​​model (contributing to the data layer D and information layer I); and Wang Wu designs a unique AI model training strategy and parameter optimization scheme, and defines the final material performance target to be achieved (contributing to the intelligence layer W and intention layer P).

[0120] When disputes arise regarding authorship and equity allocation of patent inventors, they can be resolved through the system of this invention:

[0121] Evidence submission: Each of the three parties submits evidence of their contribution, such as Zhang San's paper draft, Li Si's dataset and processing script, and Wang Wu's model design document and experimental instructions.

[0122] DIKWP Graph Generation: The system generates DIKWP graphs for each of the three parties, clearly marking the semantic elements contributed by each person and the level at which they belong.

[0123] Semantic responsibility path analysis: Semantic analysis engine 131 aims at the final "new material formula" (a complex knowledge layer node) and "target performance" (an intent layer node), tracing and integrating the contribution paths of the three parties. The algorithm finds that to achieve the final result, the contributions of the three parties are indispensable, forming a convergent and complete "D→I→K→W→P" semantic network.

[0124] Contribution quantification: The algorithm calculates the quantitative contribution of Zhang San, Li Si, and Wang Wu to the entire invention based on a preset weight model (for example, the weight of original knowledge layer and wisdom layer nodes is higher than that of regular data layer nodes), such as 30%, 20%, and 50% respectively.

[0125] Game Theory Review and Adjudication: The AI ​​review engine generates a preliminary recommendation for the allocation ratio based on this contribution. After review by arbitration experts, if there are no objections, the allocation plan is confirmed.

[0126] Smart contract execution: The final ruling establishes that the patent rights are jointly owned by the three parties, with a rights ratio of 3:2:5. Smart contract execution module 142 will invoke the "Joint Ownership Confirmation" smart contract to generate a digital asset (such as an NFT) representing the patent rights on the blockchain, and distribute the share certificates (tokens) to the digital wallets of the three parties in the 3:2:5 ratio. Any future licensing revenue generated by this patent will be automatically distributed to the three parties in real-time according to this ratio via another "Revenue Distribution" smart contract.

[0127] In this way, the present invention transforms vague, qualitative descriptions of contributions into clear, quantitative criteria for the division of rights and interests, providing a powerful and fair technical tool for solving the problem of benefit distribution in collaborative innovation.

[0128] Example 3: Expansion and Variation of Technical Solution

[0129] The core idea and architecture of this invention have good scalability and can be adapted to different application scenarios and technical implementations.

[0130] Extending to other intellectual property areas: The arbitration mechanism of this system is not limited to patents. By adjusting the parsing model of the semantic parsing module 112 and the evaluation indicators of the semantic analysis engine 131, it can be easily extended to other intellectual property areas such as copyright and trademarks.

[0131] In the copyright field: The DIKWP semantic graph can be used to represent a work's plot (knowledge layer), character relationships (information layer), creative theme (intention layer), and artistic techniques (intellectual layer). When determining plagiarism, the system can compare the semantic graphs of two works to quantify their similarity in core plots and expressions.

[0132] In the trademark field: The DIKWP map can be used to represent a trademark's design concept (intent layer), visual elements (knowledge layer), and market awareness data (data / information layer). When determining trademark similarity and the likelihood of consumer confusion, the system can comprehensively analyze the similarity of these semantic dimensions.

[0133] Different technical implementation options:

[0134] Semantic parsing models: In addition to Transformer-based models, more specialized domain knowledge graph construction techniques can be used, or they can be combined with expert systems to improve the accuracy of parsing in specific technical fields.

[0135] Blockchain platforms can choose different underlying blockchain technologies based on varying requirements for performance, cost, privacy, and regulation. For example, in highly regulated sectors like finance or the judiciary, permissioned blockchains with government or authoritative institutions as nodes can be employed.

[0136] AI review models can incorporate more sophisticated AI technologies, such as reinforcement learning, allowing the AI ​​review engine to continuously learn and optimize its adjudication and recommendation capabilities through interaction with human experts; or federated learning can be introduced to train more accurate review models using multi-party data while protecting the data privacy of all parties.

[0137] Integration with the Judicial System: This system can function as an independent third-party commercial arbitration platform, or as an adjudicative tool for courts or official arbitration commissions. Its generated quantitative analysis reports and traceable semantic link awards serve as powerful references for judges or arbitrators in making final decisions, enhancing the efficiency and technological sophistication of judicial trials. Through prior judicial approval or party agreement, the electronic evidence and awards generated by this system can be granted legal effect.

[0138] As can be seen from the detailed description of the above embodiments, the multi-party patent semantic arbitration mechanism and system proposed in this invention is not a simple combination of existing technologies, but rather a comprehensive technical solution that systematically resolves intellectual property disputes in the AI ​​era through a deep and innovative integration of the DIKWP semantic model, blockchain, and artificial intelligence technologies. It combines the rigorous logic of law with the precise calculations of computers, paving a new path for achieving fairer, more efficient, and transparent intellectual property governance.

Claims

1. A method for multi-party involved patent semantic arbitration, characterized in that, The method comprises the following steps: receiving technical scheme evidence related to the disputed patent submitted by at least two parties respectively; automatically converting the technical scheme evidence into structured DIKWP semantic graphs through a semantic analysis module, wherein each DIKWP semantic graph comprises five semantic levels of data layer, information layer, knowledge layer, wisdom layer and intention layer, and represents the semantic evolution path from original data to final technical intention; storing the digital digest of the DIKWP semantic graph on a blockchain distributed ledger, generating an unalterable timestamp and storage record for each piece of evidence; using a semantic analysis engine to execute a semantic responsibility path identification algorithm, comparing and analyzing the DIKWP semantic graphs of each party stored on the blockchain to identify and quantify the semantic contribution of each party in realizing the technical intention or the semantic similarity between technical schemes; based on the analysis result of the semantic analysis engine, generating an arbitration decision through a multi-party semantic game evaluation model of man-machine cooperation; and automatically triggering and executing the preset on-chain operation through an intelligent contract execution module according to the arbitration decision.

2. The method of claim 1, wherein, The step of automatically converting the technical scheme evidence into structured DIKWP semantic graphs specifically comprises: using natural language processing technology to perform entity recognition, relationship extraction and intention analysis on the technical scheme evidence; mapping the identified original data, experimental parameters or basic materials to the data layer; mapping the identified technical indicators, context descriptions or data processing results to the information layer; mapping the identified core technology principles, algorithm models or structural features to the knowledge layer; mapping the identified optimization strategies, technical know-how or systematic solutions to the wisdom layer; and mapping the identified problems to be solved by the technology, application targets or design intentions to the intention layer.

3. The method of claim 1, wherein, The step of the semantic responsibility path identification algorithm specifically comprises: obtaining the DIKWP semantic graphs of each party from the blockchain and performing semantic standardization processing; determining a target intention node related to the focus of the dispute; starting from the target intention node, traversing to the data layer node in each party's DIKWP semantic graph in reverse, forming at least one complete semantic responsibility path from the data layer to the intention layer; comparing the semantic responsibility paths of different parties, calculating the semantic consistency, innovation path difference and purpose convergence indexes between the paths; according to a preset weight model, assigning contribution weights to the nodes at different semantic levels in the path, and combining the indexes to calculate the final quantitative contribution degree or infringement possibility score of each party.

4. The method of claim 1, wherein, The multi-party semantic game evaluation model generates an arbitration decision, specifically including: an AI evaluation engine automatically generates an analysis report containing preliminary arbitration suggestions and reasons according to the quantified contribution degree or infringement possibility score; the analysis report, the DIKWP semantic map and the semantic responsibility path of each participant are visually presented to a human arbitration expert; the human arbitration expert audits and modifies the analysis report, and can require any participant to supplement or clarify evidence, and the semantic analysis engine updates the analysis report in real time according to the new evidence; after at least one round of evaluation interaction, the arbitration decision is finally confirmed and output by the human arbitration expert.

5. The method of claim 1, wherein, The method further comprises generating a traceable semantic link decision after generating the arbitration decision, which displays the logical reasoning path from the key evidence node to the final decision in a graphical manner, and labels the source, timestamp and storage address on the blockchain of each node on the path.

6. The method of claim 1, wherein, The step of automatically triggering and executing the preset on-chain operation by the smart contract execution module includes at least one of the following: if the arbitration decision is to determine the joint ownership of the patent right, call the right confirmation smart contract to generate digital right certificates for each participant in proportion to their contribution on the chain; if the arbitration decision is to determine that the infringement is established, call the compensation smart contract to automatically transfer the pre-frozen deposit or the funds in the specified account to the right holder according to the adjudicated amount; if the arbitration decision is to authorize the license, call the license smart contract to automatically record the license terms and periodically execute the payment of the license fee according to the agreement.

7. A multi-party engaged patent semantic arbitration system, characterized by, Comprise: One or more participant terminals for receiving technical solution evidence submitted by at least two participants respectively in relation to the disputed patent; A semantic analysis module connected with the participant terminal, configured to automatically convert the technical solution evidence into a structured DIKWP semantic map, wherein each DIKWP semantic map comprises five semantic levels of data layer, information layer, knowledge layer, wisdom layer and intention layer; a blockchain storage module configured to store a digital digest of the DIKWP semantic map on a blockchain distributed ledger and generate an unalterable timestamp; a semantic analysis engine configured to execute a semantic responsibility path identification algorithm to compare and analyze the DIKWP semantic maps of each participant stored on the blockchain to quantify the semantic contribution degree of each participant or the semantic similarity between technical solutions; a multi-party semantic game evaluation module including an AI evaluation engine and an expert evaluation subsystem, configured to generate an arbitration decision based on the analysis results of the semantic analysis engine; A smart contract execution module configured to automatically trigger and execute the preset on-chain operation according to the arbitration decision.

8. The system of claim 7, wherein, The semantic analysis engine is further configured to calculate at least one of the following indicators: semantic consistency: by comparing the coincidence of the structure and nodes of the DIKWP semantic graph of the accused infringement scheme and the DIKWP semantic graph corresponding to the disputed patent claim, the similarity of the technical content of the two is evaluated; innovation path difference: by comparing the uniqueness of the semantic responsibility path adopted by each party to achieve the same or similar technical intention, the independent innovation degree of the technical scheme is evaluated; purpose convergence: by comparing the semantic similarity of the intention layer nodes of each party's technical scheme, whether the application goals of the technical scheme are coincident is evaluated.

9. The system of claim 7, wherein, The expert review subsystem provides an interactive user interface that allows human arbitration experts to view visualized DIKWP semantic graphs, semantic responsibility path comparison results, and analysis reports generated by the AI review engine, and allows the experts to input review opinions, initiate interrogations of the parties, or require supplementary evidence.

10. The system of claim 7, wherein, The system further includes a ruling generation module configured to integrate the arbitration decision and its reasoning basis into an electronic ruling containing interactive and graphical semantic reasoning links, and each node on the semantic reasoning link is linked to its original evidence record on the blockchain.

Images