Artificial intelligence-based patent pool operation management method and system

Abstract

The present application relates to the technical field of operation management, in particular to a patent pool operation management method and system based on artificial intelligence, comprising the following steps: extracting patent text content analysis parameter behavior, identifying function jump position and cutting to generate semantic fragments, constructing vectors, judging the relationship between fragments, analyzing the permission stage and adjusting the weight configuration, analyzing member behavior to adjust the contribution weight, and identifying the conflict structure of permission request. In the present application, by constructing a directional vector and performing projection analysis, combining the keyword overlap ratio to establish a semantic connection relationship, using the permission stage and legal state label joint verification to adjust the parameter structure, using the behavior operation proportion relationship to map the member weight, optimizing the extraction and organization mode of semantic information, enhancing the response capability to the change of the permission stage, improving the precision of member behavior feature recognition, making the path conflict judgment more targeted and adaptive, and enhancing the dynamic operation management capability of the patent pool under the coupling of multiple factors.

Description

Technical Field

[0001] This invention relates to the field of operations management technology, and in particular to a patent pool operations management method and system based on artificial intelligence. Background Technology

[0002] The field of operations management technology encompasses systematic methods and technologies for coordinating, configuring, and scheduling various resources within an organization to achieve efficient operation and strategic goals. It focuses on improving organizational management efficiency and execution capabilities at the operational level through process management, decision support, resource allocation, and project tracking. This covers multiple aspects, including enterprise operational process modeling, task scheduling mechanisms, management information system architecture, and data analysis-supported decision-making models. It involves task prioritization, execution path optimization, resource consumption balancing, and risk assessment. The overall structure is typically built on a process-driven or data-driven framework, possessing the ability to describe and control the collaborative relationships of multiple roles and departments. Among these, the patent pool operations management method based on artificial intelligence refers to... In operation and management activities, a specific management mechanism is adopted to model, classify, and optimize patent information, member behavior, and collaboration rules using artificial intelligence technology. This mechanism covers patent ownership identification, patent value assessment, licensing pricing strategies, formulation and adjustment of collaboration rules among members, and the establishment of a network of relationships between patents using knowledge graph models to support patent portfolio construction and licensing strategy formulation. Specifically, it achieves the classification, integration, and dynamic management of patent resources in the patent pool by constructing a patent classification standard model, establishing a patent evaluation index system based on machine learning, designing licensing allocation rules based on patent contribution, using natural language processing to perform semantic analysis of patent texts, and adjusting weight factors based on member participation and patent cross-use frequency.

[0003] Traditional patent pool operation and management technologies employ patent evaluation and licensing allocation methods based on rule templates and experience data. In actual operation, these methods suffer from inconsistent handling when faced with semantic ambiguity, parameter overlap, and differences in behavior at different stages. They lack a structured identification mechanism for continuous behavioral segments in patent texts, making it difficult to quantitatively distinguish the operational characteristics of different members during dynamic licensing. Furthermore, the criteria for determining licensing conflicts rely on static comparison of path node information, failing to reflect the dynamic changes of multi-dimensional elements in licensing requests in real time during path structure adjustments. This can easily lead to judgment errors under conditions of geographical overlap or overlapping uses, affecting the fairness and accuracy of licensing allocation results. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide a patent pool operation and management method and system based on artificial intelligence. The technical solution is as follows:

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an artificial intelligence-based patent pool operation and management method, comprising the following steps:

[0006] S1: Extract the patent text content, analyze and compare the operation verbs and objects of adjacent continuous parameters, compare the semantic structure and target object category, identify the location of the conversion behavior, compare the verb sequence before and after the conversion point with the input and output terms, identify the function jump position and cut it, and generate a semantic fragment sequence;

[0007] S2: Based on the semantic segment sequence, obtain the keyword sequence and encode it to construct a direction vector. Project the vector onto the reference axis, compare the projection angle offset between semantic segments, determine the cross relationship based on the keyword overlap ratio, establish the semantic connection relationship of multiple segments, and generate semantic difference determination information.

[0008] S3: Based on the semantic difference determination information, analyze the interval between the time tag of the permission node and the call record, determine the permission stage and verify the status tag, adjust the parameter combination structure of the call impact and stability impact, and generate dynamic proportion control data;

[0009] S4: Based on the dynamic proportion control data, obtain member operation records, identify the proportion relationship of behavior operations in the operation set, determine the dominant behavior characteristics and perform weight adjustment, update the proportion structure of members in license revenue, and generate behavior contribution weight configuration.

[0010] S5: Based on the behavior contribution weight configuration, obtain the regional number, usage restriction and time period in the subject's permission request, analyze the intersection structure, determine the overlap state and mark the conflict segment, perform segmentation processing on the path structure, adjust the connection method and boundary classification structure between path segments, and generate path structure conflict distribution data.

[0011] As a further aspect of the present invention, the semantic segment sequence includes semantic boundary positions, verb sequence indexes, and input / output term comparisons; the semantic difference determination information includes projection angle offset parameters, keyword overlap ratios, and segment connection relationships; the dynamic proportion control data specifically includes call influence proportion structure, stability influence proportion structure, and stage weight allocation parameters; the behavior contribution weight configuration includes submission frequency proportion, combined task frequency proportion, and collaborative request operation proportion; and the path structure conflict distribution data specifically refers to regional intersection distribution, overlapping usage segments, and overlapping time period segments.

[0012] As a further aspect of the present invention, the step of obtaining the semantic segment sequence specifically includes:

[0013] S101: Obtain the patent text content, detect the operation verbs and objects corresponding to continuous parameters, analyze the semantic structure and target object categories between adjacent parameters, compare the changes in operation verbs and target objects between parameters, record the position of each category conversion, and generate a conversion behavior position index;

[0014] S102: Based on the conversion behavior position index, compare the verb sequences and input / output terms before and after the conversion point, determine the synchronous relationship of changes in operation flow, target object and input / output terms, filter function jump positions, and use the jump points as segmentation markers to obtain a semantic boundary position array;

[0015] S103: Based on the semantic boundary position array, adjust the text vector sequence structure, perform semantic segmentation of the text content, and divide the structure by combining the boundary position of each segment to obtain a semantic fragment sequence.

[0016] As a further aspect of the present invention, the process of the filtering function jumping position is specifically as follows:

[0017] Based on the conversion behavior location index, a symmetrical acquisition window with a preset parameter step size is set. The changes in verb sequences, input and output terms and operation flow before and after the conversion point within the window are compared. By judging whether the synchronous change ratio of the three exceeds the corresponding threshold, the target conversion point is marked as the function jump position.

[0018] The corresponding thresholds include semantic synchronization threshold, flow direction change threshold, and object switching threshold. The thresholds are calculated by selecting a set of positions in the patent text content that do not have category conversion as the baseline set. The distribution of verb sequence difference ratio, input / output term replacement ratio, and operation flow direction change ratio in the target set are statistically analyzed. The semantic synchronization threshold, flow direction change threshold, and object switching threshold are set at the upper quantile boundary of each ratio.

[0019] As a further aspect of the present invention, the step of obtaining the semantic difference determination information specifically includes:

[0020] S201: Based on the semantic segment sequence, obtain the keyword set corresponding to each semantic segment, count the first occurrence position of each group of keywords in the original text and arrange them in order, assign the sequential positions according to the sequence number to construct the direction vector, and obtain the sequential encoding vector group;

[0021] S202: Based on the sequential encoding vector group, call the direction reference axis vector sequence, compare the projection angle offset between multiple semantic segment vectors, combine the overlap ratio between keyword lists, calculate the cross-determination coupling value, and obtain the cross-determination coefficient group;

[0022] S203: Based on the sorting results of the cross-determination coefficient group, identify the connection paths between semantic segments and mark the cross-type, establish the semantic connection relationship of multiple segments, and generate semantic difference determination information.

[0023] As a further aspect of the present invention, the step of obtaining the dynamic proportion control data specifically includes:

[0024] S301: Based on the semantic difference determination information, analyze the time tag and call record in the license node, compare the interval span between the time tag and the call record, determine the current license stage of the patent, and generate a license stage interval;

[0025] S302: Call the aforementioned licensing stage interval, verify the stage judgment result by combining the legal status label of the license, and adjust the combination structure of the call influence ratio and the stability influence ratio in the parameter weight according to the licensing stage of the patent to obtain the stage weight adjustment interval;

[0026] S303: Based on the stage weight adjustment range, perform proportional allocation of the corresponding parameter weights for each stage, encode the weight combination relationship of multiple stages, construct a weight allocation structure, and establish dynamic proportion control data.

[0027] As a further aspect of the present invention, the step of obtaining the behavior contribution weight configuration specifically includes:

[0028] S401: Based on the dynamic proportion control data, obtain the number of times a member submits patent content, the frequency of participating in the combination construction task, and the operation record of submitting collaboration requests within the licensing period, identify the proportion relationship of each type of behavior operation in the member operation set, and generate the member behavior proportion coefficient.

[0029] S402: Based on the member behavior proportion coefficient, compare the proportions of various behaviors, determine the dominant operation characteristics among multiple behavior categories, filter the dominant operation category, and obtain the dominant operation category determination value;

[0030] S403: Based on the dominant operation category determination value, perform contribution bias mapping based on the weight dimension corresponding to the dominant behavior, perform weight adjustment and update the proportion structure of members in licensing revenue, and establish behavior contribution weight configuration.

[0031] As a further aspect of the present invention, the step of obtaining the path structure conflict distribution data specifically includes:

[0032] S501: Based on the behavior contribution weight configuration, obtain the regional number information, purpose limitation field and time period description in each subject's license request, analyze the overlap range of regional number intersection, purpose keywords and time period segments between each subject's license requests, and obtain the license parameter overlap structure quantity;

[0033] S502: Based on the overlapping structure quantity of the permission parameters, compare the regional overlap and time period repetition of the permission path segments, judge the overlapping state of the overlapping areas in the path flow, mark the distribution position of each overlapping area in the path structure, and obtain the conflict segment distribution index.

[0034] S503: Based on the conflict segment distribution index, segment the marked conflict segments in the path, adjust the connection method and boundary classification relationship between path segments, and establish path structure conflict distribution data.

[0035] On the other hand, an AI-based patent pool operation and management system is provided, which is applied to an AI-based patent pool operation and management method. This system includes:

[0036] The semantic segmentation module extracts the patent text content, analyzes and compares the operation verbs and objects of adjacent continuous parameters, compares the semantic structure and target object category, identifies the location of the conversion behavior, compares the verb sequence before and after the conversion point with the input and output terms, identifies the function jump position and segments it, and generates a semantic fragment sequence.

[0037] The vector projection module obtains the keyword sequence and encodes it to construct a direction vector based on the semantic segment sequence. It then projects the vector onto the reference axis, compares the projection angle offset between semantic segments, determines the cross relationship based on the keyword overlap ratio, establishes the semantic connection relationship of multiple segments, and generates semantic difference determination information.

[0038] The stage judgment module analyzes the interval between the time tag of the permission node and the call record based on the semantic difference judgment information, judges the permission stage and verifies the status tag, adjusts the parameter combination structure of call impact and stability impact, and generates dynamic proportion control data.

[0039] The behavior recognition module obtains member operation records based on the dynamic proportion control data, identifies the proportion relationship of behavior operations in the operation set, determines the dominant behavior characteristics and performs weight adjustment, updates the proportion structure of members in license revenue, and generates behavior contribution weight configuration.

[0040] The path control module obtains the regional number, usage restriction and time period in the subject's permission request based on the behavior contribution weight configuration, analyzes the intersection structure, determines the overlap state and marks the conflict segment, segments the path structure, adjusts the connection method and boundary classification structure between path segments, and generates path structure conflict distribution data.

[0041] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0042] By constructing direction vectors and performing projection analysis, semantic connection relationships are established by combining keyword overlap ratios. The parameter structure is adjusted by jointly verifying licensing stage and legal status labels. Member weights are mapped by utilizing the proportion of behavioral operations. The extraction and organization of semantic information are optimized, enhancing the responsiveness to changes in licensing stages and improving the accuracy of member behavioral feature recognition. This makes path conflict judgment more targeted and adaptable, and enhances the dynamic operation and management capabilities of the patent pool under the coupling of multiple factors. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0045] Figure 2 This is a detailed flowchart of S1 of the present invention;

[0046] Figure 3 This is a detailed flowchart of the S2 process of the present invention;

[0047] Figure 4 This is a detailed flowchart of the S3 process of the present invention;

[0048] Figure 5 This is a detailed flowchart of the S4 process of the present invention;

[0049] Figure 6 This is a detailed flowchart of S5 of the present invention;

[0050] Figure 7 This is a system flowchart of the present invention. Detailed Implementation

[0051] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0052] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0053] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent.

[0054] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0055] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0056] Please see Figure 1 This invention provides a technical solution: a patent pool operation and management method based on artificial intelligence, comprising the following steps:

[0057] S1: Extract the patent text content, analyze and compare the operation verbs and objects of adjacent continuous parameters, compare the semantic structure and target object category, identify the location of the conversion behavior, compare the verb sequence before and after the conversion point with the input and output terms, identify the function jump position and cut it, and generate a semantic fragment sequence;

[0058] S2: Based on the semantic segment sequence, obtain the keyword sequence and encode it to construct a direction vector. Project the vector onto the reference axis, compare the projection angle offset between semantic segments, determine the cross relationship based on the keyword overlap ratio, establish the semantic connection relationship of multiple segments, and generate semantic difference determination information.

[0059] S3: Based on semantic difference judgment information, analyze the interval between the time tag of the permission node and the call record, determine the permission stage and verify the status tag, adjust the parameter combination structure of call impact and stability impact, and generate dynamic proportion control data;

[0060] S4: Based on the dynamic proportion adjustment data, obtain member operation records, identify the proportion relationship of behavior operations in the operation set, determine the dominant behavior characteristics and perform weight adjustment, update the proportion structure of members in license revenue, and generate behavior contribution weight configuration.

[0061] S5: Based on the behavior contribution weight configuration, obtain the regional number, usage restriction and time period in the subject's permission request, analyze the intersection structure, determine the overlap status and mark the conflict segment, segment the path structure, adjust the connection method and boundary classification structure between path segments, and generate path structure conflict distribution data.

[0062] The semantic fragment sequence includes semantic boundary positions, verb sequence indexes, and input / output term comparisons. Semantic difference determination information includes projection angle offset parameters, keyword overlap ratios, and fragment connection relationships. Dynamic proportion control data specifically includes call impact proportion structure, stability impact proportion structure, and stage weight allocation parameters. Behavioral contribution weight configuration includes submission frequency proportion, combined task frequency proportion, and collaborative request operation proportion. Path structure conflict distribution data specifically refers to regional intersection distribution, paragraphs with overlapping uses, and overlapping fragments with time periods.

[0063] Please see Figure 2 The specific steps for obtaining the semantic fragment sequence are as follows:

[0064] S101: Obtain the patent text content, detect the operation verbs and objects corresponding to continuous parameters, analyze the semantic structure and target object categories between adjacent parameters, compare the changes in operation verbs and target objects between parameters, record the position of each category conversion, and generate a conversion behavior position index;

[0065] To obtain the patent text content, such as "a semiconductor wet etching process," which contains a total of 15,000 characters, firstly, extract continuous technical parameters from the claims and specification sections of the text. Specifically, identify the parameter "etching solution temperature" at character position 350, with a value of "". Following this, the parameter "etching time" was identified at character position 380, with a value of "". Then, the operation verbs and their objects corresponding to these two consecutive parameters are analyzed. The operation verb associated with "etching solution temperature" is identified as "heating," and its object is "etching solution." The operation verb associated with "etching time" is "immersion," and its object is "silicon wafer." Next, by comparing the semantic structure and target object category in the actions associated with adjacent parameters, it is determined that the target object category of "heating the etching solution" belongs to "process condition control," while the target object category of "immersing the silicon wafer" belongs to "wafer processing." Since "process condition control" and "wafer processing" are two different categories, this change is recorded, and the character position 380 where the category transition occurs is marked. Continuing to process the text further, the parameter "photoresist thickness" is identified at character position 850, with a value of "". The operation verb is "spin coating", the target is "silicon wafer", and the parameter "nitrogen flow rate" is identified at character position 890, with a value of "". The operation verb is "purge", and the target is "process chamber". Here, the target category is changed from "wafer processing" to "environmental control". Therefore, the marked character position 890 is another category conversion position. All character positions where category conversion occurs, such as 380, 890, 1520, etc., are arranged in the order of their appearance in the text to generate a conversion behavior position index.

[0066] S102: Based on the conversion behavior position index, compare the verb sequences and input / output terms before and after the conversion point to determine the synchronization relationship between the operation flow, target object and input / output terms, filter the function jump position, and use the jump point as a segmentation mark to obtain the semantic boundary position array;

[0067] Based on the conversion behavior location index [380, 890, 1520, ...], the first conversion point 380 in the index is processed first. A symmetrical acquisition window with a preset parameter step size is set. This step size is determined by statistical analysis of 100 patent documents in the same field, taking 50% of the average length of the technical step description. Each character, specifically, a symmetrical acquisition window with a range of [330, 430] is defined centered at character position 380. Then, the changes in verb sequences, input / output terms, and operation flow before and after the transition point 380 within the window are compared. Before the transition point, within the interval [330, 379], the identified verb sequences are [heating, maintaining, monitoring], the input / output terms are [heater power, temperature sensor reading], and the operation flow is "equipment to solution". After the transition point, within the interval [380, 430], the verb sequences are [immersion, shaking, removing], the input / output terms are [robotic arm position, etched wafer], and the operation flow is "wafer to solution". Then, by determining whether the synchronous change ratio of the three exceeds the corresponding threshold, the function jump position is filtered. The process of setting the corresponding threshold is as follows: In the patent text content, select... A set of position points that do not contain any category transformations, such as character positions 200, 500, 1100, etc., constitutes a baseline set. Similarly, a set is configured for each position point within this baseline set. A symmetrical window containing 1 character is used to calculate the Jaccard similarity of verb sequences within the window, the overlap of input and output terms, and the vector angle of the operation flow. This yields the verb sequence difference ratio, the input / output term replacement ratio, and the operation flow change ratio. For example, the 10 sets of verb sequence difference ratios calculated in the baseline set are [0.1, 0.15, 0.05, 0.2, 0.1, 0.1, 0.05, 0.15, 0.2, 0.15]. These are sorted, and the upper quartile (the 8th value) is used as the boundary for the semantic synchronization threshold. The semantic synchronization threshold is set to [value missing]. (Difference, 1 - Similarity), similarly, the threshold for change in flow direction is set as... The object switching threshold is Returning to the analysis of transition point 380, the verb sequence difference ratio is calculated as follows: The input / output term replacement ratio is The proportion of operation flow change is ,because , , Since all three change ratios exceed the corresponding thresholds, the transition point 380 is marked as a functional jump position. This process is repeated for all transition points in the index. All transition points marked as functional jump positions, such as [380, 890, 1520], are integrated and processed to obtain a semantic boundary position array.

[0068] S103: Based on the semantic boundary position array, adjust the text vector sequence structure, perform semantic segmentation of the text content, and divide the structure by combining the boundary position of each segment to obtain the semantic fragment sequence.

[0069] Based on the semantic boundary position array [380, 890, 1520], the vectorized representation sequence of the original patent text is structurally adjusted. This text vector sequence was initially a continuous vector stream generated by sentences or a fixed length. Now, based on the precise character positions recorded in the semantic boundary position array, the vector sequence is re-aligned and segmented to ensure that the boundaries of each vector subsequence perfectly match the boundaries of functional jumps. Specifically, the content of the original text from character position 0 to 379 and its corresponding vector set are defined as the first semantic segment, the content from character position 380 to 889 and its vector set are defined as the second semantic segment, and the content from character position 890 to 1519 and its vector set are defined as the third semantic segment. Three semantic segments are used, and so on, to complete the semantic segmentation of the entire text. After segmentation, each segment is assigned a structured attribute, and the structure is divided based on the boundary position of each segment. For example, segment one is marked as {ID:Seg01,Start:0,End:379,Type:Process Condition Control}, segment two is marked as {ID:Seg02,Start:380,End:889,Type:Wafer Processing}, and segment three is marked as {ID:Seg03,Start:890,End:1519,Type:Environmental Control}. All these structured segments are arranged according to their order of appearance in the original text to obtain the semantic segment sequence.

[0070] Please see Figure 3 The specific steps for obtaining semantic difference determination information are as follows:

[0071] S201: Based on the semantic segment sequence, obtain the keyword set corresponding to each semantic segment, count the first occurrence position of each group of keywords in the original text and arrange them in order, assign the sequential positions according to the sequence number to construct the direction vector, and obtain the sequential encoding vector group;

[0072] Based on the semantic segment sequence [Seg01, Seg02, Seg03], the first semantic segment is processed first. By using the TF-IDF algorithm and combining it with a domain dictionary, the corresponding keyword set is extracted from its text content [0:379]. Next, the positions of the first occurrence of these keywords in the original patent text were counted. The results showed that "etching" first appeared at character position 56, "solution" at 81, "temperature" at 350, "heating" at 365, and "control" at 120. These keywords were then sorted in ascending order according to their first occurrence position, resulting in the ordered sequence [etching, solution, control, temperature, heating]. This ordered position was then assigned values ​​1, 2, 3, 4, 5 to construct... The direction vector is obtained. For the second semantic segment Performing the same operation, its keyword set is The first occurrences of these terms in the original text are at positions 150, 390, 56, 410, and 910, respectively. The sorted sequence is [etching, wafer, immersion, shaking, drying], and the corresponding direction vector is... For the third semantic segment Performing the same operation, its keyword set is The first occurrences are at positions 895, 892, 900, 910, and 150. The sorted sequence is [wafer, nitrogen, chamber, purging, drying], with the corresponding direction vectors being... , generate direction vectors for all semantic segments When these are combined, a sequential encoding vector group is obtained.

[0073] S202: Based on the sequential encoding vector group, the directional reference axis vector sequence is called to compare the projection angle offset between multiple semantic segment vectors. Combined with the overlap ratio between keyword lists, the formula is used:

[0074] ;

[0075] Calculate the cross-determination coupling value to obtain the cross-determination coefficient set;

[0076] in, For the first The semantic fragment and the first The cross-determination coupling value between semantic segments is dimensionless and is obtained by calculating the ratio of the intersection of the normalized value of the directional offset angle and the keyword intersection. For the first The semantic fragment and the first The normalized value of the projection angle between the direction vectors of each semantic segment is obtained by taking the projection angle between two direction vectors and dividing it by a preset maximum reference angle. For the first The normalized value of the number of keywords in a semantic segment is obtained by counting the length of the keyword sequence and dividing it by the maximum number of keywords in the semantic segment set. For the first The normalized value of the number of keywords in a semantic segment is obtained by counting the length of the keyword sequence and dividing it by the maximum number of keywords in the semantic segment set. For the first The semantic fragment and the first The normalized value of the intersection of the keyword sets of two semantic segments is obtained by extracting the intersection of the keyword sequences of the two segments and dividing it by the maximum number of keywords in the semantic segment set. This is the index number of the semantic segment to be calculated. This is the index number of the currently compared semantic segment;

[0077] Based on sequential encoding vector group ,in , It then invokes a sequence of direction reference axis vectors, which is a standard direction benchmark generated by statistical modeling the technological evolution paths of a large number of core patents in the field. Here, one of the principal axes is selected. ,Compare and To determine the projection angle offset between these two semantic segment vectors, first calculate the cosine of the angle between the two vectors. Then the included angle Approximately radians, i.e. The preset maximum reference angle is determined based on the 90th percentile of the statistical distribution of the included angles of patent vectors in unrelated technical fields, and is set as follows: The normalized value of the projected angle is obtained after normalization. Next, combine the keyword lists from the two segments. and The overlap ratio between segments was analyzed, and the segments containing the most keywords in the entire semantic segment set were identified. The maximum number of keywords is 8. The number of keywords is After normalization , The number of keywords is After normalization The intersection of the two sets is {etch}, and the number of etched sets is . After normalization Then, the formula was used. ; Calculate the cross-determination coupling value, the formula is based on the difference in direction. Differences in content Multiplication increases the coupling value when there is significant directional divergence or minimal content overlap, thus quantifying the semantic cross-strength between segments. Substituting the numerical values ​​from the above examples into the formula:

[0078] ;

[0079] The cross-coupling value is a numerical indicator used to measure the semantic cross-strength between any two semantic segments in terms of both structural evolution direction and content overlap. A higher value indicates a more significant divergence between the two segments in their semantic evolution paths, with less content overlap. This value serves as an edge selection criterion when building a semantic connection graph. It can identify semantic segment combinations that have shifted in structural direction and are no longer consistent in content. By eliminating connection paths with excessively high coupling values, it helps to accurately separate topic evolution nodes and delineate the boundary positions of multiple technical modules in a patent. This provides a clear structural input foundation for subsequent operations such as licensing path planning, modular combination strategies, and semantic clustering, significantly improving the accuracy and applicability of semantic structure recognition. These results demonstrate that… and The low cross-determination coupling value between them indicates that they have a certain degree of semantic continuity. By calculating the cross-determination coefficients between all fragments in the patent pool, a set of cross-determination coefficients is obtained.

[0080] S203: Based on the sorting results of the cross-determination coefficient group, identify the connection path between semantic segments and mark the cross type, establish the semantic connection relationship of multiple segments, and generate semantic difference determination information;

[0081] Based on the cross-determination coefficient set, this array is a The matrix, where The total number of semantic fragments, the matrix Location storage The value, for example, a partial value obtained after calculation. , , The array is sorted in ascending order for each row to identify the other segments most closely connected to each semantic segment. A connection threshold is set, determined based on statistics of the coupling value distribution, using the 25th percentile, and set to 0.20. ,because Then in and Establish a connection path between them, and If they are not directly connected, then no direct connection is established between them. Next, the connection path is marked with an intersection type, and the marking rule is: if If so, it is marked as "strong association - sequential evolution"; if If so, it is marked as a "weakly associated parallel module"; if If it is, it will be marked as "unrelated - topic jump". and The connections were labeled as "strong association-sequential evolution". and Connection ( It is labeled as a "weakly associated parallel module". By performing this determination on all fragment pairs, a graph structure containing all fragment nodes and typed connection edges is established, that is, the semantic connection relationship of multiple fragments. This graph structure and the attributes of all its nodes, the weights and types of its edges are stored together to generate semantic difference determination information.

[0082] Please see Figure 4 The specific steps for obtaining dynamic proportion control data are as follows:

[0083] S301: Based on semantic difference determination information, analyze the time stamps and call records in the licensing node, compare the interval span between the time stamps and call records, determine the current licensing stage of the patent, and generate a licensing stage interval;

[0084] Based on the semantic difference determination information, the technical topic modules are constructed. One core technical module, "Wafer Etching Process" (including Seg02, Seg04, and Seg05), is selected for analysis. First, all licensing node records related to the patents within this module are retrieved, as shown in Table 1.

[0085] Table 1. Patent P8890 Licensing Record Table

[0086] Call Record ID Time tags Call type L001 2023-08-15 Due diligence inquiry L002 2024-06-20 Licensing Intent Contact L003 2025-01-10 Composite Builder Call L004 2025-02-15 Composite Builder Call L005 2025-03-05 Licensing Intent Contact

[0087] As shown in Table 1, the time stamps and call records in the permission node are analyzed, and the time stamp interval between two adjacent call records is compared. and The interval is sky, and The interval is sky, and The interval is sky, and The interval is On that day, it was observed that the interval span showed a clear and accelerating shortening trend. Then, the current stage of the patent licensing was determined, and the stage division criteria were: the interval span was greater than A period of time during which the number of days does not show a shortening trend is defined as a "dormant period"; the interval span is... Heavenly The period between days, or a period that shows a slow shortening trend, is defined as the "introduction period"; the interval span is... Heavenly A period of time that is shortening rapidly between days is defined as the "growth period"; intervals shorter than [a certain number of days] are defined as [a certain number of days]. If the interval remains stable, it is defined as the "maturity period". Based on the currently calculated interval span sequence [310,204,36,18] and its accelerating shortening trend, the patent is determined to be in the "growth period". This determination [growth period] and the corresponding start and end time [2025-01-10, present] are used to generate the licensing stage interval.

[0088] S302: Invoke the licensing stage interval, verify the stage judgment result by combining the legal status label of the license, and adjust the combination structure of the invoking influence ratio and the stability influence ratio in the parameter weight according to the licensing stage of the patent to obtain the stage weight adjustment interval;

[0089] The patent's licensing stage [Growth Stage, 2025-01-10, Present] was invoked, and its current legal status tag was obtained. A database query revealed its status to be "Authorized - Valid - No Litigation." This stable and positive legal status verifies the validity of its commercial activities (invocation records), confirming the "Growth Stage" assessment result is correct. Based on the patent's licensing stage being "Growth Stage," the weighting of invocation impact and stability impact in its value assessment parameters was adjusted. The preset weighting structure for each stage is based on the patent pool exceeding... The system was established through regression analysis of the historical value curve of patents and licensing revenue data. Specifically, the combined structure for the "Introduction Phase" is {Call Influence: 0.4, Stability Influence: 0.6}; for the "Growth Phase," it is {Call Influence: 0.7, Stability Influence: 0.3}; for the "Maturity Phase," it is {Call Influence: 0.5, Stability Influence: 0.5}; and for the "Decline Phase," it is {Call Influence: 0.2, Stability Influence: 0.8}. Since the current stage is the "Growth Phase," the system selects the corresponding {Call Influence: 0.7, Stability Influence: 0.3} as the basic weight. Simultaneously, the system will fine-tune based on the sub-labels of legal status; for example, the "No Litigation" label will increase the stability influence weight. The adjusted combined structure is {call impact: 0.7, stability impact: 0.3*(1+0.05)=0.315}, which is then normalized to obtain {call impact: 0.7 / 1.015, stability impact: 0.315 / 1.015}, or {call impact: 0.69, stability impact: 0.31}, thus obtaining the stage weight adjustment range.

[0090] S303: Based on the stage weight adjustment range, perform proportional allocation of the corresponding parameter weights for each stage, encode the weight combination relationship of multiple stages, construct a weight allocation structure, and establish dynamic proportion control data.

[0091] Based on the stage weight adjustment interval {call impact: 0.69, stability impact: 0.31}, the weights of the corresponding parameters for each stage are proportionally allocated. This weight ratio is then applied to the patent's valuation model. This means that when calculating its current market value, the total weight of parameters related to call records (such as query frequency and the number of times it has been included in a license package) is 0.69, while the total weight of parameters related to legal stability (such as patent family size, remaining protection years, and litigation records) is 0.31. Next, the relationships between multiple stage weight combinations are encoded and stored in a structured form. For example, the weight adjustment for the "growth stage" is encoded as follows: ,in Represents the growth stage. Represents a timestamp. Represents the patent number. The system represents a weight structure, constructs a weight allocation structure, and aggregates the weight allocation coding sequences of all patents at different time points to establish dynamic proportion control data.

[0092] Please see Figure 5 The specific steps for obtaining the behavior contribution weight configuration are as follows:

[0093] S401: Based on dynamic proportion adjustment data, obtain the number of times members submit patent content, the frequency of participating in portfolio construction tasks, and the operation records of submitting collaboration requests during the licensing period, identify the proportion relationship of each type of behavior operation in the member operation set, and generate member behavior proportion coefficient;

[0094] Based on dynamic proportion adjustment data, within the current revenue distribution cycle (e.g., Q3 2025), the operation records of patent pool member "Innovative Technology Company" are obtained, and the number of times they submit patent content with high value weight (i.e., in the growth or maturity stage) is counted. The frequency of participating in system-initiated tasks aimed at building high-value patent portfolios around emerging technology themes is [number]. Each instance of submitting technical clarification or licensing requests to other members or patent pool operators through the collaboration platform is recorded as... The total number of operations was [number] times. Next, identify the proportion of each type of behavior in the set of member operations, and calculate the proportion of "content contribution". The proportion of "combined construction contribution" is The proportion of "collaborative contributions" is Generate member behavior ratio coefficients {Content: 0.16, Combination: 0.64, Collaboration: 0.20}.

[0095] S402: Based on the member behavior proportion coefficient, compare the proportions of various behaviors, determine the dominant operation characteristics among multiple behavior categories, filter the dominant operation category, and obtain the dominant operation category judgment value;

[0096] Based on the member behavior proportion coefficients {Content: 0.16, Combination: 0.64, Collaboration: 0.20}, the proportion values ​​of these three behaviors are directly compared, i.e., a comparison is made. , and The size of the ratio is used to determine the dominant operational characteristic among multiple behavioral categories. The criterion for determining the dominant operation is: if the ratio of a certain behavior category exceeds the sum of the ratios of all behaviors... (Right now If so, the action is considered the dominant operation. In this example, Therefore, "combined construction contribution" is identified as the dominant operational feature of "innovative technology company" in this period. If no single type of behavior accounts for more than 50%, the two types of behavior with the highest proportion are jointly identified as the main operational features. The dominant operational category is selected to obtain the dominant operational category judgment value, which is "combined construction".

[0097] S403: Based on the dominant operation category judgment value, perform contribution bias mapping based on the weight dimension corresponding to the dominant behavior, perform weight adjustment and update the proportion structure of members in licensing revenue, and establish behavior contribution weight configuration.

[0098] Based on the dominant operation category judgment value "Combined Construction," contribution weighting is mapped according to the weight dimension corresponding to the dominant behavior. The benchmark weight structure for patent pool revenue distribution is set as {Content Contribution Weight: 0.5, Combined Contribution Weight: 0.3, Collaboration Contribution Weight: 0.2}. This benchmark weight is determined according to the benchmark importance of each contribution type as stipulated in the patent pool charter. Since the dominant behavior of "Innovative Technology Company" is "Combined Construction," the system performs an upward adjustment on its "Combined Contribution Weight" dimension. The adjustment coefficient is positively correlated with the proportion coefficient of the dominant behavior, and is set to [value missing]. Multiply the original combined contribution weights by this coefficient to obtain the adjusted weights. At the same time, the weights of other non-dominant behaviors are reduced proportionally, while maintaining a total weight of 1. The reduced content contribution weight is... Collaborative contribution weight is After the update, the member's revenue distribution structure is {Content: 0.47, Combination: 0.342, Collaboration: 0.188}. This personalized weight structure is archived to establish a behavior contribution weight configuration.

[0099] Please see Figure 6 The specific steps for obtaining path structure conflict distribution data are as follows:

[0100] S501: Based on the behavior contribution weight configuration, obtain the regional number information, purpose restriction field and time period description in each subject's license request, analyze the overlap range of regional number intersection, purpose keywords and time period segments between each subject's license requests, and obtain the overlap structure of license parameters;

[0101] Based on the behavioral contribution weight configuration, when processing a new licensing transaction, license requests from two different entities (Company A and Company B) are obtained. Company A's request content is {Region ID: [DE,FR], Purpose limitation field: "In-vehicle infotainment system", Time period description: [2026-01-01, 2030-12-31]}, and Company B's request content is {Region ID: [DE,UK], Purpose limitation field: "Vehicle assisted driving", Time period description: [2028-06-01, 2032-05-31]}. The overlap of region IDs, purpose keywords, and time period segments between the license requests of each entity is analyzed. The region ID intersection is [DE]. The purpose keywords are calculated using semantic similarity. The semantic distance between "In-vehicle infotainment system" and "Vehicle assisted driving" in the preset automotive technology knowledge graph is [missing information]. (Values ​​range from 0 to 1; larger values ​​indicate greater distance and lower similarity). The overlapping range of time period segments is [2028-06-01, 2030-12-31], with an overlap duration of [missing information]. In 2010, this overlapping information was quantified, and an overlap metric ranging from 0 to 1 was defined, with the regional overlap degree being... (Company A requests 2 regions, 1 of which overlaps) = The degree of overlap in usage is The time overlap is (Total duration requested by Company A) = The number of overlapping structures with permissible parameters is obtained through comprehensive calculation.

[0102] S502: Based on the overlapping structure quantity of the permission parameters, compare the regional overlap and time period repetition of the permission path segments, judge the overlapping state of the overlapping areas in the path flow, mark the distribution position of each overlapping area in the path structure, and obtain the conflict segment distribution index.

[0103] Based on the overlapping structure of the permit parameters {region: 0.5, purpose: 0.4, time: 0.5}, the regional overlap and time period repetition of permitted path segments are compared, and a conflict judgment matrix is ​​set. The regional overlap is then used to determine the conflict. And time overlap And the degree of overlap in use When a potential overlapping state is identified, all current calculated values... All greater than Therefore, the overlapping status of overlapping areas in the path flow is judged and determined to be "moderate risk conflict". The distribution position of each overlapping area in the path structure is marked. Specifically, it is marked on the patent licensing path involving the region "DE", the use "automotive related" and the time period "2028-06-01 to 2030-12-31". A conflict report is generated, which includes the specific parameters of the conflict, the patent technology module involved (such as the aforementioned "wafer etching process" module, if it is a key technology for automotive chip manufacturing), and the conflict level "moderate", and the distribution index of conflict segments is obtained.

[0104] S503: Based on the conflict segment distribution index, segment the marked conflict segments in the path, adjust the connection method and boundary classification relationship between path segments, and establish path structure conflict distribution data;

[0105] Based on the conflict segment distribution index, which indicates a conflict between node A and node B in the licensing graph on the edge attribute {DE, Automobile, 2028-2030}, the marked conflict segments in the path are segmented. Specifically, the license package containing the "wafer etching process" module originally planned to be granted to companies A and B is split within the region "DE" and the time period "2028-06-01 to 2030-12-31". The connection methods and boundary classification relationships between path segments are adjusted. One adjustment method is to reduce the license of the conflict segment... The licensing method can be changed from "ordinary license" to "non-exclusive license with restrictions," and companies A and B can be clearly informed that there are other licensees in this specific area and time period. Alternatively, the boundary classification can be refined, for example, the use of "automotive related" can be further divided into "cockpit electronics" and "autonomous driving control," and these two sub-uses can be exclusively licensed to companies A and B respectively, thereby eliminating direct conflicts. This adjustment plan, including the revised licensing roadmap, updated boundary classification, and conflict resolution records, should be integrated and archived to establish road structure conflict distribution data.

[0106] Please see Figure 7 An AI-based patent pool operation and management system is used to execute the aforementioned AI-based patent pool operation and management method. The system includes:

[0107] The semantic segmentation module extracts the patent text content, analyzes and compares the operation verbs and objects of adjacent continuous parameters, compares the semantic structure and target object category, identifies the location of the conversion behavior, compares the verb sequence before and after the conversion point with the input and output terms, identifies the function jump position and segments it, and generates a semantic fragment sequence.

[0108] The vector projection module obtains the keyword sequence and encodes it to construct a direction vector based on the semantic segment sequence. It then projects the vector onto the reference axis, compares the projection angle offset between semantic segments, determines the cross relationship based on the keyword overlap ratio, establishes the semantic connection relationship of multiple segments, and generates semantic difference determination information.

[0109] The stage judgment module determines the information based on semantic differences, analyzes the interval between the time tag of the license node and the call record, determines the license stage and verifies the status tag, adjusts the parameter combination structure of call impact and stability impact, and generates dynamic proportion control data.

[0110] The behavior recognition module obtains member operation records based on dynamic proportion adjustment data, identifies the proportion relationship of behavior operations in the operation set, determines the dominant behavior characteristics and performs weight adjustment, updates the proportion structure of members in license revenue, and generates behavior contribution weight configuration.

[0111] The path control module configures the behavior contribution weight, obtains the regional number, usage restriction and time period in the subject's permission request, analyzes the intersection structure, determines the overlap status and marks the conflict segment, segments the path structure, adjusts the connection method and boundary classification structure between path segments, and generates path structure conflict distribution data.

[0112] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0113] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0114] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0115] It should be understood that, in various embodiments of the present invention, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0116] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0117] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0118] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0119] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0120] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0121] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0122] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A patent pool operation and management method based on artificial intelligence, characterized in that, The method includes: S1: Extract the patent text content, analyze and compare the operation verbs and objects of adjacent continuous parameters, compare the semantic structure and target object category, identify the location of the conversion behavior, compare the verb sequence before and after the conversion point with the input and output terms, identify the function jump position and cut it, and generate a semantic fragment sequence. S2: Based on the semantic segment sequence, obtain the keyword sequence and encode it to construct a direction vector. Project the vector onto the reference axis, compare the projection angle offset between semantic segments, determine the cross relationship based on the keyword overlap ratio, establish the semantic connection relationship of multiple segments, and generate semantic difference determination information. S3: Based on the semantic difference determination information, analyze the interval between the time tag of the permission node and the call record, determine the permission stage and verify the status tag, adjust the parameter combination structure of the call impact and stability impact, and generate dynamic proportion control data; S4: Based on the dynamic proportion control data, obtain member operation records, identify the proportion relationship of behavior operations in the operation set, determine the dominant behavior characteristics and perform weight adjustment, update the proportion structure of members in license revenue, and generate behavior contribution weight configuration. S5: Based on the behavior contribution weight configuration, obtain the regional number, usage restriction and time period in the subject's permission request, analyze the intersection structure, determine the overlap state and mark the conflict segment, perform segmentation processing on the path structure, adjust the connection method and boundary classification structure between path segments, and generate path structure conflict distribution data. The specific steps for obtaining the path structure conflict distribution data are as follows: S501: Based on the behavior contribution weight configuration, obtain the regional number information, purpose limitation field and time period description in each subject's license request, analyze the overlap range of regional number intersection, purpose keywords and time period segments between each subject's license requests, and obtain the license parameter overlap structure quantity; S502: Based on the overlapping structure quantity of the permission parameters, compare the regional overlap and time period repetition of the permission path segments, judge the overlapping state of the overlapping areas in the path flow, mark the distribution position of each overlapping area in the path structure, and obtain the conflict segment distribution index. S503: Based on the conflict segment distribution index, segment the marked conflict segments in the path, adjust the connection method and boundary classification relationship between path segments, and establish path structure conflict distribution data.

2. The patent pool operation and management method based on artificial intelligence according to claim 1, characterized in that, The semantic segment sequence includes semantic boundary positions, verb sequence indexes, and input / output term comparisons. The semantic difference determination information includes projection angle offset parameters, keyword overlap ratios, and segment connection relationships. The dynamic proportion control data specifically includes call influence proportion structure, stability influence proportion structure, and stage weight allocation parameters. The behavior contribution weight configuration includes submission frequency proportion, combined task frequency proportion, and collaborative request operation proportion. The path structure conflict distribution data specifically refers to regional intersection distribution, paragraphs with overlapping uses, and overlapping segments with time periods.

3. The patent pool operation and management method based on artificial intelligence according to claim 1, characterized in that, The specific steps for obtaining the semantic fragment sequence are as follows: S101: Obtain the patent text content, detect the operation verbs and objects corresponding to continuous parameters, analyze the semantic structure and target object categories between adjacent parameters, compare the changes in operation verbs and target objects between parameters, record the position of each category conversion, and generate a conversion behavior position index; S102: Based on the conversion behavior position index, compare the verb sequences and input / output terms before and after the conversion point, determine the synchronous relationship of changes in operation flow, target object and input / output terms, filter function jump positions, and use the jump points as segmentation markers to obtain a semantic boundary position array; S103: Based on the semantic boundary position array, adjust the text vector sequence structure, perform semantic segmentation of the text content, and divide the structure by combining the boundary position of each segment to obtain a semantic fragment sequence.

4. The patent pool operation and management method based on artificial intelligence according to claim 3, characterized in that, The process of redirecting to the selected location is as follows: Based on the conversion behavior location index, a symmetrical acquisition window with a preset parameter step size is set. The changes in verb sequences, input and output terms and operation flow before and after the conversion point within the window are compared. By judging whether the synchronous change ratio of the three exceeds the corresponding threshold, the target conversion point is marked as the function jump position. The corresponding thresholds include semantic synchronization threshold, flow direction change threshold, and object switching threshold. The thresholds are calculated by selecting a set of positions in the patent text content that do not have category conversion as the baseline set. The distribution of verb sequence difference ratio, input / output term replacement ratio, and operation flow direction change ratio in the target set are statistically analyzed. The semantic synchronization threshold, flow direction change threshold, and object switching threshold are set at the upper quantile boundary of each ratio.

5. The patent pool operation and management method based on artificial intelligence according to claim 3, characterized in that, The steps for obtaining the semantic difference determination information are as follows: S201: Based on the semantic segment sequence, obtain the keyword set corresponding to each semantic segment, count the first occurrence position of each group of keywords in the original text and arrange them in order, assign the sequential positions according to the sequence number to construct the direction vector, and obtain the sequential encoding vector group; S202: Based on the sequential encoding vector group, call the direction reference axis vector sequence, compare the projection angle offset between multiple semantic segment vectors, combine the overlap ratio between keyword lists, calculate the cross-determination coupling value, and obtain the cross-determination coefficient group. S203: Based on the sorting results of the cross-determination coefficient group, identify the connection paths between semantic segments and mark the cross-type, establish the semantic connection relationship of multiple segments, and generate semantic difference determination information.

6. The patent pool operation and management method based on artificial intelligence according to claim 5, characterized in that, The specific steps for obtaining the dynamic proportion control data are as follows: S301: Based on the semantic difference determination information, analyze the time tag and call record in the license node, compare the interval span between the time tag and the call record, determine the current license stage of the patent, and generate a license stage interval; S302: Call the aforementioned licensing stage interval, verify the stage judgment result by combining the legal status label of the license, and adjust the combination structure of the call influence ratio and the stability influence ratio in the parameter weight according to the licensing stage of the patent to obtain the stage weight adjustment interval; S303: Based on the stage weight adjustment range, perform proportional allocation processing of the corresponding parameter weights for each stage, encode the weight combination relationship of multiple stages, construct a weight allocation structure, and establish dynamic proportion control data.

7. The patent pool operation and management method based on artificial intelligence according to claim 6, characterized in that, The specific steps for obtaining the behavior contribution weight configuration are as follows: S401: Based on the dynamic proportion control data, obtain the number of times members submit patent content, the frequency of participating in combination construction tasks, and the operation records of submitting collaboration requests within the licensing period, identify the proportion relationship of each type of behavior operation in the member operation set, and generate member behavior proportion coefficient. S402: Based on the member behavior proportion coefficient, compare the proportions of various behaviors, determine the dominant operation characteristics among multiple behavior categories, filter the dominant operation category, and obtain the dominant operation category determination value; S403: Based on the dominant operation category determination value, perform contribution bias mapping based on the weight dimension corresponding to the dominant behavior, perform weight adjustment and update the proportion structure of members in licensing revenue, and establish behavior contribution weight configuration.

8. A patent pool operation and management system based on artificial intelligence, characterized in that, The system is used to implement the artificial intelligence-based patent pool operation and management method according to any one of claims 1-7, and the system includes: The semantic segmentation module extracts the patent text content, analyzes and compares the operation verbs and objects of adjacent continuous parameters, compares the semantic structure and target object category, identifies the location of the conversion behavior, compares the verb sequence before and after the conversion point with the input and output terms, identifies the function jump position and segments it, and generates a semantic fragment sequence. The vector projection module obtains the keyword sequence and encodes it to construct a direction vector based on the semantic segment sequence. It then projects the vector onto the reference axis, compares the projection angle offset between semantic segments, determines the cross relationship based on the keyword overlap ratio, establishes the semantic connection relationship of multiple segments, and generates semantic difference determination information. The stage judgment module analyzes the interval between the time tag of the permission node and the call record based on the semantic difference judgment information, judges the permission stage and verifies the status tag, adjusts the parameter combination structure of call impact and stability impact, and generates dynamic proportion control data. The behavior recognition module obtains member operation records based on the dynamic proportion control data, identifies the proportion relationship of behavior operations in the operation set, determines the dominant behavior characteristics and performs weight adjustment, updates the proportion structure of members in license revenue, and generates behavior contribution weight configuration. The path control module obtains the regional number, usage restriction and time period in the subject's permission request based on the behavior contribution weight configuration, analyzes the intersection structure, determines the overlap state and marks the conflict segment, segments the path structure, adjusts the connection method and boundary classification structure between path segments, and generates path structure conflict distribution data.

Images