Personalized recommendation method and system based on large language model and domain model cooperation

Abstract

The application discloses a personalized recommendation method and system based on large language model and field model cooperation, and relates to the technical field of information recommendation. In the target recommendation scene, the method processes unstructured data through a large language model to obtain semantic mode features, and simultaneously analyzes user quantifiable operation records by using a specified field model to output behavior mode features, thereby realizing the synchronous extraction of unstructured semantic information and structured behavior information. Two types of features are subjected to bidirectional information supplement and knowledge transfer to construct a cooperatively optimized feature mapping set, thereby realizing the unified conversion and synchronous scheduling of cross-modal features. In combination with real-time interaction information reflecting the current intention of a user and the change of scene demand, an initial recommendation list is generated through double-model cooperative reasoning, and after dynamic sorting, it is determined whether to output a personalized recommendation result, thereby realizing the accurate adaptation of real-time demand of a cross-scene user and personalized recommendation, improving the timeliness of recommendation, and further effectively improving the rapid adaptability of personalized recommendation.

Description

Technical Field

[0001] This invention relates to the field of information recommendation technology, and in particular to a personalized recommendation method and system based on the collaboration of a large language model and a domain model. Background Technology

[0002] As the volume of digital content and services continues to expand, users' need to efficiently obtain target content from massive amounts of information is becoming increasingly urgent, making personalized recommendations a core link connecting users and information. Meanwhile, Large Language Model (LLM) technology is developing rapidly, providing a new paradigm for deeply mining users' latent intentions and optimizing the recommendation interaction experience thanks to its powerful semantic understanding, natural language interaction, and cross-domain knowledge integration capabilities. Domain models, through long-term practice, have accumulated professional data processing experience in specific industries, scenario-based rules, and the stability of recommendation results. The synergistic integration of these two technologies has become a key driving force for upgrading personalized recommendation technology towards greater accuracy and more scenario-specific tailoring.

[0003] Current personalized recommendation technologies are primarily implemented through three paths. The first relies on collaborative filtering and matrix factorization techniques, mining the association between users and specified targets based on historical user behavior data to achieve basic interest matching, suitable for early e-commerce and content platform scenarios. The second employs deep learning models, such as Neural Collaborative Filtering (NCF), integrating user profiles, specified target attributes, and scene features. It enhances feature interaction through neural networks to improve recommendation accuracy. The third approach begins to incorporate large language model capabilities. Some solutions utilize LLM to generate user interest text and specified target knowledge tags as supplementary features, or leverage LLM to optimize the natural language interpretation of recommendation results, exploring a combined application model.

[0004] For example, the Chinese invention patent CN118551123B discloses a method and system for personalized movie recommendation that integrates a large language model and a knowledge graph. The method includes: collecting a dataset containing detailed movie information and preprocessing it; constructing a knowledge graph based on the preprocessed dataset; fine-tuning the large language model by constructing prompt words based on the knowledge graph, and simultaneously constructing a fusion rating model by combining it with traditional recommendation algorithms; training the fusion rating model; and using the trained fusion rating model to predict ratings and make personalized recommendations.

[0005] For example, the Chinese invention patent application CN120492590A discloses a personalized content recommendation method, system, and electronic device based on a large language model, which includes: in response to receiving a content query instruction, generating question text corresponding to the content query instruction using at least one large language model; determining the corresponding text block based on the similarity between each question text and text blocks in the content library; generating a recommendation result corresponding to the question text using at least one large language model based on a first prompt word generated according to each question text and the corresponding text block; and outputting the recommendation result.

[0006] In cross-scenario recommendation, it is necessary to achieve seamless connection of user needs and consistent preference continuity in different scenarios. However, the features output by large language models are usually unstructured vectors. Existing technologies usually do not fully consider a unified feature transformation and synchronous scheduling mechanism, which leads to temporal misalignment between unstructured semantic features and structured behavioral features of the domain model in the transmission and parsing process. This further exacerbates the asynchronous problem that the large language model has already output recommendation candidates before the domain model has completed behavioral calculation. This reduces the timeliness of cross-modal feature collaboration and results in poor adaptability of personalized recommendations in cross-scenario recommendation. Summary of the Invention

[0007] To address the problem of low computational efficiency in existing technologies when dynamically expanding large-scale, multi-source data, this invention provides a personalized recommendation method and system based on the collaboration of a large language model and a domain model. The technical solution is as follows:

[0008] On the one hand, a personalized recommendation method based on the collaboration of a large language model and a domain model is provided. This method includes: Step 1, in the target recommendation scenario, using a large language model to perform semantic understanding on the acquired unstructured data to output semantic pattern features, and simultaneously using a specified domain model to perform collaborative analysis on the acquired user behavior data to output behavioral pattern features. The user behavior data represents the quantifiable operation records of users in the target recommendation scenario; Step 2, bidirectional information supplementation is performed between the semantic pattern features output by the large language model and the behavioral pattern features output by the specified domain model to form a collaboratively optimized feature mapping set; Step 3, combining real-time user interaction information and the formed feature mapping set, an initial recommendation list is generated through collaborative reasoning of the large language model and the specified domain model. After dynamic sorting, it is determined whether to output personalized recommendation results. Real-time user interaction information is used to reflect the user's current operation status and changes in scenario-based needs, and personalized recommendation results are used to support rapid adaptation in cross-modal scenarios.

[0009] On the other hand, a personalized recommendation system based on the collaboration of a large language model and a domain model is provided. This system includes: a dual-model pattern feature output module, which is used to perform semantic understanding on the acquired unstructured data in the target recommendation scenario using a large language model to output semantic pattern features, and at the same time to perform collaborative analysis on the acquired user behavior data using a specified domain model to output behavior pattern features.

[0010] The bidirectional information supplementation and mapping set construction module is used to supplement the semantic pattern features output by the large language model and the behavioral pattern features output by the specified domain model in two directions to form a collaboratively optimized feature mapping set. The initial recommendation list generation and judgment module is used to combine the user's real-time interaction information and the formed feature mapping set to generate an initial recommendation list through collaborative reasoning between the large language model and the specified domain model. After dynamic sorting, it is used to determine whether to output personalized recommendation results.

[0011] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0012] 1. This invention processes data in parallel using two models within a target recommendation scenario: a large language model performs semantic understanding processing on unstructured data, outputting semantic pattern features; simultaneously, a domain-specific model analyzes quantifiable user operation records, outputting behavioral pattern features. This step achieves simultaneous extraction of two types of features, avoiding the one-sidedness of a single model focusing only on semantics or behavior, and eliminating information bias at the data source for subsequent cross-modal collaboration. Bidirectional information supplementation and knowledge transfer are performed on semantic and behavioral pattern features to construct a collaboratively optimized feature mapping set. This process, through a unified feature transformation and scheduling mechanism, solves the temporal misalignment problem of feature transmission and parsing in existing technologies, fundamentally alleviating the pain point of low timeliness in cross-modal feature collaboration. Combining real-time interactive information reflecting the user's current operation state and changes in scenario needs, an initial recommendation list is generated based on dual-model collaborative reasoning, and after dynamic sorting, it is determined whether to output personalized recommendation results. This step allows for seamless connection of user needs across scenarios and consistent preference continuity, effectively improving the problem of low cross-scenario recommendation adaptability in existing technologies.

[0013] 2. First, the semantic content of the unstructured data is determined. If it exceeds a preset limit, the sub-features are refined to a preset number according to the semantic granularity ratio. If it does not exceed the preset limit, the semantic vectors are sorted and coupled to obtain the results. Then, semantic pattern features are generated and cross-text verification is performed to ensure that the semantics stably match the real needs. Next, behavioral pattern features are output: the frequency of specified behaviors is obtained from the temporal dimension to calculate temporal difference parameters, and behavioral trend changes are judged to adjust temporal weights or directly output according to patterns. For new users, the behavior of the same-label group is referenced. This process solves the problem of unstructured semantic feature parsing through dynamic adjustment of semantic granularity and cross-text verification. It optimizes the output of structured behavioral features by using temporal difference parameters and weight control, and constructs a unified feature transformation and synchronous scheduling mechanism to avoid temporal misalignment of semantic and behavioral feature transmission and parsing, reduce asynchronous model output problems, improve the timeliness of cross-modal feature collaboration, achieve seamless connection of user needs and consistency of preferences in cross-scenario recommendations, and improve the adaptability of personalized recommendations.

[0014] 3. When forming a collaboratively optimized feature mapping set, the semantic and behavioral pattern features are first standardized in terms of dimension. The feature mapping matrix transforms features of different dimensions to the same target space, reducing dimensional differences. Then, the semantic-behavioral feature correlation is calculated. First, a frequency domain intensity value is generated by Fourier transform to capture temporal correlation. Then, a discrete matching degree index is formed by binary encoding and cosine similarity. Finally, the two are balanced by geometric mean to avoid interference from a single index. Next, the feature weights are adjusted according to the deviation value between the correlation degree and the midpoint of the preset range. If the deviation exceeds the upper limit, the semantic pattern feature weights are reduced and the behavioral pattern feature weights are increased. If the deviation value is negative and the opposite number exceeds the midpoint, the semantic pattern feature weights are increased and the behavioral pattern feature weights are reduced. If the deviation value is within the allowable range, the weights are maintained. Finally, the feature mapping set is obtained by alignment and matching. This process addresses the difference between unstructured and structured feature dimensions through unified dimensional transformation, ensures accurate correlation calculation through geometric mean, achieves feature synergy through dynamic weight adjustment, constructs a unified feature transformation mechanism, avoids timing misalignment in transmission and parsing and asynchronous model output, improves the timeliness of cross-modal feature synergy, facilitates seamless connection of user needs and continuity of preferences in cross-scenario recommendations, and enhances the adaptability of personalized recommendations.

[0015] 4. When generating the initial recommendation list, the system first retrieves operation data from the user's real-time interaction logs and converts it into real-time feature vectors. These vectors are then matched with the feature mapping set to calculate the relevance score. Simultaneously, the system compares the candidate recommendation table generated based on the large language model with the candidate list generated based on the specified domain model to obtain the overlap rate of the specified targets. If the relevance score exceeds the preset relevance score and the overlap rate of the specified targets is lower than the preset overlap rate, it indicates that the user's current needs have a high degree of matching with historical data, but the lists differ significantly. Differential cross-validation is needed to complete the missing targets and perform adaptability verification. If the relevance score does not exceed the preset threshold and the overlap rate is not lower than the preset threshold, an initial recommendation list containing IDs, scores, and recommendation reasons is directly generated based on the overlapping targets. When determining whether to output personalized recommendation results, the system considers the recommendation priority of the targets in the initial list and the proportion of specified targets. If the criteria are met, the results are output and sorted by priority; otherwise, optimization or manual verification is prompted. This process relies on feature mapping sets to achieve real-time matching with historical features. It solves the problem of collaboration between unstructured semantics and structured behavioral features through differentiated verification and adaptability verification, avoids temporal misalignment and model asynchrony, improves the timeliness of cross-modal collaboration, helps seamlessly connect user needs and continue preferences across scenarios, and improves the adaptability of personalized recommendations. Attached Figure Description

[0016] 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.

[0017] Figure 1 A flowchart illustrating a personalized recommendation method based on the collaboration of a large language model and a domain model, provided in an embodiment of the present invention;

[0018] Figure 2 A flowchart illustrating the output of behavioral pattern features provided in embodiments of the present invention;

[0019] Figure 3 A flowchart illustrating the generation process of the initial recommendation list provided in this embodiment of the invention;

[0020] Figure 4 This is a schematic diagram of the structure of a personalized recommendation system based on the collaboration of a large language model and a domain model, provided in an embodiment of the present invention. Detailed Implementation

[0021] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0022] 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.

[0023] 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.

[0024] This invention provides a personalized recommendation method based on the collaboration of a large language model and a domain model, such as... Figure 1 The flowchart shown is for a personalized recommendation method based on the collaboration of a large language model and a domain model. The processing flow of this method may include the following steps:

[0025] Step one involves using a large language model to perform semantic understanding on the acquired unstructured data in the target recommendation scenario, outputting semantic pattern features. Simultaneously, a domain-specific model is used to collaboratively analyze the acquired user behavior data, outputting behavioral pattern features. This achieves the simultaneous extraction of unstructured semantic information and structured behavioral information. Unstructured data represents information that cannot be stored or parsed in a fixed format, typically including user comments, consultation text, browsing content descriptions, and scene dialogue records. User behavior data represents quantifiable operation records of users in the target recommendation scenario, typically including click frequency, dwell time, favorite / unfavorite actions, and purchase / browsing history.

[0026] Step two involves bidirectional information supplementation between the semantic pattern features output by the large language model and the behavioral pattern features output by the specified domain model to form a collaboratively optimized feature mapping set. This achieves unified conversion and synchronous scheduling of cross-modal features. Semantic pattern features are used to capture users' potential semantic needs, such as understanding user needs and scenarios through text. Behavioral pattern features are used to quantify users' explicit operational patterns, such as reflecting users' actual needs through behavioral frequency and temporal changes. The feature mapping set is used to integrate general semantic knowledge and domain behavioral patterns, providing a unified feature benchmark for subsequent real-time feature matching and recommendation list generation.

[0027] Step three involves combining real-time user interaction information and the resulting feature mapping set to generate an initial recommendation list through collaborative reasoning between a large language model and a specified domain model. After dynamic sorting, a decision is made on whether to output personalized recommendation results. This achieves seamless integration of user needs across scenarios and accurate adaptation of personalized recommendations, improving the timeliness and adaptability of recommendations. Real-time user interaction information reflects the user's current operational status and changes in scenario-based needs. Personalized recommendation results support rapid adaptation across modal scenarios. The initial recommendation list carries the core information of candidate recommendation targets, typically including the specified target ID, matching score, and recommendation reason.

[0028] Specifically, the large language model is built using the Transformer-BERT architecture: First, it is pre-trained using massive amounts of general text data (web pages, books, dialogue corpora) to capture semantic, grammatical, and common-sense associations through self-supervised learning (such as masked language modeling tasks). Then, it is fine-tuned using unstructured data from recommendation scenarios (user comments, consultation texts) to adapt to the needs of contextualized semantic understanding, ultimately resulting in a model that can output structured semantic pattern features. Its input is unstructured data from the target recommendation scenario, and its output is semantic pattern features such as user demand vectors and contextual semantic labels.

[0029] The specific domain model is constructed using a temporal collaborative filtering model: It uses historical user behavior data (clicks, purchase records) as training samples, learns user behavior patterns and temporal changes through the model, and incorporates recommendation scenario business rules (such as product category weights and user segmentation rules) to optimize model parameters, forming a targeted domain model. Its input is user behavior data (click frequency, dwell time), and its output is user behavior trend vectors, preference weight values, and other behavioral pattern features.

[0030] The two types of models have a clear division of labor. By inputting different data sources, they output semantic and behavioral pattern features respectively, which provides a foundation for subsequent feature fusion and recommendation list generation, solves the problem of asynchronous cross-modal features, and improves recommendation adaptability.

[0031] In a specific implementation, taking the online course recommendation scenario as an example: unstructured data includes user questions in the course consultation area (e.g., looking for a Python data analysis course suitable for beginners, preferably with practical examples) and course evaluation text (e.g., this course is too fast-paced, I hope there will be more explanations of basic concepts); user behavior data includes the user's course browsing history in the past 15 days (clicked on Python courses 8 times and data analysis courses 6 times), video viewing time (watched a single episode of a basic Python course for more than 40 minutes), and favorites (favorited 2 courses with practical tags).

[0032] First, unstructured data is processed using a large language model based on Transformer-BERT, outputting semantic pattern features such as Python needs for beginners, emphasis on practical examples, and preference for slow-paced explanations. Then, a temporal collaborative filtering domain model is used to analyze behavioral data, outputting behavioral pattern features such as a Python course preference weight of 0.8 and behavioral trend vectors for practical courses. Subsequently, both types of features are dimensionality-standardized (mapped to 512 dimensions), and weights are adjusted based on relevance to form a feature mapping set. When a user searches for Python practical courses in real time, the real-time features are matched against the mapping set, and the two models collaboratively generate an initial list containing course IDs, matching scores (e.g., a matching score of 91 for a Python practical course for beginners), and recommendation reasons. Personalized recommendations are then output after sorting.

[0033] It solves the problem of the disconnect between users' semantic needs and behavioral data in online education scenarios. The recommendations not only match the learning preferences expressed by users in the text, but also conform to actual browsing and operation habits. The connection between needs across learning stages (such as from basic to advanced) is smoother, which effectively improves the accuracy of course recommendations and user learning conversion rate.

[0034] Furthermore, the specific output process of semantic pattern features is as follows: First, the semantic content volume in the unstructured data is obtained. If the semantic content volume is greater than a preset semantic content volume, the semantic granularity ratio is calculated based on the obtained semantic content volume deviation, using the semantic granularity ratio formula. This ratio is the ratio of the semantic content volume deviation to the preset semantic content volume. The semantic content is then progressively divided into dimensions corresponding to the semantic content volume, decomposing the semantic content into a preset number (usually 10-20) of semantic sub-features. If the semantic content volume is not greater than the preset semantic content volume, the semantic vectors corresponding to the semantic content in the unstructured data are sorted by similarity. Each sorted semantic vector is assigned a corresponding weight coefficient and coupled, i.e., each semantic vector is multiplied by its corresponding weight coefficient, and all results are summed to obtain the semantic feature result. The weight coefficient is usually set by the pre-defined personnel based on the business objectives of historical recommendation scenarios and historical semantic matching effects, used to quantify the importance of each semantic vector. The vector sequence of the semantic feature result is obtained, and semantic pattern features are generated sequentially from high to low semantic association strength. Cross-text semantic consistency verification is then performed to improve the matching degree between the semantic pattern features and user needs.

[0035] The semantic granularity ratio is a quantitative representation of the difference between the preset semantic content amount and the acquired semantic content amount, used to reflect the fineness of semantic parsing of unstructured data. The preset semantic content amount is set by analyzing the average effective semantic length of historical unstructured data and combining it with the accuracy requirements of semantic parsing in the recommendation scenario to ensure that it is adapted to the needs of the scenario. The semantic content amount deviation represents the difference between the acquired semantic content amount and the preset semantic content amount. The vector sequence represents the ordered vector set formed by the semantic feature results according to the preset semantic logic. The semantic association strength represents the degree of matching between each ordered vector and the user's current needs. The semantic association strength is obtained by calculating the cosine similarity between each ordered vector and the user's current needs vector.

[0036] like Figure 2 The flowchart for outputting behavioral pattern features, as shown, follows this logic: First, obtain the temporal difference parameters. Then, determine if their absolute values ​​are not greater than a preset difference threshold. If so, directly output the behavioral pattern feature. If not, increase the temporal weight corresponding to the deviation of the temporal difference parameters and check if the temporal weight exceeds a preset range. If it does, pause the feature calculation process; if it does not, continue monitoring the temporal difference parameters. Furthermore, for special cases such as new users, behavioral pattern features can also be output, but these are based on the behavioral distribution patterns of user groups with the same feature label, ensuring the comprehensiveness and adaptability of the process.

[0037] It is important to understand that the specific output process of behavioral pattern features is as follows: From the time-series dimension of user behavior data, the frequency of specified behaviors within two adjacent statistical periods is retrieved. The difference between the frequency of specified behaviors in the later period and the frequency of specified behaviors in the previous period is defined as the time-series difference parameter, which is used to quantify the changing trend of user behavior over time. If the absolute value of the time-series difference parameter is greater than the preset difference threshold, it is determined that the behavior trend of the corresponding user has changed significantly, and a time-series weight corresponding to the deviation of the time-series difference parameter is added to the specified domain model. At the same time, it is monitored whether the time-series weight exceeds the preset time-series weight range. The preset difference threshold is set by analyzing the frequency of user's historical behaviors and calculating the upper limit of the frequency difference of 90%. The preset time-series weight range is set by referring to the optimal weight range of the same scenario domain model. The optimal weight range is usually 0.3-0.8.

[0038] If the temporal weight exceeds the preset temporal weight range, the feature calculation process based on that weight is paused to prevent excessive interference with the behavioral feature output process. If the temporal weight is within the preset temporal weight range, monitoring continues until the absolute value of the temporal difference parameter is not greater than the preset difference threshold, at which point the behavioral pattern feature is output. If the absolute value of the temporal difference parameter is not greater than the preset difference threshold, the corresponding user's behavioral trend is determined to be stable, and behavioral pattern features are output based solely on the corresponding user's historical behavioral patterns. If it is a new user, behavioral pattern features are output based on the behavioral distribution patterns of user groups with the same feature label. The same feature label represents a set of labels consistent with the new user's current scenario and behavioral patterns.

[0039] The cross-text semantic consistency verification process involves: acquiring semantic fragments from semantic pattern features, generating semantic vectors corresponding to each semantic fragment using a large language model, and calculating the cosine similarity of each semantic vector; if the cosine similarity is higher than a preset cosine similarity, the semantic features are considered consistent, and the semantic pattern features are output to ensure output stability in cross-text scenarios. The preset cosine similarity is represented by the sum and average of historical cosine similarities of historical cross-text semantic vectors; otherwise, the process goes back to the semantic pattern feature output stage, re-decomposes the semantic content into a preset number of semantic sub-features, and after a preset number of decompositions, if the re-acquired cosine similarity is still not higher than the preset cosine similarity, the preset personnel are prompted to perform semantic calibration until the semantic features are considered consistent.

[0040] In this embodiment, when outputting semantic pattern features, the processing method is dynamically adjusted according to the amount of semantic content. When the amount of semantic content exceeds the preset amount, semantic sub-features are split proportionally. When the amount is insufficient, semantic vectors are weighted and coupled. Semantic accuracy is also ensured through cross-text consistency verification, effectively improving the parsing accuracy of unstructured data and avoiding semantic bias caused by a single processing method. When outputting behavioral pattern features, the temporal difference parameters are combined with the preset difference threshold to dynamically adjust the temporal weights. New users are referenced to the behavior of groups with the same tag to accurately capture changes in user behavior trends. Both types of feature outputs rely on historical data and scene adaptation to set parameters, reducing the temporal misalignment between unstructured semantic and structured behavioral features, alleviating the asynchronous problem of the two models, laying the foundation for subsequent cross-modal feature collaboration, and significantly improving the adaptability of cross-scene personalized recommendations.

[0041] Furthermore, a collaboratively optimized feature mapping set is formed, specifically including: performing dimensionality standardization on the output semantic pattern features and behavioral pattern features to transform the two types of features to a specified dimensional space through a feature mapping matrix, such as mapping the 1024-dimensional vector of semantic features and the 256-dimensional vector of behavioral features to a 512-dimensional target space, reducing the dimensionality difference caused by different feature sources; based on the semantic-behavioral feature correlation degree in the current recommendation scenario, calculating the deviation value between the current semantic-behavioral feature correlation degree and the midpoint of the preset feature correlation degree range, where the preset feature correlation degree range is obtained by considering the historical semantic-behavioral feature correlation degrees of the target recommendation scenario over the past 3-6 months. Minimum and maximum values ​​are used to determine the feature correlation range after removing extreme outliers (such as data below the 1st percentile or above the 99th percentile). The minimum value of the remaining data is used as the lower limit and the maximum value is used as the upper limit. The midpoint of the preset feature correlation range represents the median of the historical semantic-behavioral feature correlation within the preset feature correlation range for this scenario. If the deviation value is within the allowable range, the current semantic and behavioral pattern feature weights are kept unchanged. The two types of features are mapped and matched according to the dimension corresponding to the current weight through a feature alignment algorithm to eliminate dimensional deviation. The dimension vector after mapping and matching is statistically analyzed to obtain a feature mapping set that integrates general semantic knowledge and domain behavioral rules.

[0042] A positive deviation value indicates that if the deviation value is not less than the upper limit of the allowable deviation range, the coupling redundancy is determined to be too high. Based on the first deviation value, the semantic pattern feature weights are adjusted downwards by a certain percentage, while the behavioral pattern feature weights are adjusted upwards by the same percentage. The sum of the semantic pattern feature weights and the behavioral pattern feature weights is 1. The first deviation value represents the ratio of the difference between the obtained deviation value and the upper limit of the allowable deviation range to the upper limit of the allowable deviation range. If the deviation value is less than the upper limit of the allowable deviation range, the current semantic pattern feature weights and behavioral pattern feature weights remain unchanged. The upper limit of the allowable deviation range is determined by statistically analyzing the maximum deviation value without redundancy risk in historical feature coupling, adding a 10%-15% tolerance interval, and finally determining the upper limit to ensure that coupling redundancy is only determined when the deviation significantly exceeds the safe range. The lower limit of the allowable deviation range is usually set to 0.

[0043] The deviation value is negative: If the opposite of the deviation value is not less than the midpoint of the allowable deviation value range, the coupling depth is determined to be insufficient. The semantic pattern feature weights are adjusted upward based on the second deviation value, while the behavioral pattern feature weights are adjusted downward by the same proportion. The second deviation value represents the ratio of the difference between the midpoint of the allowable deviation value range and the obtained deviation value to the midpoint of the allowable deviation value range. If the opposite of the deviation value is less than the midpoint of the allowable deviation value range, the current semantic pattern feature weights and behavioral pattern feature weights remain unchanged. Both the first and second deviation values ​​are used to quantify the degree of feature coupling deviation from the expected state, and the numerical range is limited to [0,1] to ensure that the weight adjustment ratio is controllable. The midpoint of the allowable deviation value range is usually taken as 50% of the upper limit of the allowable deviation value range, which reflects the balanced expectation of feature coupling and can be compared with the deviation value (positive / negative) to accurately determine whether the coupling is redundant or insufficient in depth.

[0044] The semantic-behavioral feature correlation is obtained as follows: Fourier transform is performed on the semantic pattern features and behavioral pattern features after dimensionality standardization to convert the two types of features from the time domain to the frequency domain, presenting the periodic pattern of the two types of features changing over time. Then, the amplitude and phase of each frequency point in the spectrum of the two types of features are extracted. For each frequency point, the Pearson correlation coefficient of the amplitude of the semantic and behavioral pattern features (quantifying the consistency of the amplitude change trend, with a value of [-1,1], the closer to 1, the higher the amplitude overlap) and the mean of the phase difference between the two are calculated (measuring phase coordination, the smaller the mean, the more synchronized the phase). Then, the Pearson correlation coefficient (forwardly mapped to [0,1]) and the mean phase difference (backwardly mapped to [0,1], the smaller the difference, the larger the mapped value) are normalized and then weighted and summed according to a 4:6 ratio (the degree of amplitude overlap is more critical to the periodic correlation) to obtain a comprehensive quantized frequency domain intensity value, which is set to [0,1], in order to capture the correlation pattern of the above two types of features over time. The higher the value, the closer the correlation between the two types of features in the periodic trend.

[0045] Binary encoding is used to convert the median of continuous values ​​of two types of features into a threshold. Continuous values ​​greater than the threshold are mapped to 1, and continuous values ​​less than or equal to the threshold are mapped to 0, thus transforming both types of features into equal-length 0-1 binary sequences. The two binary sequences are then treated as vectors, and the cosine similarity between the sequences is calculated, which is the vector dot product divided by the product of the magnitudes of the two vectors. This forms a discrete matching index used to quantify the degree of matching between discretized features. For example, if the semantic feature sequence is [1,0,1,1] and the behavioral feature sequence is [1,0,0,1], the cosine similarity formula yields a value that reflects the directional consistency between the two features in the discrete dimension. The discrete matching index represents the degree of directional consistency between the two types of features in the vector space after discretization; a higher value indicates a stronger matching between the two types of features in the discrete feature dimension.

[0046] Since the influence of frequency domain intensity (reflecting periodic correlation) and discrete matching degree index (reflecting discrete dimensional matching) on ​​the correlation degree needs to be balanced, a geometric mean calculation is used for fusion: the values ​​of the two indices are multiplied, and then the square root of the product is taken to obtain the semantic-behavioral feature correlation degree. If the frequency domain intensity value is 0.8 and the discrete matching degree index is 0.9, the correlation degree is... =0.847, this result can avoid the excessive dominance of a single indicator (such as an excessively high frequency domain intensity value) on the overall correlation, and more objectively reflect the degree of synergy between the two types of features.

[0047] In this embodiment, dimensional standardization eliminates the dimensional differences between semantic and behavioral features, avoiding fusion barriers caused by different sources and laying the foundation for subsequent collaboration. The semantic-behavioral feature correlation calculation integrates the periodicity of the frequency domain and the matching of discrete dimensions, and uses geometric averaging to balance the influence of the index, ensuring that the correlation evaluation is objective and accurate. For cases where the deviation value is positive, the semantic weight is lowered and the behavioral weight is raised when the upper limit is exceeded, which can effectively reduce coupling redundancy and avoid the excessive dominance of semantic features. When the deviation value is negative and the absolute value exceeds the midpoint, the semantic weight is raised and the behavioral weight is lowered, which can compensate for insufficient coupling depth and strengthen the collaboration between the two types of features. At the same time, the weight adjustment is based on the quantized first and second deviation values ​​and is limited to the range [0,1], ensuring that the adjustment is controllable and fits the scenario expectations. The final feature mapping set integrates general semantic knowledge and domain behavioral rules, which not only alleviates the problem of cross-modal feature temporal misalignment, but also provides high-quality feature support for subsequent recommendations, significantly improving the adaptability and accuracy of cross-scenario recommendations.

[0048] like Figure 3 The flowchart shown illustrates the initial recommendation list generation process. Its design logic is as follows: First, obtain the relevance score and the specified target overlap rate. By comparing the relevance score with the corresponding preset value, determine whether to proceed directly to the next step. If the relevance score is not greater than the corresponding preset value, then based on the comparison of the specified target overlap rate with the preset value, decide whether to perform differentiated cross-validation and suitability verification. If the specified target overlap rate exceeds the corresponding preset value, prompt for manual review. Only after verification or meeting the conditions will the initial recommendation list be generated. This logic ensures that recommendations are based on reasonable relevance and overlap rates, balancing automation and manual intervention to guarantee recommendation quality.

[0049] Further understanding is needed regarding the specific steps for generating the initial recommendation list: Retrieving current operation data from real-time user interaction logs and converting it into a real-time feature vector; then performing feature matching calculations with the existing feature mapping set to obtain the correlation score between real-time features and historical mapping features; comparing the candidate recommendation list generated based on a large language model with the candidate list generated based on a specified domain model to obtain the specified target overlap rate; the correlation score, obtained through the Euclidean distance between real-time features and historical mapping features, quantifies the closeness of the match between them, with higher scores indicating a higher degree of fit between current needs and historical features; if the correlation score is greater than a preset correlation score, and the specified target overlap rate is less than a preset overlap rate, it indicates a high degree of match between the user's current needs and historical features, but significant differences between the two recommendation lists, requiring differentiated cross-validation to supplement the corresponding information in the two lists. While identifying non-overlapping target information, the system improves the matching degree between non-overlapping targets and users' real-time needs. Simultaneously, it performs adaptive verification of differentiated cross-validation results. The preset relevance score is determined by statistically analyzing the mean Euclidean distance of users' historical features during the recommendation process, mapping it back to the base score, and adding a 5%-8% tolerance range. If the relevance score is not greater than the preset relevance score, and the overlap rate of the specified targets is not less than the preset overlap rate set by preset personnel, it indicates that the user's current needs have a low matching degree with historical features, but the consistency between the two types of recommendation lists is high. An initial recommendation list containing the specified target ID, matching score, and recommendation reason is directly generated based on the overlapping target information. If the above conditions are not met, it indicates that the matching degree between the user's current needs and historical features, and the consistency between the two types of recommendation lists, are in an intermediate state. Preset personnel are then prompted to intervene and manually review the information to avoid fluctuations in recommendation accuracy due to the intermediate state.

[0050] The adaptation verification of the differentiated cross-validation results is specifically as follows: The specified targets corresponding to the non-overlapping specified target information in the candidate recommendation table are designated as semantic-priority targets, and the specified targets corresponding to the non-overlapping specified target information in the candidate list are designated as behavior-priority targets. Behavioral pattern features in the feature mapping set are called to obtain the length of the longest common subsequence between the feature pattern features and the corresponding user's historical behavior sequence. Combined with the total length of the historical behavior sequence, the length ratio is obtained, i.e., the ratio of the longest common subsequence length to the total length of the historical behavior sequence. Simultaneously, the user behavior weight coefficients in the feature mapping set are combined to generate the behavioral fit of the semantic-priority targets through weighted summation. Simultaneously, the semantic pattern features in the feature mapping set are called to calculate the cosine similarity between the attribute labels of the behavior-priority targets and the semantic vector of the user's current operation state, thus obtaining the semantic fit of the behavior-priority targets.

[0051] Only targets with behavioral fit no less than a preset behavioral fit and semantic fit no less than a preset semantic fit, corresponding to semantic priority targets and behavioral priority targets, are retained. Targets that do not meet the conditions (i.e., behavioral fit less than a preset behavioral fit, and semantic fit less than a preset semantic fit, corresponding to semantic priority targets and behavioral priority targets) are removed, and the reasons for removal are synchronized to the designated personnel. The preset behavioral fit is determined by analyzing the behavioral fit data of behavioral priority targets clicked / converted in the historical recommendations of the target scenario, taking the average value and lowering it by 5%-8% to ensure coverage of most effective behavioral matching targets. The preset semantic fit is determined by statistically analyzing the average semantic fit of semantic priority targets recognized by historical users, plus a 5%-10% error tolerance range, which avoids missing high-quality targets and reduces interference from low-fit targets.

[0052] The determination of whether to output personalized recommendation results is as follows: First, obtain the recommendation priority and percentage of each specified target in the initial recommendation list. Recommendation priority includes allowing and disallowing recommendations. If the recommendation priority is allowing recommendations and the percentage of specified targets is not less than a preset percentage, then personalized recommendation results are output. This means using the generated initial recommendation list and sorting the corresponding initial recommendation lists in descending order of recommendation priority. The preset percentage is typically set to 60%-70% of the total number of targets in the initial recommendation list. This percentage is determined by considering user browsing habits in the target recommendation scenario (e.g., the average number of recommended targets a user focuses on per visit) and the required recommendation accuracy. This ensures that the output personalized recommendation list contains enough high-quality targets (allowed recommendations) while avoiding limited user choices due to insufficient quantity. If the recommendation priority is disallowed recommendations and the percentage of specified targets is less than the preset percentage, then the initial recommendation list needs optimization, and a recommendation anomaly warning is issued. Otherwise, it indicates that neither the output nor optimization conditions are met, and a manual verification warning is issued.

[0053] Specifically, the target overlap rate represents the number of identical specified targets in the candidate recommendation table generated by the large language model and the candidate list generated by the specified domain model. It is used to quantify the consistency of the recommendation results from both models. Both the candidate recommendation table and the candidate list output targets from two dimensions: semantic understanding and behavioral analysis, based on user recommendation needs. The overlap rate reflects the consistency of the recommendations from both models.

[0054] Large language models perform semantic understanding on real-time user interaction information and unstructured content (such as user reviews, product descriptions, and content tags), and combine this with their built-in contextual association and intent reasoning capabilities to generate a set of recommendation candidates that highly matches the user's semantic needs, i.e., a candidate recommendation table. Domain-specific models focus on in-depth analysis of user behavior data, such as user purchase history, browsing time, and add-to-cart behavior in the e-commerce field, and user collection, likes, and forwarding sequences in the content field. Through reasoning on this behavior data, they generate a candidate list based on behavioral trajectories.

[0055] In this embodiment, a dual-indicator approach of correlation score and overlap rate is used for judgment. A correlation threshold is set based on historical data, and manual review is triggered in intermediate states. This avoids misjudgment based on a single indicator and reduces accuracy fluctuations. Differentiated cross-validation completes the list of non-overlapping targets, and adaptability verification filters for behavioral / semantic fit, eliminating low-quality targets to ensure that recommended targets align with user needs. Result judgment is based on allowable recommendation priority and a 60%-70% proportion of recommended targets, satisfying user selection needs while improving adaptability accuracy. When targets are not met, optimization prompts or warnings are issued to adapt to recommendation requirements in different scenarios. The entire process relies on historical data to set thresholds and record the reasons for elimination, achieving standardized recommendations while facilitating manual traceability and adjustment. This effectively alleviates the problems of poor demand coordination and insufficient feature collaboration across scenarios, improving the reliability of personalized recommendations and user satisfaction.

[0056] This invention provides a personalized recommendation system based on the collaboration of a large language model and a domain model, such as... Figure 4 The diagram shows the structure of a personalized recommendation system based on the collaboration of a large language model and a domain model. This system can include: a dual-model pattern feature output module, used to perform semantic understanding on acquired unstructured data in a target recommendation scenario using a large language model to output semantic pattern features, and simultaneously using a specified domain model to collaboratively analyze acquired user behavior data to output behavioral pattern features; a bidirectional information supplementation and mapping set construction module, used to supplement the semantic pattern features output by the large language model and the behavioral pattern features output by the specified domain model with bidirectional information to form a collaboratively optimized feature mapping set; and an initial recommendation list generation and judgment module, used to combine real-time user interaction information and the formed feature mapping set to generate an initial recommendation list through collaborative reasoning between the large language model and the specified domain model, and then dynamically sort it to determine whether to output a personalized recommendation result.

[0057] In this embodiment, the dual-model pattern feature output module parses unstructured data using a large language model to output semantic pattern features, and analyzes user behavior data using a domain model to output behavioral pattern features. The two models perform their respective functions while cooperating, avoiding the limitations of a single model in processing data and providing an accurate feature foundation for subsequent recommendations. The bidirectional information supplementation and mapping set construction module supplements the two types of features bidirectionally, constructing a collaboratively optimized feature mapping set to eliminate information discrepancies between semantic and behavioral features, making the features more comprehensive and aligned with user needs. The initial recommendation list generation and judgment module combines real-time user interaction information and the feature mapping set, generates an initial recommendation list through collaborative reasoning of the two models, and dynamically sorts and judges the output results. This not only quickly responds to real-time user preferences but also improves recommendation accuracy, reduces invalid recommendations, and significantly optimizes user experience and recommendation conversion efficiency.

[0058] 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. A 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 flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, 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. A 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. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0059] 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.

[0060] In various embodiments of the present invention, the order of the above-mentioned process numbers 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.

[0061] 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.

[0062] The above are merely specific embodiments 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 personalized recommendation method based on a large language model and a domain model, characterized in that, The method comprises: Step one, in the target recommendation scene, the semantic understanding of the obtained unstructured data is carried out by using a large language model to output semantic pattern features, and the collaborative analysis of the obtained user behavior data is carried out by using a specified field model to output behavior pattern features, wherein the user behavior data represents quantifiable operation records of the user in the target recommendation scene; Step two, the semantic pattern features output by the large language model and the behavior pattern features output by the specified field model are supplemented with bidirectional information to form a feature mapping set optimized in collaboration; Step three, combining the real-time interaction information of the user and the formed feature mapping set, an initial recommendation list is generated through the collaborative reasoning of the large language model and the specified field model, and after dynamic sorting, it is determined whether to output a personalized recommendation result, wherein the real-time interaction information of the user is used to reflect the current operation state and the change of the scenario demand, and the personalized recommendation result is used to support the rapid adaptation in the cross-modal scene; The formation of the feature mapping set optimized in collaboration specifically comprises: First, the output semantic pattern features and behavior pattern features are subjected to dimension standardization processing to convert the two types of features to a specified dimension space through a feature mapping matrix, thereby reducing the dimension difference caused by different feature sources; Second, based on the semantic-behavior feature correlation degree in the current recommendation scene, the deviation value of the current semantic-behavior feature correlation degree from the midpoint of the preset feature correlation degree range is calculated; Third, if the deviation value is not less than the upper limit of the deviation value allowed range, it is determined that the coupling redundancy is too high, and the first deviation value is used to obtain the down-regulation proportion of the semantic pattern feature weight, while the behavior pattern feature weight is up-regulated by the same proportion, wherein the sum of the semantic pattern feature weight and the behavior pattern feature weight is 1; If the deviation value is less than the upper limit of the deviation value allowed range, the current semantic pattern feature weight and the behavior pattern feature weight remain unchanged; Fourth, if the inverse of the deviation value is not less than the midpoint of the deviation value allowed range, it is determined that the coupling depth is insufficient, and the second deviation value is used to obtain the up-regulation proportion of the semantic pattern feature weight, while the behavior pattern feature weight is down-regulated by the same proportion; If the inverse of the deviation value is less than the midpoint of the deviation value allowed range, the current semantic pattern feature weight and the behavior pattern feature weight remain unchanged; The first deviation value and the second deviation value are used to quantify the degree of deviation of the feature coupling degree from the expected state; Fifth, if the deviation value is within the deviation value allowed range, the current semantic and behavior pattern feature weights remain unchanged, the two types of features are aligned and matched according to the dimensions corresponding to the current weights, and a feature mapping set that fuses general semantic knowledge and field behavior rules is obtained.

2. The personalized recommendation method based on large language model and domain model collaboration of claim 1, wherein, The specific output process of the semantic pattern features is as follows: The semantic content quantity in the unstructured data is obtained, if the semantic content quantity is greater than the preset semantic content quantity, the semantic granularity proportion is calculated based on the obtained semantic content quantity deviation, combined with the semantic granularity proportion formula, to convert the dimension of the semantic content corresponding to the semantic content quantity by gradually splitting the semantic content quantity, and the semantic content is decomposed into a preset number of semantic sub-features; The semantic granularity ratio is obtained by quantifying the difference between the preset semantic content amount and the obtained semantic content amount with respect to the preset semantic content amount, and is used to reflect the fine degree of unstructured data semantic analysis. If the semantic content amount is not greater than the preset semantic content amount, the semantic vectors corresponding to the semantic content in the unstructured data are sorted by similarity, each semantic vector is assigned a corresponding weight coefficient and coupled to obtain a semantic feature result. The vector sequence of the semantic feature result is obtained, and the semantic pattern features are generated in turn from high to low according to the semantic correlation strength, and cross-text semantic consistency verification is performed to improve the matching degree of the semantic pattern features and the user demand, wherein the vector sequence represents an ordered vector set formed by the semantic feature result according to a preset semantic logic, and the semantic correlation strength represents the matching closeness of each ordered vector to the current user demand.

3. The personalized recommendation method based on large language model and domain model collaboration according to claim 2, characterized in that, The cross-text semantic consistency verification is specifically as follows: The semantic fragments in the semantic pattern features are obtained, the semantic vectors corresponding to each semantic fragment are generated by a large language model, and the cosine similarity of each semantic vector is calculated. If the cosine similarity is higher than the preset cosine similarity, it is determined that the semantic features are consistent, and the semantic pattern features are output to ensure the output stability in the cross-text scenario. Otherwise, the semantic content is re-decomposed into a preset number of semantic sub-features at the semantic pattern feature output link, and after a preset number of decompositions, if the re-obtained cosine similarity is still not higher than the preset cosine similarity, the preset personnel is prompted to perform semantic calibration until the semantic features are determined to be consistent.

4. The personalized recommendation method based on large language model and domain model collaboration of claim 1, wherein, The specific output process of the behavior pattern feature is as follows: From the time sequence dimension of the user behavior data, the specified behavior frequency in the adjacent two statistical periods is retrieved, the difference between the specified behavior frequency in the latter period and the specified behavior frequency in the former period is defined as a time sequence difference parameter, which is used to quantify the change trend of the user behavior over time. If the absolute value of the time sequence difference parameter is greater than the preset difference threshold, the time sequence weight corresponding to the proportion of the time sequence difference parameter deviation is added in the specified field model, and whether the time sequence weight exceeds the preset time sequence weight range is monitored. If the time sequence weight exceeds the preset time sequence weight range, the feature calculation process based on the weight is suspended to prevent excessive interference with the behavior feature output process. If the time sequence weight is within the preset time sequence weight range, the monitoring continues until the absolute value of the time sequence difference parameter is not greater than the preset difference threshold, and the behavior pattern feature is output. If the absolute value of the time sequence difference parameter is not greater than the preset difference threshold, the behavior pattern feature is output based on the historical behavior rule of the corresponding user. If it is a new user, the behavior pattern feature is output based on the behavior distribution rule of the user group corresponding to the same feature label, wherein the same feature label represents a label set consistent with the current scenario and behavior rule of the new user.

5. The personalized recommendation method based on large language model and domain model collaboration of claim 1, wherein, The semantic-behavior feature correlation degree is obtained as follows: The semantic pattern features and the behavior pattern features processed by dimension standardization are subjected to Fourier transform to convert the two types of features from time domain to frequency domain, and the frequency domain intensity values are generated by analyzing the corresponding frequency spectrum distribution states to capture the correlation rules of the two types of features changing over time; The frequency domain intensity values represent the amplitude coincidence degree and phase cooperativity of each frequency component in the frequency domain space of the two types of features; The continuous values of the two types of features are mapped into 0-1 sequences by binary coding, and the cosine similarity between the binary sequences of the two types of features is calculated to form a discrete matching degree index for quantifying the matching degree of the discretized features; The discrete matching degree index represents the direction consistency degree of the two types of features in the vector space after discretization; The frequency domain intensity values and the discrete matching degree index are subjected to geometric mean operation to obtain the semantic-behavior feature correlation degree, and the geometric mean operation is used to balance the influence of the frequency domain intensity values and the discrete matching degree index on the correlation degree result.

6. The personalized recommendation method based on large language model and domain model collaboration of claim 1, wherein, The specific generation steps of the initial recommendation list are: The current operation data is retrieved from the user real-time interaction log and converted into a real-time feature vector, which is then subjected to feature matching calculation with the formed feature mapping set to obtain the correlation degree score of the real-time feature and the historical mapping feature, and the candidate recommendation list generated based on the large language model and the to-be-selected list generated based on the specified domain model are compared to obtain the specified target coincidence rate; The correlation degree score is obtained by the Euclidean distance between the real-time feature and the historical mapping feature, and is used to quantify the matching closeness of the real-time feature and the historical mapping feature; If the correlation degree score is greater than the preset correlation degree score and the specified target coincidence rate is less than the preset coincidence rate, differential cross-validation is performed to improve the matching degree of the non-coincidence specified target and the user real-time demand while supplementing the non-coincidence specified target information in the two types of lists, and the adaptability of the differential cross-validation result is verified; If the correlation degree score is not greater than the preset correlation degree score and the specified target coincidence rate is not less than the preset coincidence rate, an initial recommendation list containing the specified target ID, matching score and recommendation reason is generated based on the coincident specified target information; If the above conditions are not met, the preset personnel is prompted for manual review.

7. The personalized recommendation method based on large language model and domain model collaboration of claim 6, wherein, The adaptability verification of the differential cross-validation result is specifically: The specified target corresponding to the non-coincidence specified target information in the candidate recommendation list is recorded as a semantic priority target, and the specified target corresponding to the non-coincidence specified target information in the to-be-selected list is recorded as a behavior priority target; The behavior pattern features in the feature mapping set are called to obtain the length of the longest common subsequence with the corresponding user historical behavior sequence, and the length ratio is obtained by combining the total length of the historical behavior sequence, and the behavior fit degree of the semantic priority target is generated by weighted summation combined with the user behavior weight coefficient; The semantic pattern features in the feature mapping set are called synchronously, the cosine similarity between the attribute label of the behavior priority target and the semantic vector of the user current operation state is calculated to obtain the semantic fit degree of the behavior priority target; Only the behavior fit degree is not less than the preset behavior fit degree, and the semantic fit degree is not less than the preset semantic fit degree, the corresponding semantic priority target and the behavior priority target, the specified target which does not meet the condition is eliminated, and the elimination reason is synchronized to the preset personnel.

8. The personalized recommendation method based on large language model and domain model collaboration of claim 7, wherein, The method comprises the following steps: Obtaining the recommendation priority of each specified target in the initial recommendation list and the specified target proportion quantity, the recommendation priority including allowed recommendation results and disallowed recommendation results; If the recommendation priority is allowed recommendation results, and the specified target proportion quantity is not less than the preset proportion quantity, it is determined to output the personalized recommendation results, and the corresponding initial recommendation list is arranged in descending order according to the recommendation priority; If the recommendation priority is disallowed recommendation results, and the specified target proportion quantity is less than the preset proportion quantity, it is determined to optimize the initial recommendation list, and a recommendation exception prompt is given; Otherwise, it is indicated that neither the output condition nor the optimization condition is met, and an artificial verification warning is given.

9. The personalized recommendation system based on the cooperation of the large language model and the domain model, the personalized recommendation method based on the cooperation of the large language model and the domain model is applied according to any one of claims 1-8, characterized in that, It comprises: A double-model mode feature output module is used to perform semantic understanding on the obtained unstructured data by using a large language model to output semantic mode features, and to perform collaborative analysis on the obtained user behavior data by using a specified field model to output behavior mode features in a target recommendation scene; A bidirectional information supplement and mapping set construction module is used to supplement the semantic mode features output by the large language model and the behavior mode features output by the specified field model in a bidirectional manner to form a collaborative optimization feature mapping set; An initial recommendation list generation and determination module is used to generate an initial recommendation list by collaborative reasoning of the large language model and the specified field model in combination with real-time interaction information of the user and the formed feature mapping set, and to determine whether to output a personalized recommendation result after dynamic sorting.

Images