Document intelligent recommendation method and system fusing multi-dimensional evaluation and dynamic weight

Abstract

The invention belongs to the field of document recommendation, and provides an intelligent document recommendation method and system fusing multi-dimensional evaluation and dynamic weight in order to solve the problem that a current recommendation result is not ideal in individuation and precision. The intelligent document recommendation method integrating multi-dimensional evaluation and dynamic weight comprises the following steps: in a cold start stage, recommending a document template by utilizing a three-layer decision model; in the hot data stage, according to interaction behavior data of a user and a document template, user portrait features are dynamically updated so as to adaptively and dynamically adjust the weight of recommendation influence factors, and a document template recommendation list is generated according to a recommendation influence factor weighted summation method. The problem of accurate recommendation when a new user has no behavior data can be effectively solved; meanwhile, a dynamic weight adjustment mechanism is designed, smooth transition from cold start to a hot data stage is achieved, the bottleneck that a traditional recommendation system cannot adapt to dynamic changes of user behaviors is broken through, and various professional documents can be processed.

Description

Technical Field

[0001] The present invention belongs to the field of document recommendation, and in particular relates to a document intelligent recommendation method and system integrating multi-dimensional evaluation and dynamic weighting. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] When it comes to recommending professional documents (such as contract templates and legal documents), existing recommendation systems typically use natural language processing (NLP) technology to perform semantic analysis on document content and collaborative filtering algorithms to match users with documents. Existing recommendation algorithms perform poorly when dealing with cold start issues. When faced with new users, due to a lack of sufficient behavioral data, the system can only provide random or popularity-based recommendations, failing to objectively assess the quality of professional documents. This makes it difficult for new documents or new users to receive effective recommendations. Furthermore, when processing multi-source data, existing systems struggle to fully integrate multiple factors, such as document quality, user industry characteristics, and frequency of use. This results in suboptimal personalization and accuracy in recommendation results. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides a document intelligent recommendation method and system that integrates multi-dimensional evaluation and dynamic weights, which can effectively solve the problem of accurate recommendation when new users have no behavioral data; at the same time, a dynamic weight adjustment mechanism is designed to achieve a smooth transition from cold start to hot data stage. The system can automatically optimize the weight distribution of each factor according to user behavior data, so that the recommendation results can be upgraded from universal recommendations to personalized and accurate recommendations as user preferences evolve, breaking through the bottleneck that traditional recommendation systems cannot adapt to dynamic changes in user behavior, and can handle multiple types of professional documents.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: A first aspect of the present invention provides a document intelligent recommendation method that integrates multi-dimensional evaluation and dynamic weighting.

[0006] In one or more embodiments, a document intelligent recommendation method integrating multi-dimensional evaluation and dynamic weighting is provided, including: During the cold start phase, a three-layer decision model is used to recommend document templates. The model consists of a target layer, a criterion layer, and an indicator layer. The target layer takes recommending the best document template as its decision goal. The criterion layer includes three evaluation dimensions: quality score, industry matching, and random exploration factor. The indicator layer includes indicators for each evaluation dimension of the criterion layer. In the hot data stage, the user portrait features are dynamically updated based on the interaction behavior data between users and document templates to adaptively and dynamically adjust the weights of recommendation influencing factors. A document template recommendation list is generated based on the weighted summation method of recommendation influencing factors.

[0007] As an implementation method, the user portrait features include industry characteristics, preference characteristics and user behavior patterns.

[0008] As an implementation method, recommendation influencing factors include quality score, industry matching, usage frequency and collaborative filtering.

[0009] As an implementation method, in the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, a minimum weight threshold of the quality score is set, and the amount of interaction behavior data between users and document templates is negatively correlated with the weight of the quality score.

[0010] As an implementation method, in the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of industry matching is positively correlated with industry concentration.

[0011] As an implementation method, in the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of usage frequency is represented by a piecewise function. When the usage frequency is lower than or equal to the set threshold, the weight of usage frequency adopts the preset basic weight; when the usage frequency is higher than the set threshold, the weight of usage frequency is positively correlated with the total amount of interactive behavior data.

[0012] As an implementation method, in the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of collaborative filtering is positively correlated with the size of similar user groups.

[0013] As an implementation method, the weights of the recommendation influencing factors are dynamically optimized based on online learning, which is expressed as: ; in, is the learning rate; Functions are built based on user satisfaction indicators; Indicates the weight of the recommendation influencing factors before the update; Indicates the updated weight of the recommendation influencing factors.

[0014] A second aspect of the present invention provides a document intelligent recommendation system that integrates multi-dimensional evaluation and dynamic weighting.

[0015] In one or more embodiments, a document intelligent recommendation system integrating multi-dimensional evaluation and dynamic weighting includes: The cold start recommendation module is used in the cold start phase to recommend document templates using a three-layer decision model comprising a target layer, a criterion layer, and an indicator layer. The target layer takes recommending the optimal document template as its decision goal; the criterion layer comprises three evaluation dimensions: quality score, industry matching, and random exploration factor; and the indicator layer comprises indicators for each evaluation dimension of the criterion layer. The hot data recommendation module is used in the hot data stage. It dynamically updates the user portrait features based on the interaction behavior data between the user and the document template, adaptively and dynamically adjusts the weights of the recommendation influencing factors, and generates a document template recommendation list based on the weighted summation method of the recommendation influencing factors.

[0016] A third aspect of the present invention provides an electronic device.

[0017] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the document intelligent recommendation method integrating multi-dimensional evaluation and dynamic weighting as described above are implemented.

[0018] Compared with the prior art, the present invention has the following beneficial effects: In the cold start phase, the present invention adopts a three-layer decision model to recommend document templates, with quality score as the dominant basis and industry matching as an auxiliary reference factor while retaining the necessary random exploration space. This not only ensures the priority recommendation of high-quality contract templates, but also retains a small amount of random exploration space to avoid the recommendation results falling into local optimality, thereby ensuring that the recommendation results are both professional and diverse.

[0019] In the hot data stage, the present invention dynamically updates user portrait features based on the interaction behavior data between users and document templates, adaptively and dynamically adjusts the weights of recommendation influencing factors, and generates a document template recommendation list based on the weighted summation method of recommendation influencing factors, so that the recommendation system can achieve personalized and precise recommendations when facing different users and different scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0021] Figure 1 1 is a flow chart of a method for intelligent document recommendation integrating multi-dimensional evaluation and dynamic weighting according to an embodiment of the present invention; Figure 2 Schematic diagram of the structure of the document intelligent recommendation system integrating multi-dimensional evaluation and dynamic weighting according to an embodiment of the present invention; Figure 3 is a schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0024] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0025] Figure 1 This is a flow chart of a method for intelligent document recommendation that integrates multi-dimensional evaluation and dynamic weighting according to an embodiment of the present invention. Figure 1 The document intelligent recommendation method integrating multi-dimensional evaluation and dynamic weighting in this embodiment may include the following steps S101 to S102.

[0026] The specific implementation process of steps S101 to S102 is as follows: Step S101: In the cold start phase, a three-layer decision model is used to recommend document templates; the three-layer decision model includes a target layer, a criterion layer, and an indicator layer; the target layer takes recommending the optimal choice of document template as the decision goal; the criterion layer includes three evaluation dimensions: quality score, industry matching, and random exploration factor; the indicator layer includes indicators of each evaluation dimension of the criterion layer.

[0027] In an embodiment of the present invention, a three-tier decision model is constructed based on the Analytic Hierarchy Process (AHP). To address the uncertainty in identifying industry characteristics caused by the lack of user behavior data during the cold start phase, an embodiment of the present invention adopts a conservative weight allocation strategy, using quality rating as the dominant basis and industry matching as an auxiliary reference factor while retaining the necessary random exploration space. This weight allocation scheme ensures the priority recommendation of high-quality contract templates while retaining a small amount of random exploration space to prevent the recommendation results from falling into local optimality, thereby ensuring that the recommendation results are both professional and diverse. The system uses the eigenvector method to calculate the judgment matrix at each level. By solving the maximum eigenvalue and the corresponding eigenvector, and after normalization, the final weight of each evaluation factor is obtained. The calculation process uses a numerically stable eigenvalue solution algorithm, and sets a convergence threshold to ensure the calculation accuracy of the weight allocation results.

[0028] In the indicator layer design, the quality scoring dimension is further refined into five sub-indicators: completeness of clauses, compliance, structural standardization, balance of rights and obligations, and completeness of risk control. The industry matching dimension includes two sub-indicators: keyword coverage and professional term matching. The weights of each sub-indicator are preset through the experience of field experts, forming a systematic and quantifiable evaluation framework, which effectively achieves a precise match between legal professional requirements and actual user needs.

[0029] The following takes the document template as an example to explain in detail: In the quality scoring dimension, all templates in the contract template library are traversed and quality scores are given based on five dimensions: completeness of clauses, compliance, structural standardization, balance of rights and obligations, and risk control.

[0030] Clause completeness is assessed by comparing the legal knowledge graph to ensure coverage of essential clauses (e.g., sales contracts must include basic clauses such as subject matter, price, and method of delivery). Compliance is assessed by matching mandatory regulations with a rule engine and identifying high-risk expressions through a large language model. Structural standardization is assessed by ensuring clarity and consistency in the contract clause hierarchy and formatting to enhance readability. Balance of rights and obligations is assessed through a large model to assess the reasonableness of the allocation of rights and obligations between the parties. Risk control completeness is assessed by examining risk control clauses such as breach of contract clauses, dispute resolution mechanisms, and confidentiality obligations, and assessing the reasonableness of limitation of liability clauses. In the quality scoring model, dynamic weights are assigned to five evaluation dimensions based on the nature of the document. Weights are dynamically adjusted based on industry attributes, document complexity, and user role, enabling the quality assessment model to automatically adapt to different scenarios. During the cold start phase, recommendations are quickly generated using pre-set industry benchmark weights. As user behavior data accumulates, personalized weightings are gradually optimized. For example, sales contracts may prioritize clause completeness and risk control, employment contracts may prioritize compliance and a balance of rights and obligations, and lease contracts may prioritize structural standardization and compliance, with corresponding weightings assigned accordingly.

[0031] In the industry match assessment dimension, we extract characteristics of the user's industry and perform keyword matching using pre-built legal terminology dictionaries for industries such as finance, manufacturing, and IT. We first use the TF-IDF algorithm to calculate the coverage of the template and the user's industry keywords. Then, we use a small model (such as BERT) to calculate the semantic similarity between the template terms and the industry terms. Finally, we weight these two indicators and sum them according to a preset ratio to obtain an industry match score ranging from 0 to 1. This enables preliminary personalized recommendations even without user behavior data.

[0032] In the random exploration factor evaluation dimension, a random exploration mechanism is implemented according to a certain weight ratio (assuming it is 10%), and the ε-greedy strategy is used to randomly select candidate templates from the contract template pool with high quality scores.

[0033] The specific implementation process is as follows: Generate a random number between 0 and 1. If the value is less than 0.1, randomly select a template from the high-quality pool; otherwise, select a template based on the weighted score. Step 3: Avoid over-concentration of recommendation results on a few high-scoring templates to ensure recommendation diversity.

[0034] The quality score ( , corresponding weight ), industry matching ( , corresponding weight ) and the random exploration factor ( , corresponding weight ) is linearly weighted summed, and the recommendation score calculation formula is: .

[0035] All candidate templates are sorted in descending order by score, and the top-N recommendation results are screened and presented to the user. The embodiment of the present invention integrates multi-dimensional evaluation factors to achieve high-quality recommendations in the cold start phase.

[0036] In some optional embodiments, a pre-classified index of contract templates is pre-built to cluster high-quality templates by industry, reducing real-time computing overhead. A caching mechanism is also introduced to cache recommendation results for frequently visited industries (such as manufacturing) for 30 minutes. This significantly improves the recommendation response speed during the cold start phase and ensures system performance.

[0037] Step S102: In the hot data stage, based on the interactive behavior data between the user and the document template, the user portrait features are dynamically updated to adaptively and dynamically adjust the weights of the recommendation influencing factors, and a document template recommendation list is generated based on the weighted summation method of the recommendation influencing factors.

[0038] As user behavior data accumulates, the weight ratios of factors such as quality score, industry relevance, usage frequency and collaborative filtering are automatically adjusted according to the richness of user behavior data, achieving a smooth transition from the cold start phase dominated by quality assessment to the hot data phase with multi-factor collaborative optimization.

[0039] Continuously track and record the interactive behavior data of users and document templates. The interactive behavior data is divided into explicit behavior data (click to browse, download, favorite, actual use, etc.) and implicit behavior data (page browsing time, number of views of the same template). Based on user behavior, a behavior weight mapping function is designed to map different types of behavior to basic scores. The higher the importance of the behavior, the larger the basic score. At the same time, a time decay coefficient is introduced. Correct the historical behavior value, the time interval between the behavior occurrence time and the current time The longer the time, the more significant the decline in value, which is specifically reflected in the behavior score. An exponential decline over time: ,in, Initially set the basic score for user behavior.

[0040] Extract user portrait features from the interaction behavior data between users and document templates.

[0041] Among them, the user portrait characteristics include industry characteristics, preference characteristics and user behavior patterns.

[0042] Industry characteristics: Count the industries to which the contract templates most frequently used by users belong and calculate the industry concentration score; Preference characteristics: We can use existing large language models to analyze the characteristics of the terms in the templates downloaded by users, build personalized term preference labels, and accurately capture users' special needs in terms of term content (such as "tolerance for high-risk terms" and "dispute resolution mechanism preference"); User behavior patterns: Analyze user behavior sequences through existing clustering algorithms to identify different types of user behavior patterns (such as "urgent" or "cautious" based on operation frequency and operation interval duration).

[0043] In an embodiment of the present invention, recommendation influencing factors include quality score, industry matching, usage frequency and collaborative filtering.

[0044] In the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, a minimum weight threshold for the quality score is set, and the amount of interactive behavior data between users and document templates is negatively correlated with the weight of the quality score.

[0045] For example, the weight of the quality score ; in, The amount of user-document interaction behavior data (such as number of clicks, browsing time, etc.); is the preset minimum weight threshold (e.g. 0.3); is the adjustment coefficient, which is used to control the decay speed ( ); is the smoothing constant (to prevent The weight explodes when , and is usually set to 1).

[0046] It should be noted that the minimum weight threshold of the quality score can be set according to actual conditions. The amount of user interaction behavior data with the document template is negatively correlated with the weight of the quality score, and a specific negative correlation function can be calculated and determined according to actual conditions.

[0047] Specifically, in the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of industry matching is positively correlated with industry concentration.

[0048] For example, the weight of industry matching is ; ; in, is a positive correlation coefficient ( ), used to control the weight range (if set to 0.5, the weight range is mapped to [0,0.5]); is the industry concentration (the value range is generally [0,1]), which can be calculated by the Herfindahl-Hirschman Index (HHI) or the entropy index. Indicates the The proportion of each industry segment in user behavior or document collection; is the total number of industries. Industry concentration The higher the value (users or documents focus on a few industries), the higher the weight of the corresponding industry matching degree. The bigger.

[0049] The industry matching degree here can be represented by the similarity between the industry of the document template and the industry in the user interaction behavior data. The weight of the industry matching degree is positively correlated with the industry concentration and can be set according to the actual situation.

[0050] In the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of usage frequency is represented by a piecewise function. When the usage frequency is lower than or equal to the set threshold, the weight of usage frequency adopts the preset basic weight; when the usage frequency is higher than the set threshold, the weight of usage frequency is positively correlated with the total amount of interactive behavior data.

[0051] ; in, is the weight of the frequency of use, is the usage frequency of the current user or document (such as the number of clicks, visits, etc.), is a preset threshold that distinguishes low-frequency and high-frequency behaviors. is the basic weight (fixed weight in the low-frequency stage), is the total amount of user-document interaction data (such as total clicks, total usage times, etc.), is the adjustment coefficient (controls the extent to which the weight increases with the amount of data, which must satisfy ).

[0052] The preset basic weight here can be set specifically according to the actual situation. The weight of the frequency of use is positively correlated with the total amount of interactive behavior data, and the specific positive correlation function can be calculated and determined according to the actual situation.

[0053] In the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of collaborative filtering is positively correlated with the size of similar user groups.

[0054] ; in, is the weight of collaborative filtering; Indicates the size of the similar user group (i.e., the number of users with similar behaviors to the target user); The maximum weight limit (controls the weight range, for example, set to 0.6); is a smoothing coefficient used to control the shape of the growth curve. For example, its value range is usually [5,20].

[0055] The size of the similar user group here can be represented by the number of similar users. The weight of collaborative filtering is positively correlated with the size of the similar user group, and a specific positive correlation function can be calculated and determined based on actual conditions.

[0056] The final recommendation list is generated through a multi-factor weighted fusion calculation, taking into account factors such as quality score, industry relevance, usage frequency, and collaborative filtering. The weight of each factor is adjusted in real time according to a dynamic weight calculation model to adapt to the evolution of user preferences, thereby generating a recommendation score: ,in, 、 、 、 They are the weights of quality score, industry relevance, frequency of use and collaborative filtering score. 、 、 、 They are quality score, industry relevance, usage frequency score and collaborative filtering score.

[0057] On this basis, in order to improve the recommendation effect, a composite optimization mechanism is adopted: First, we introduce the random perturbation term , for the recommendation score After fine-tuning, the score is adjusted to , thereby avoiding excessive concentration of recommendation results on high-frequency templates and ensuring recommendation diversity; Secondly, set a time decay compensation factor for newly entered documents , acting on the recommendation score, the score after compensation: ,in, is the time when the document is stored, through the constant coefficient To adjust the decay speed to enhance the exposure opportunity of the new template. At the same time, the time decay compensation factor Can ensure the original recommendation score ranking by adjusting the constant coefficient and constant coefficient , which can balance content fluidity and recommendation stability; Finally, set mandatory filtering rules to automatically exclude high-risk document templates whose risk scores exceed the preset threshold: ,in, is the indicator function, when 1 when it is, otherwise 0; The above methods ensure the reliability of recommendations from three dimensions: score optimization, novelty assurance, and legal compliance.

[0058] In other embodiments, the weights of the recommendation influencing factors are dynamically optimized based on online learning, which is expressed as: ; in, is the learning rate; Functions are built based on user satisfaction indicators; Indicates the weight of the recommendation influencing factors before the update; Indicates the updated weight of the recommendation influencing factors.

[0059] The embodiment of the present invention collects user behavioral feedback data such as click-through conversion rate and usage time in real time, constructs a loss function based on preset satisfaction evaluation indicators, and uses optimization algorithms such as gradient descent to adjust the weight parameters of dimensions such as quality score, industry relevance, usage frequency and collaborative filtering, so as to regularly update the weight parameters of each dimension.

[0060] like Figure 2 As shown, the document intelligent recommendation system that integrates multi-dimensional evaluation and dynamic weights provided by the embodiment of the present invention can be implemented in a software manner. The document intelligent recommendation system that integrates multi-dimensional evaluation and dynamic weights includes the following software modules: a cold start recommendation module 201 and a hot data recommendation module 202.

[0061] The following is an introduction to the functions of each software module in the document intelligent recommendation system that integrates multi-dimensional evaluation and dynamic weighting: The cold start recommendation module 201 is used to recommend document templates during the cold start phase using a three-layer decision model comprising a target layer, a criterion layer, and an indicator layer. The target layer takes recommending the optimal document template as its decision goal; the criterion layer comprises three evaluation dimensions: quality score, industry matching, and random exploration factor; and the indicator layer comprises indicators for each evaluation dimension of the criterion layer. The hot data recommendation module 202 is used to dynamically update the user portrait features according to the interaction behavior data between the user and the document template in the hot data stage, so as to adaptively and dynamically adjust the weights of the recommendation influencing factors, and generate a document template recommendation list according to the weighted summation method of the recommendation influencing factors.

[0062] In the hot data recommendation module 202, the user portrait features include industry features, preference features and user behavior patterns.

[0063] Recommendation influencing factors include quality score, industry matching, usage frequency, and collaborative filtering.

[0064] In the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, a minimum weight threshold for the quality score is set, and the amount of interactive behavior data between users and document templates is negatively correlated with the weight of the quality score.

[0065] In the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of industry matching is positively correlated with industry concentration.

[0066] In the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of usage frequency is represented by a piecewise function. When the usage frequency is lower than or equal to the set threshold, the weight of usage frequency adopts the preset basic weight; when the usage frequency is higher than the set threshold, the weight of usage frequency is positively correlated with the total amount of interactive behavior data.

[0067] In the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of collaborative filtering is positively correlated with the size of similar user groups.

[0068] In some optional embodiments, the weights of the recommendation influencing factors are dynamically optimized based on online learning, which is expressed as: ; in, is the learning rate; Functions are built based on user satisfaction indicators; Indicates the weight of the recommendation influencing factors before the update; Indicates the updated weight of the recommendation influencing factors.

[0069] It should be noted here that the various modules in the document intelligent recommendation system that integrates multi-dimensional evaluation and dynamic weights in an embodiment of the present invention correspond one-to-one to the various steps in the document intelligent recommendation method that integrates multi-dimensional evaluation and dynamic weights in the above-mentioned embodiment, and the specific implementation process is the same, which will not be repeated here.

[0070] In the cold start phase, the present invention adopts a three-layer decision model to recommend document templates, with quality score as the dominant basis and industry matching as an auxiliary reference factor while retaining the necessary random exploration space. This not only ensures the priority recommendation of high-quality contract templates, but also retains a small amount of random exploration space to avoid the recommendation results falling into local optimality, thereby ensuring that the recommendation results are both professional and diverse.

[0071] In the hot data stage, the present invention dynamically updates user portrait features based on the interaction behavior data between users and document templates, adaptively and dynamically adjusts the weights of recommendation influencing factors, and generates a document template recommendation list based on the weighted summation method of recommendation influencing factors, so that the recommendation system can achieve personalized and precise recommendations when facing different users and different scenarios.

[0072] The structure of the electronic device according to the embodiment of the present invention is described in detail below. Figure 3 The schematic diagram of the structure of the electronic device provided in the embodiment of the present invention can be understood as follows: Figure 3 Only exemplary structures of the electronic device are shown, not all structures. Part or all of the shown structures may be implemented as needed.

[0073] The electronic device provided by the embodiment of the present invention includes: at least one processor 301, a memory 302, a user interface 303 and at least one network interface 304. The various components in the document intelligent recommendation system integrating multi-dimensional evaluation and dynamic weighting are coupled together through a bus system 305. It can be understood that the bus system 305 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 305 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, Figure 3 Various buses are labeled as bus system 305 .

[0074] The user interface 303 may include a display, a keyboard, a mouse, a trackball, a click wheel, keys, buttons, a touch pad or a touch screen.

[0075] It will be appreciated that the memory 302 may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memory. The memory 302 in the embodiment of the present invention can store data to support the operation of the terminal. Examples of such data include any computer program used to operate on the terminal, such as an operating system and application programs. The operating system includes various system programs, such as a framework layer, a core library layer, and a driver layer, which are used to implement various basic services and handle hardware-based tasks. The application program may include various application programs.

[0076] In some embodiments, the intelligent document recommendation system that integrates multi-dimensional evaluation and dynamic weighting provided by the embodiments of the present invention can be implemented using a combination of software and hardware. As an example, the intelligent document recommendation system that integrates multi-dimensional evaluation and dynamic weighting provided by the embodiments of the present invention can be a processor in the form of a hardware decoding processor, which is programmed to execute the intelligent document recommendation method that integrates multi-dimensional evaluation and dynamic weighting provided by the embodiments of the present invention. For example, the processor in the form of a hardware decoding processor can be one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0077] As an example, the processor 301 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., where the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0078] As an example of hardware implementation of the document intelligent recommendation system that integrates multi-dimensional evaluation and dynamic weights provided in an embodiment of the present invention, the device provided in an embodiment of the present invention can be directly executed using a processor 301 in the form of a hardware decoding processor. For example, it can be executed by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components to implement the document intelligent recommendation method that integrates multi-dimensional evaluation and dynamic weights provided in an embodiment of the present invention.

[0079] The memory 302 in the embodiment of the present invention is used to store various types of data to support the operation of the document intelligent recommendation system integrating multi-dimensional evaluation and dynamic weight, or to store data for executing Figure 1Program code for the method shown. Examples of such data include any executable instructions for operating on a document intelligent recommendation system that integrates multi-dimensional evaluation and dynamic weighting, such as executable instructions. A program that implements the document intelligent recommendation method that integrates multi-dimensional evaluation and dynamic weighting according to an embodiment of the present invention may be included in the executable instructions.

[0080] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer readable medium, the computer program including a computer program for executing Figure 1 In such an embodiment, the computer program can be downloaded and installed from a network via the communication portion and / or installed from a removable medium. When the computer program is executed by the central processing unit, the various functions defined in the apparatus of the present application are performed.

[0081] The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products of the embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0082] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A document intelligent recommendation method integrating multi-dimensional evaluation and dynamic weighting, characterized in that: include: In the cold start phase, a three-tier decision model is used to recommend document templates; The three-layer decision model includes a target layer, a criterion layer, and an indicator layer; the target layer takes the recommendation of the optimal choice of the document template as the decision-making goal; the criterion layer includes three evaluation dimensions: quality score, industry matching, and random exploration factor; and the indicator layer includes indicators for each evaluation dimension of the criterion layer; In the hot data stage, the user portrait features are dynamically updated based on the interaction behavior data between users and document templates to adaptively and dynamically adjust the weights of recommendation influencing factors. A document template recommendation list is generated based on the weighted summation method of recommendation influencing factors.

2. The method for intelligent document recommendation integrating multi-dimensional evaluation and dynamic weighting according to claim 1, characterized in that: The user portrait characteristics include industry characteristics, preference characteristics and user behavior patterns.

3. The document intelligent recommendation method integrating multi-dimensional evaluation and dynamic weighting according to claim 1, characterized in that: Recommendation influencing factors include quality score, industry matching, usage frequency, and collaborative filtering.

4. The method for intelligent document recommendation integrating multi-dimensional evaluation and dynamic weighting according to claim 3, characterized in that: In the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, a minimum weight threshold for the quality score is set, and the amount of user-document interaction behavior data is negatively correlated with the weight of the quality score.

5. The method for intelligent document recommendation integrating multi-dimensional evaluation and dynamic weighting according to claim 3, characterized in that: In the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of industry matching is positively correlated with industry concentration.

6. The method for intelligent document recommendation integrating multi-dimensional evaluation and dynamic weighting according to claim 3, characterized in that: In the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of usage frequency is represented by a piecewise function. When the usage frequency is lower than or equal to the set threshold, the weight of usage frequency adopts the preset basic weight; When the usage frequency is higher than the set threshold, the weight of the usage frequency is positively correlated with the total amount of interactive behavior data.

7. The method for intelligent document recommendation integrating multi-dimensional evaluation and dynamic weighting according to claim 3, characterized in that: In the process of adaptively and dynamically adjusting the weights of recommendation influencing factors, the weight of collaborative filtering is positively correlated with the size of similar user groups.

8. The method for intelligent document recommendation integrating multi-dimensional evaluation and dynamic weighting according to claim 1, characterized in that: The weights of the recommendation influencing factors are dynamically optimized based on online learning and are expressed as: ； in, is the learning rate; Functions are built based on user satisfaction indicators; Indicates the weight of the recommendation influencing factors before the update; Indicates the updated weight of the recommendation influencing factors.

9. A document intelligent recommendation system integrating multi-dimensional evaluation and dynamic weighting, characterized in that: include: The cold start recommendation module is used to recommend document templates using a three-layer decision model during the cold start phase. The three-layer decision model includes a target layer, a criterion layer, and an indicator layer; the target layer takes the recommendation of the optimal choice of the document template as the decision-making goal; the criterion layer includes three evaluation dimensions: quality score, industry matching, and random exploration factor; and the indicator layer includes indicators for each evaluation dimension of the criterion layer; The hot data recommendation module is used to dynamically update user portrait features based on the interaction behavior data between users and document templates in the hot data stage, adaptively and dynamically adjust the weights of recommendation influencing factors, and generate a document template recommendation list based on the weighted summation method of recommendation influencing factors.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the document intelligent recommendation method integrating multi-dimensional evaluation and dynamic weighting are implemented as described in any one of claims 1-8.

Images