Internet platform task processing method and system based on labor distribution

Abstract

The invention relates to the technical field of internet data processing, and particularly discloses a labor-allocated internet platform task processing method, which comprises the steps of digital medal generation, capability portrait generation, task intelligent matching, visual task decomposition, collaborative scheduling, contract signing, execution monitoring, quality evaluation, reward settlement, reputation updating, growth incentive and compliance settlement. According to the method, smooth weighting of historical reputation and latest evaluation is adopted, so that the long-term reputation of the user is prevented from being influenced by single abnormality; the reputation value is an important basis for matching priorities, and is also an access threshold for users to participate in high-level tasks and obtain more training resources, and a virtuous circle is formed. Customized courses, actual combat items or higher-level authentication tasks are automatically recommended for users whose reputation and experience reach different stages; through a closed loop of learning-practice-evaluation-upgrading, the method can help the user to continuously improve skills, and can also guarantee that the platform continuously outputs high-quality labor force.

Description

Technical Field

[0001] The present invention belongs to the technical field of Internet data processing, and in particular relates to an Internet platform-based task processing method and system with distribution according to work. Background Art

[0002] In today's digital age, internet platforms have become a crucial bridge connecting task publishers and workers, significantly promoting the optimal allocation of labor resources. However, existing internet platform-based task processing methods still have many shortcomings in terms of the core principle of "pay according to work," making it difficult to fully guarantee reasonable compensation for workers and efficient operation of the platform.

[0003] When it comes to reputation system development, existing platforms' reputation evaluation mechanisms are often inadequate. On the one hand, reputation evaluations are sparsely sized, focusing primarily on simple metrics like the number of completed tasks while neglecting important factors like task quality and user feedback. This results in reputation evaluations that fail to fully and truly reflect users' capabilities and reputation levels. On the other hand, reputation update mechanisms are illogical, with limited resilience to single, anomalous evaluations. This makes it easy for individual extreme cases to impact a user's long-term reputation, making it difficult for the reputation system to effectively incentivize and constrain users. Furthermore, reputation value plays a limited role in platform resource allocation, failing to integrate closely with other aspects like task recommendation and reward calculation, preventing the formation of an effective closed-loop incentive mechanism. This hinders users from continuously improving their capabilities and reputation.

[0004] In response to this, the inventors proposed an Internet platform-based task processing method and system based on distribution according to work to solve the above problems. Summary of the Invention

[0005] The purpose of the present invention is to provide an Internet platform task processing method and system with work-based distribution to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The Internet platform-based task processing method based on distribution according to work includes:

[0008] The digital medal generation step is used to authenticate the identity and labor qualifications of platform users to obtain digital labor certificates; the capability profile generation step is used to analyze the skills and behavior data of the digital labor certificates to obtain capability labels;

[0009] The task intelligent matching step is used to filter and sort the tasks according to the capability tags, calculate the priority, and obtain a recommended task list according to the priority sorting; the visual task decomposition step is used to perform structured decomposition on the selected tasks in the recommended task list to obtain a subtask set;

[0010] A collaborative scheduling step is used to match and dispatch the subtask set to form a work order set; a contract signing step is used to deploy and generate a corresponding blockchain smart contract for the work order set to obtain a contract instance;

[0011] An execution monitoring step is used to track the execution behavior of the contract instance to obtain an execution snapshot; a quality assessment step is used to perform multi-dimensional quality assessment aggregation on the execution snapshot to obtain a quality assessment result;

[0012] A reward settlement step is used to price and distribute the quality assessment results to obtain a reward amount; a reputation update step is used to perform reputation weighting and multi-dimensional feedback evaluation on the reward amount to obtain an updated user reputation value;

[0013] The growth incentive step is used to generate personalized growth suggestions for the reputation value and obtain a growth plan; the compliance settlement step is used to convert the remuneration amount into multiple currencies and conduct tax compliance processing, and output a payment report.

[0014] Preferably, filtering and ranking the capability tags for tasks, calculating priorities, and obtaining a recommended task list according to the priority ranking includes: using a task-user intelligent matching scoring formula to calculate the priority, and the task-user intelligent matching scoring formula is:

[0015]

[0016] Where, u: user to be matched;

[0017] t: task to be assigned;

[0018] simskill(u,t): the similarity between the user's ability label and the task requirement label (range 0,1), calculated by cosine similarity or deep matching model;

[0019] simpref(u,t): the matching degree between user preference labels (such as areas of expertise, working time periods) and task characteristics (range 0, 1);

[0020] Ru: user reputation value;

[0021] Rmax: The highest reputation value of all users on the platform, used for normalization;

[0022] Lt: Task urgency level, the larger the value, the more urgent it is;

[0023] Lmax: the maximum urgency level in the platform task pool, used for normalization;

[0024] ω1~4:weight coefficient, satisfying ∑ iwi=1, dynamically adjusted by the platform based on operating strategies.

[0025] When a new task t enters the task pool, the platform calculates Smatch for each u in the candidate user set and recommends several users with the highest scores to the task publisher or assigns them directly.

[0026] Preferably, performing multi-dimensional quality assessment aggregation on the execution snapshot to obtain a quality assessment result includes: performing multi-dimensional quality assessment aggregation using a multi-dimensional quality assessment aggregation formula to obtain a quality assessment result; the multi-dimensional quality assessment aggregation formula is:

[0027] Qtotal=αQacc+βQcom+γQinn+δQfb

[0028] Where Qacc: the accuracy score of the submitted results (such as text accuracy or image recognition accuracy), ranging from 0 to 1;

[0029] Qcom: task completeness score (the coverage of the submitted content relative to the requirements), ranging from 0 to 1;

[0030] Qinn: innovation score (for creative tasks, novelty is measured by text / image feature distance), range 0,1;

[0031] Qfb: subjective / manual score of task publishers and platform spot checks, normalized to the range of 0,1;

[0032] α, β, γ, δ: scoring weight coefficients, α + β + γ + δ = 1, depending on the task type (for example, α and β can be increased for "pure translation").

[0033] Preferably, the pricing and profit-sharing processing of the quality assessment results to obtain the remuneration amount includes: adopting an intelligent dynamic remuneration pricing formula for pricing and profit-sharing, and the intelligent dynamic remuneration pricing formula is:

[0034] Pi=Pbase×(1+μQtotal)×fweigh(t)

[0035] Where Pi: the final reward amount received by user i;

[0036] Pbase: The base remuneration for this subtask, set by the task publisher or provided by the platform standard library;

[0037] Qtotal: multidimensional quality assessment results;

[0038] μ: quality gain coefficient (e.g. 0.2 means that a full quality score can result in an additional 20% increase in revenue);

[0039] fweight(t): Work order weight function, which takes into account the comprehensive addition coefficient of factors such as task difficulty, resource scarcity, and urgency.

[0040] Preferably, performing reputation weighting and multi-dimensional feedback evaluation on the remuneration amount I to obtain an updated user reputation value includes: performing reputation weighting using a user reputation dynamic update formula, wherein the user reputation dynamic update formula is:

[0041]

[0042] Among them, Ru(old), Ru(new): user's reputation value before and after update;

[0043] λ: reputation update rate (e.g. 0.3 means new reviews have a 30% weight);

[0044] g(·): Reputation score mapping function, usually a linear or nonlinear mapping, such as

[0045] g(Q,E)=η1Q+η2E

[0046] in:

[0047] Q≡Qtotal;

[0048] Efb: subjective evaluation of the task publisher (0,1);

[0049] η1+η2=1, which can be adjusted according to the platform's risk preference.

[0050] Preferably, generating personalized growth suggestions for the reputation value J to obtain a growth plan includes: using a personalized growth plan recommendation formula to generate personalized growth suggestions, and the personalized growth plan recommendation formula is:

[0051]

[0052] Where, u: user;

[0053] c: Candidate growth resources (online courses, practical projects, etc.);

[0054] simskill(u,c): the matching degree between resource requirement skills and user capability labels;

[0055] Lc: resource difficulty level;

[0056] Iu: the user's current experience value or the number of similar resources completed;

[0057] Lmax, Imax: maximum values used for normalization;

[0058] δ1~3: weight coefficient, ∑δi=1.

[0059] Preferably, the task intelligent matching step includes:

[0060] Combining the capability tags, task requirement tags, and user reputation levels, a graph matching algorithm or a deep recommendation model is used to generate a personalized task recommendation list.

[0061] The description information of the recommended task is parsed in natural language, the task is decomposed into multiple subtasks, the logical relationships and execution priorities between the tasks are marked, and a subtask set is obtained.

[0062] Another aspect of the present invention is to provide an Internet platform-based task processing system with distribution based on work, comprising:

[0063] The digital medal generation module is used to authenticate the identity and labor qualifications of platform users to obtain digital labor certificates; the capability profile generation module is used to analyze the skills and behavior data of the digital labor certificates to obtain capability labels;

[0064] The task intelligent matching module is used to filter and sort tasks based on the capability tags, calculate priorities, and obtain a recommended task list based on the priority ranking; the visual task decomposition module is used to perform structured decomposition on the selected tasks in the recommended task list to obtain a subtask set;

[0065] A collaborative scheduling module is used to match and dispatch the subtask set to form a work order set; a contract signing module is used to deploy and generate a corresponding blockchain smart contract for the work order set to obtain a contract instance;

[0066] An execution monitoring module is used to track the execution behavior of the contract instance and obtain an execution snapshot; a quality assessment module is used to perform multi-dimensional quality assessment aggregation on the execution snapshot and obtain a quality assessment result;

[0067] A remuneration settlement module is used to perform pricing and profit sharing on the quality assessment results to obtain a remuneration amount; a reputation update module is used to perform reputation weighting and multi-dimensional feedback evaluation on the remuneration amount to obtain an updated user reputation value;

[0068] The growth incentive module is used to generate personalized growth suggestions for the reputation value and obtain a growth plan; the compliance settlement module is used to convert the remuneration amount into multiple currencies and handle tax compliance, and output a payment report.

[0069] Compared with the prior art, the present invention has the following beneficial effects:

[0070] (1) The various remuneration terms of the present invention (basic remuneration, quality gain, expedited weight, etc.) are solidified in the blockchain smart contract at the beginning of the task signing, and neither party can change it at will; the remuneration settlement and payment process are automatically triggered on the chain without human intervention, which greatly reduces the risk of human intervention and interest transfer. Accuracy, completeness, innovation, manual feedback and other multi-dimensional indicators can be quantified and combined to ensure that remuneration is linked to labor results; the weights of each dimension are flexibly configured according to the task type, so that different types of labor can obtain evaluation standards that are consistent with their characteristics, preventing "one-size-fits-all" scoring bias.

[0071] (2) The present invention uses a deep learning model to integrate the user's historical task completion performance, skill test results, and preference tags to accurately portray the individual's ability profile; achieve high-dimensional vector matching of talents and task requirements, and can quickly screen the best candidates from a large number of users and tasks, reducing the "match-reject-rematch" cycle. By assigning adjustable weights to multiple factors such as reputation value, task urgency, and difficulty level, the platform can adjust the recommendation tendency in real time according to operational strategies or market changes; it can not only take into account multiple goals such as "high-quality users first" and "urgent orders first", but also flexibly switch between different stages to improve resource utilization.

[0072] (3) This invention uses a smooth weighting of historical reputation and the latest evaluation to prevent a single anomaly from affecting a user's long-term reputation. Reputation value is not only an important basis for matching priority, but also the entry threshold for users to participate in high-level tasks and obtain more training resources, forming a virtuous cycle. For users at different levels of reputation and experience, customized courses, practical projects or higher-level certification tasks are automatically recommended. Through the closed loop of "learning-practice-evaluation-upgrade", users can continuously improve their skills and ensure that the platform continuously outputs high-quality labor. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 This is a flow chart of the Internet platform-based task processing method for distribution according to work of the present invention;

[0074] Figure 2 This is a block diagram of the Internet platform-based task processing system with work-based distribution according to the present invention. DETAILED DESCRIPTION

[0075] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0076] Example 1:

[0077] See also Figure 1 As shown in FIG, the Internet platform task processing method of distribution according to work includes:

[0078] The digital medal generation step is used to authenticate the identity and labor qualifications of platform users to obtain digital labor certificates; the capability profile generation step is used to analyze the skills and behavior data of the digital labor certificate A to obtain capability labels;

[0079] The task intelligent matching step is used to filter and sort the tasks according to the capability tags, calculate the priority, and obtain a recommended task list according to the priority sorting; the visual task decomposition step is used to perform structured decomposition on the selected tasks in the recommended task list to obtain a subtask set;

[0080] A collaborative scheduling step is used to match and dispatch the subtask set to form a work order set; a contract signing step is used to deploy and generate a corresponding blockchain smart contract for the work order set to obtain a contract instance;

[0081] An execution monitoring step is used to track the execution behavior of the contract instance to obtain an execution snapshot; a quality assessment step is used to perform multi-dimensional quality assessment aggregation on the execution snapshot to obtain a quality assessment result;

[0082] A reward settlement step is used to price and distribute the quality assessment results to obtain a reward amount; a reputation update step is used to perform reputation weighting and multi-dimensional feedback evaluation on the reward amount to obtain an updated user reputation value;

[0083] The growth incentive step is used to generate personalized growth suggestions for the reputation value and obtain a growth plan; the compliance settlement step is used to convert the remuneration amount into multiple currencies and conduct tax compliance processing, and output a payment report.

[0084] Specifically, filtering and ranking the tasks on the capability tags, calculating the priorities, and obtaining a recommended task list according to the priority ranking includes: calculating the priorities using a task-user intelligent matching scoring formula, wherein the task-user intelligent matching scoring formula is:

[0085]

[0086] Where, u: user to be matched;

[0087] t: task to be assigned;

[0088] simskill(u,t): the similarity between the user's ability label and the task requirement label (range 0,1), calculated by cosine similarity or deep matching model;

[0089] simpref(u,t): the matching degree between user preference labels (such as areas of expertise, working time periods) and task characteristics (range 0, 1);

[0090] Ru: user reputation value;

[0091] Rmax: The highest reputation value of all users on the platform, used for normalization;

[0092] Lt: Task urgency level, the larger the value, the more urgent it is;

[0093] Lmax: the maximum urgency level in the platform task pool, used for normalization;

[0094] ω1~4:weight coefficient, satisfying ∑ i wi=1, dynamically adjusted by the platform based on operating strategies.

[0095] When a new task t enters the task pool, the platform calculates Smatch for each u in the candidate user set and recommends several users with the highest scores to the task publisher or assigns them directly.

[0096] Comprehensively consider skills, preferences, reputation and task urgency to achieve "precise recommendations";

[0097] The weights are adjustable, and the platform can prioritize "high credibility first" or "urgent orders first" according to different time periods;

[0098] Effectively combine users' past behaviors with real-time needs to improve task completion efficiency and satisfaction.

[0099] Specifically, performing multi-dimensional quality assessment aggregation on the execution snapshot to obtain a quality assessment result includes: performing multi-dimensional quality assessment aggregation using a multi-dimensional quality assessment aggregation formula to obtain a quality assessment result; the multi-dimensional quality assessment aggregation formula is:

[0100] Qtotal=αQacc+βQcom+γQinn+δQfb

[0101] Where Qacc: the accuracy score of the submitted results (such as text accuracy or image recognition accuracy), ranging from 0 to 1;

[0102] Qcom: task completeness score (the coverage of the submitted content relative to the requirements), ranging from 0 to 1;

[0103] Qinn: innovation score (for creative tasks, novelty is measured by text / image feature distance), range 0,1;

[0104] Qfb: subjective / manual score of task publishers and platform spot checks, normalized to the range of 0,1;

[0105] α, β, γ, δ: scoring weight coefficients, α + β + γ + δ = 1, depending on the task type (for example, α and β can be increased for "pure translation").

[0106] The system automatically calls the AI model to calculate the first four sub-scores and aggregates them by weight to obtain the final quality score Qtotal; it combines objective and subjective evaluations to ensure the scalability of machine evaluation while incorporating the flexibility of manual review; weights can be dynamically configured for different task categories to improve the pertinence and fairness of the evaluation, providing a quantitative basis for subsequent remuneration settlement.

[0107] Specifically, the pricing and profit-sharing processing of the quality assessment results to obtain the remuneration amount includes: adopting an intelligent dynamic remuneration pricing formula for pricing and profit-sharing, and the intelligent dynamic remuneration pricing formula is:

[0108] Pi=Pbase×(1+μQtotal)×fweigh(t)

[0109] Where Pi: the final reward amount received by user i;

[0110] Pbase: The base remuneration for this subtask, set by the task publisher or provided by the platform standard library;

[0111] Qtotal: multidimensional quality assessment results;

[0112] μ: quality gain coefficient (e.g. 0.2 means that a full quality score can result in an additional 20% increase in revenue);

[0113] fweight(t): Work order weight function, which takes into account the comprehensive addition coefficient of factors such as task difficulty, resource scarcity, and urgency.

[0114] Based on the submitted task's Qtotal and pre-set parameters, the system automatically calculates each participant's Pi and pays them through a smart contract. High-quality work is rewarded, achieving a "pay-for-work, pay-for-performance" approach. Flexible bonuses for urgent and difficult tasks are introduced to motivate users to take on high-value tasks. Smart contract execution ensures fairness and transparency.

[0115] Specifically, performing reputation weighting and multi-dimensional feedback evaluation on the remuneration amount I to obtain an updated user reputation value includes: performing reputation weighting using a user reputation dynamic update formula, wherein the user reputation dynamic update formula is:

[0116]

[0117] Among them, Ru(old), Ru(new): user's reputation value before and after update;

[0118] λ: reputation update rate (e.g. 0.3 means new reviews have a 30% weight);

[0119] g(·): Reputation score mapping function, usually a linear or nonlinear mapping, for example:

[0120] g(Q,E)=η1Q+η2E

[0121] in:

[0122] Q≡Qtotal;

[0123] Efb: subjective evaluation of the task publisher (0,1);

[0124] η1+η2=1, which can be adjusted according to the platform's risk preference.

[0125] Integrate the ratings and feedback of recently completed tasks into the user's historical reputation in real time. This smooths historical reputation fluctuations to prevent a single abnormal evaluation from significantly impacting users. It distinguishes the contributions of AI evaluations and human feedback, taking into account both objective and subjective factors. This provides a stable and reliable reputation reference for task matching and incentive mechanisms.

[0126] Specifically, generating personalized growth suggestions for the reputation value J to obtain a growth plan includes: using a personalized growth plan recommendation formula to generate personalized growth suggestions, and the personalized growth plan recommendation formula is:

[0127]

[0128] Where, u: user;

[0129] c: Candidate growth resources (online courses, practical projects, etc.);

[0130] simskill(u,c): the matching degree between resource requirement skills and user capability labels;

[0131] Lc: resource difficulty level;

[0132] Iu: the user's current experience value or the number of similar resources completed;

[0133] Lmax, Imax: maximum values used for normalization;

[0134] δ1~3: weight coefficient, ∑δi=1.

[0135] In the "Growth Incentive Step," all available training programs are scored and the highest-scoring options are selected as recommendations. Recommendations are tailored to user interests and abilities while also offering appropriate challenges. This balances the needs of both experienced users seeking advancement and beginners seeking a basic understanding, creating a closed-loop incentive system that prioritizes both profit and growth.

[0136] Specifically, the task intelligent matching step includes:

[0137] Combining the capability tags, task requirement tags, and user reputation levels, a graph matching algorithm or a deep recommendation model is used to generate a personalized task recommendation list.

[0138] The description information of the recommended task is parsed in natural language, the task is decomposed into multiple subtasks, the logical relationships and execution priorities between the tasks are marked, and a subtask set is obtained.

[0139] As can be seen above, various remuneration terms (basic remuneration, quality bonus, expedited weight, etc.) are fixed in the blockchain smart contract at the beginning of the task signing and cannot be arbitrarily changed by either party. The remuneration settlement and payment process is automatically triggered on the chain, without human intervention, greatly reducing the risk of human intervention and profit transfer. Multi-dimensional indicators such as accuracy, completeness, innovation, and manual feedback can be quantified and combined to ensure that remuneration is linked to labor results. The weights of each dimension are flexibly configured according to the task type, so that different types of labor can receive evaluation standards that match their characteristics, preventing "one-size-fits-all" scoring bias.

[0140] Utilizing deep learning models, the platform integrates users' historical task performance, skill test results, and preference tags to accurately profile their individual abilities. This enables high-dimensional vector matching of talent and task requirements, enabling rapid screening of optimal candidates from a vast pool of users and tasks, reducing the "match-reject-rematch" cycle. By assigning adjustable weights to multiple factors, such as reputation, task urgency, and difficulty level, the platform can adjust its recommendations in real time based on operational strategies or market changes. This balances multiple objectives, such as prioritizing high-quality users and urgent orders, while flexibly switching between different phases to improve resource utilization.

[0141] Example 2:

[0142] Text translation tasks

[0143] Scenario Description

[0144] A content platform released an English to Chinese short article translation task T1, with a basic remuneration of P x -base=100 yuan, task urgency level Lt=2 (5-level system), maximum urgency Lmax=5.

[0145] 1. Data acquisition methods

[0146] User Skill Profile (simskill): Based on the user's historical translation samples, the cosine similarity is calculated in the sentence vector space.

[0147] User preference matching (simpref): derived from the user preference questionnaire (field, time period).

[0148] Reputation value (R u ): A comprehensive normalized score of the platform’s historical completion rate and positive review rate.

[0149] Task urgency (Lt): Scored by the task publisher when publishing.

[0150] 2. Smart matching score calculation is shown in Table 1 below:

[0151] Table 1

[0152]

[0153] The weights are: ω1=0.4, ω2=0.2, ω3=0.3, ω4=0.1.

[0154] Smatch=0.4·simskill+0.2·simpref+0.3·Ru / Rmax-0.1·Lt / Lmax

[0155] For U1: S1 = 0.4 × 0.85 + 0.2 × 0.70 + 0.3 × 0.80 - 0.1 × 0.40 = 0.34 + 0.14 + 0.24 - 0.04 = 0.68

[0156] For U2: S2 = 0.4 × 0.90 + 0.2 × 0.50 + 0.3 × 0.60 - 0.1 × 0.40 = 0.36 + 0.10 + 0.18 - 0.04 = 0.60

[0157] Result: U1 is recommended as the first choice.

[0158] 3. Quality assessment and aggregation are shown in Table 2 below:

[0159] Table 2

[0160] index U1 score illustrate Weight Accuracy Qacc 0.95 AI semantic alignment + manual proofreading 0.5 Completeness Qcom 0.90 Compare original text coverage 0.3 Innovation 0.20 Translate and polish novel short sentences 0.1 Publisher Feedback Qfb 0.90 Manual scoring by publisher 0.1

[0161] Qtotal=0.5×0.95+0.3×0.90+0.1×0.20+0.1×0.90

[0162] =0.475+0.27+0.02+0.09

[0163] =0.855

[0164] 4. Remuneration Settlement

[0165] Basic remuneration P x -base=100 yuan; quality gain coefficient μ=0.2; work order weight fweight=1.1 (slightly expedited).

[0166] PU1=100×(1+0.2×0.855)×1.1=100×(1+0.171)×1.1=100×1.171×1.1≈128.81 yuan

[0167] 5. Reputation Update

[0168] Old reputation Rold = 0.80; update rate λ = 0.3; feedback mapping g(Q, E) = 0.7Q + 0.3E, E = publisher feedback = 0.90.

[0169] g=0.7×0.855+0.3×0.90=0.5985+0.27=0.8685

[0170] Rnew=(1-0.3)×0.80+0.3×0.8685=0.56+0.2606=0.8206

[0171] From the above, we can see that the implementation of the above solution can bring the following effects:

[0172] Improved matching accuracy: U1's matching score is 13% higher than U2 → Reduced invalid recommendations;

[0173] Significant quality incentives: high-quality submissions can earn an additional ≈28.8%;

[0174] Smooth credit update: A single task increases new credit by ≈2.06%, ensuring the continuity of users' long-term performance.

[0175] Example 3:

[0176] UI design tasks

[0177] Scenario Description

[0178] The platform releases UI design task T2, with a total reward of P 2- Base = 2000 yuan, urgency Lt = 4, Lmax = 5, split into two subtasks: D1 (prototype design) and D2 (high-fidelity cut-out), with work order weights f1 = 1.0 and f2 = 1.2 respectively.

[0179] 1. Data Acquisition

[0180] Decomposition subtask (D): NLP automatic parsing requirement document generation;

[0181] Abilities and preferences: User scores on the "Visual Design" and "Interaction Design" tests, as well as self-assessment of "Tool Expertise";

[0182] Execution monitoring snapshot: The SDK automatically collects the number of file submissions and reviews during the design process.

[0183] 2. Smart matching calculation is shown in Table 3 below:

[0184] Table 3

[0185] user simskill simpref Ru Ru / Rmax Lt / Lmax U3 0.80 0.60 0.85 0.85 0.80 U4 0.75 0.70 0.90 0.90 0.80

[0186] Similarly, weights ω1-4 = [0.4, 0.2, 0.3, 0.1]:

[0187] U3: 0.4×0.80+0.2×0.60+0.3×0.85-0.1×0.80=0.32+0.12+0.255-0.08=0.615

[0188] U4: 0.4×0.75+0.2×0.70+0.3×0.90-0.1×0.80=0.30+0.14+0.27-0.08=0.63

[0189] Result: D1 is assigned to U4 first, and D2 is assigned to U3.

[0190] 3. Quality assessment (molecular tasks) is shown in Table 4 below:

[0191] Table 4

[0192]

[0193] 4. Remuneration Settlement

[0194] D1:P 1- base=1200, μ=0.2, f1=1.0

[0195] PU4 = 1200 × (1 + 0.2 × 0.883) × 1.0 = 1200 × 1.1766 ≈ 1411.9 yuan

[0196] D2:P 2- base=800, μ=0.2, f2=1.2

[0197] PU3 = 800 × (1 + 0.2 × 0.816) × 1.2 = 800 × 1.1632 × 1.2 ≈ 1115.9 yuan

[0198] 5. Reputation updates are shown in Table 5 below:

[0199] Table 5

[0200]

[0201]

[0202] 6. The comparison of technical effects is shown in Table 6 below:

[0203] Table 6

[0204] U4(D1) U3(D2) Matching score 0.63 0.615 Quality score Qtotal 0.883 0.816 Final remuneration (yuan) 1411.9 1115.9 Reputation Enhancement ±0.0054 (basically maintained) ±0.0059 (basically maintained)

[0205] Efficient allocation: approximate matching scores, automatic and detailed order splitting and dispatching;

[0206] Differentiated incentives: High-fidelity cutting maps still receive +39.5% revenue due to secondary innovation and work order bonuses;

[0207] Smooth reputation: After multiple rounds of tasks, the reputation fluctuation of two users is less than 0.01, ensuring long-term stability.

[0208] Through the above two embodiments, the practical application and technical effects of the method of the present invention in the whole process of "intelligent matching → automatic splitting → quality assessment → dynamic settlement → reputation maintenance" are clearly demonstrated.

[0209] As can be seen above, a smooth weighting of historical reputation and recent reviews is used to prevent a single anomaly from impacting a user's long-term reputation. Reputation serves as both an important basis for matching priorities and a barrier to entry for participating in higher-level tasks and gaining access to more training resources, creating a virtuous cycle. Customized courses, practical projects, or higher-level certification tasks are automatically recommended for users at different levels of reputation and experience. This closed-loop "learning-practice-evaluation-upgrade" not only helps users continuously improve their skills but also ensures a continuous flow of high-quality labor on the platform.

[0210] Example 4:

[0211] The internet platform-based task processing system with distribution based on work includes:

[0212] The digital medal generation module is used to authenticate the identity and labor qualifications of platform users to obtain digital labor certificates; the capability profile generation module is used to analyze the skills and behavior data of the digital labor certificates to obtain capability labels;

[0213] The task intelligent matching module is used to filter and sort tasks based on the capability tags, calculate priorities, and obtain a recommended task list based on the priority ranking; the visual task decomposition module is used to perform structured decomposition on the selected tasks in the recommended task list to obtain a subtask set;

[0214] A collaborative scheduling module is used to match and dispatch the subtask set to form a work order set; a contract signing module is used to deploy and generate a corresponding blockchain smart contract for the work order set to obtain a contract instance;

[0215] An execution monitoring module is used to track the execution behavior of the contract instance and obtain an execution snapshot; a quality assessment module is used to perform multi-dimensional quality assessment aggregation on the execution snapshot and obtain a quality assessment result;

[0216] A remuneration settlement module is used to perform pricing and profit sharing on the quality assessment results to obtain a remuneration amount; a reputation update module is used to perform reputation weighting and multi-dimensional feedback evaluation on the remuneration amount to obtain an updated user reputation value;

[0217] The growth incentive module is used to generate personalized growth suggestions for the reputation value and obtain a growth plan; the compliance settlement module is used to convert the remuneration amount into multiple currencies and handle tax compliance, and output a payment report.

[0218] Its beneficial effects are the same as the technical effects of the Internet platform task processing method with distribution according to work, which will not be repeated here.

[0219] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An Internet platform-based task processing method based on distribution according to work, characterized in that: include: Authenticate the identity and labor qualifications of platform users and obtain digital labor certificates; Perform skill and behavior data analysis on the digital labor credentials to obtain capability labels; Filter and sort the tasks based on the capability tags, calculate the priorities, and obtain a recommended task list based on the priorities; perform structural splitting on the selected tasks in the recommended task list to obtain a subtask set; Matching and assigning the subtask set to form a work order set; deploying and generating a corresponding blockchain smart contract for the work order set to obtain a contract instance; Tracking the execution behavior of the contract instance to obtain an execution snapshot; performing multi-dimensional quality assessment aggregation on the execution snapshot to obtain a quality assessment result; The quality assessment results are priced and distributed to obtain a remuneration amount; Performing reputation weighting and multi-dimensional feedback evaluation on the remuneration amount to obtain an updated user reputation value; Generate personalized growth suggestions for the reputation value to obtain a growth plan; Perform multi-currency conversion and tax compliance processing on the remuneration amount and output payment report.

2. The Internet platform task processing method based on distribution according to work according to claim 1 is characterized in that: The filtering and sorting of tasks based on the capability tags, the calculating of priorities, and the obtaining of a recommended task list based on the priority sorting include: using a task-user intelligent matching scoring formula to calculate the priority, wherein the task-user intelligent matching scoring formula is: Where, u: user to be matched; t: task to be assigned; simskill(u,t): the similarity between the user's ability label and the task requirement label, calculated using cosine similarity or deep matching model; simpref(u,t): the matching degree between user preference label and task characteristics; Ru: user reputation value; Rmax: The highest reputation value of all users on the platform, used for normalization; Lt: Task urgency level, the larger the value, the more urgent it is; Lmax: the maximum urgency level in the platform task pool, used for normalization; ω1~4:weight coefficient, satisfying ∑ i wi=1, dynamically adjusted by the platform based on operating strategies.

3. The Internet platform task processing method based on distribution according to work according to claim 1 is characterized in that: The performing multi-dimensional quality assessment aggregation on the execution snapshot to obtain a quality assessment result includes: performing multi-dimensional quality assessment aggregation using a multi-dimensional quality assessment aggregation formula to obtain a quality assessment result; the multi-dimensional quality assessment aggregation formula is: Qtotal=αQacc+βQcom+γQinn+δQfb Where Qacc: the accuracy score of the submitted results; Qcom: task completeness score; Qinn: innovation score; Qfb: subjective and manual scoring by task publishers and platform spot checks; α, β, γ, δ: scoring weight coefficients, α + β + γ + δ = 1, based on the task type.

4. The Internet platform task processing method based on distribution according to work according to claim 1 is characterized in that: The pricing and profit sharing processing of the quality assessment results to obtain the remuneration amount includes: adopting an intelligent dynamic remuneration pricing formula to perform pricing and profit sharing processing, and the intelligent dynamic remuneration pricing formula is: Pi=Pbase×(1+μQtotal)×fweigh(t) Where Pi: the final reward amount received by user i; Pbase: The base remuneration for this subtask, set by the task publisher or provided by the platform standard library; Qtotal: multidimensional quality assessment results; μ: mass gain coefficient; fweight(t): Work order weight function, which takes into account the comprehensive addition coefficient of task difficulty, resource scarcity, and urgency factors.

5. The Internet platform task processing method based on distribution according to work according to claim 1 is characterized in that: The performing of reputation weighting and multi-dimensional feedback evaluation on the remuneration amount I to obtain an updated user reputation value includes: performing reputation weighting using a user reputation dynamic update formula, wherein the user reputation dynamic update formula is: in, and Respectively, the user's reputation value before and after update; λ: reputation update rate; g(·): reputation score mapping function, Efb: subjective evaluation of the task publisher; η1+η2=1, adjusted according to the platform's risk preference.

6. The Internet platform task processing method based on distribution according to work according to claim 1 is characterized in that: Generating personalized growth suggestions for the reputation value J to obtain a growth plan includes: using a personalized growth plan recommendation formula to generate personalized growth suggestions, wherein the personalized growth plan recommendation formula is: Where, u: user; c: candidate growth resources; simskill(u,c): the matching degree between resource requirement skills and user capability labels; Lc: resource difficulty level; Iu: the user's current experience value or the number of similar resources completed; Lmax, Imax: maximum values used for normalization; δ1~3: weight coefficient, ∑δi=1.

7. The Internet platform task processing method based on distribution according to work according to claim 1 is characterized in that: The task intelligent matching step includes: Combining the capability tags, task requirement tags, and user reputation levels, a graph matching algorithm or a deep recommendation model is used to generate a personalized task recommendation list. The description information of the recommended task is parsed in natural language, the task is decomposed into multiple subtasks, the logical relationships and execution priorities between the tasks are marked, and a subtask set is obtained.

8. An Internet platform-based task processing system based on distribution according to work, characterized in that: include: The digital medal generation module is used to authenticate the identity and labor qualifications of platform users and obtain digital labor certificates; A capability profile generation module, used to analyze the skills and behavior data of the digital labor certificate to obtain a capability label; A task intelligent matching module is used to filter and sort tasks based on the capability tags, calculate priorities, and obtain a recommended task list based on the priority order; A visual task decomposition module is used to perform structural decomposition on the selected tasks in the recommended task list to obtain a subtask set; A collaborative scheduling module is used to match and dispatch the subtask set to form a work order set; a contract signing module is used to deploy and generate a corresponding blockchain smart contract for the work order set to obtain a contract instance; An execution monitoring module is used to track the execution process of the contract instance and obtain an execution snapshot; A quality assessment module, configured to perform multi-dimensional quality assessment aggregation on the execution snapshot to obtain a quality assessment result; A remuneration settlement module is used to perform pricing and profit sharing on the quality assessment results to obtain a remuneration amount; A reputation update module, configured to perform reputation weighting and multi-dimensional feedback evaluation on the reward amount to obtain an updated user reputation value; A growth incentive module, configured to generate personalized growth suggestions for the credit value and obtain a growth plan; The compliance settlement module is used to perform multi-currency conversion and tax compliance processing on the remuneration amount and output a payment report.

Images