Task allocation method, apparatus and electronic device

Abstract

Embodiments of the present application provide a task allocation method and device and electronic equipment, comprising: obtaining a user label of each user in a plurality of users and a task label of a target task to be allocated; determining a first similarity of each user according to the user label of each user and the task label; determining at least one second similarity corresponding to at least one historical task of each user; determining a task completion quality level of each user according to the at least one second similarity and at least one first accuracy rate corresponding to the at least one historical task of each user; determining a task matching degree of each user and the target task according to the first similarity and the task completion quality level of each user, to obtain a plurality of task matching degrees corresponding to the plurality of users; determining a target user from the plurality of users according to the plurality of task matching degrees; and allocating the target task to the target user. The above technical solution can ensure the completion quality of the task under reasonable task allocation.

Description

Technical Field

[0001] This application relates to the field of task allocation technology, and in particular to a task allocation method, apparatus and electronic device. Background Technology

[0002] In the data annotation task allocation process, common allocation methods currently include manual allocation, automatic allocation, and dynamic retrieval. Manual allocation involves administrators assigning tasks one by one based on the annotators' expertise, experience, or available time. Automatic allocation utilizes preset rules, such as average distribution or allocation based on workload, to distribute tasks to annotators. Dynamic retrieval allows annotators to actively retrieve tasks from a task pool. However, both manual allocation and dynamic retrieval, along with automatic allocation using fixed rules, can lead to inaccurate task allocation. Therefore, improving the accuracy of task allocation is a pressing issue that needs to be addressed. Summary of the Invention

[0003] This application provides a task allocation method, apparatus, and electronic device that can improve the accuracy of task allocation.

[0004] A first aspect of this application provides a task allocation method, the method comprising:

[0005] Obtain the user tags and task tags of the target tasks to be assigned for each user among multiple users. The user tags are obtained based on the user profile information, and the task tags are obtained based on the task requirements of the target tasks.

[0006] Based on the user tag and the task tag of each user, a first similarity is determined for each user, and the first similarity is used to characterize the similarity between the target task and each user;

[0007] Determine at least one second similarity for at least one historical task of each user, wherein the second similarity is used to characterize the similarity between the target task and the corresponding historical task;

[0008] Based on the at least one second similarity and at least one first accuracy corresponding to the at least one historical task of each user, the task completion quality level of each user is determined, wherein the first accuracy is used to characterize the completion quality of the corresponding historical task;

[0009] Based on the first similarity and the task completion quality level of each user, the task matching degree between each user and the target task is determined to obtain multiple task matching degrees corresponding to the multiple users;

[0010] Based on the multiple task matching degrees, the target user is determined from the multiple users;

[0011] Assign the target task to the target user.

[0012] In some possible embodiments, determining the task completion quality level of each user based on the at least one second similarity and at least one first accuracy corresponding to the at least one historical task of each user includes:

[0013] Based on the at least one first accuracy rate of each user, determine the average accuracy rate of the completion quality of the historical tasks of each user;

[0014] Determine the difference between each of the at least one first accuracy rates and the average accuracy rate to obtain at least one difference corresponding to the at least one first accuracy rate;

[0015] The task completion quality level of each user is determined by weighting and summing the at least one second similarity with the at least one difference and the average accuracy.

[0016] In some possible embodiments, determining at least one second similarity corresponding to at least one historical task of each user includes:

[0017] Based on the task tag of the target task and the task tags of at least one historical task of each user, determine the similarity of at least one second tag corresponding to the at least one historical task;

[0018] Based on the task requirements of the target task and the task requirements of each user's at least one historical task, at least one second text similarity is determined for the at least one historical task.

[0019] The at least one second similarity of each user is determined based on the at least one second tag similarity and the at least one second text similarity.

[0020] In some possible embodiments, determining the first similarity of each user based on the user tag and the task tag includes:

[0021] Based on the user tags and the task tags, determine the first tag similarity between each user and the target task;

[0022] Based on the user profile information and the task requirements, determine the first text similarity between each user and the target task;

[0023] The first similarity of each user is determined based on the first tag similarity and the first text similarity.

[0024] In some possible embodiments, determining the task matching degree between each user and the target task based on the first similarity and the task completion quality level of each user includes:

[0025] Determine the positive and negative ideal solutions for each evaluation index in the first similarity and the task completion quality level index, respectively;

[0026] The group utility value and individual regret value are calculated using the positive ideal solution, negative ideal solution, and weight of each evaluation index.

[0027] Based on the group utility value and the individual regret value, the task matching degree between each user and the target task is determined.

[0028] In some possible embodiments, the method further includes:

[0029] The task order of the multiple tasks to be assigned is determined based on the task priority of each task to be assigned and / or the task start time of each task to be assigned, wherein the multiple tasks to be assigned include the target task.

[0030] The step of determining the target user from the multiple users based on the multiple task matching degrees includes:

[0031] Based on the order of the target tasks in the task sequence, and according to the matching degree of the multiple tasks, the target user is determined from the multiple users.

[0032] In some possible embodiments, determining the target user from the plurality of users based on the plurality of task matching degrees includes:

[0033] The target user is determined from the multiple users based on the multiple task matching degrees and the workload value of each user.

[0034] In some possible embodiments, determining the target user from the plurality of users based on the plurality of task matching degrees and the workload value of each user includes:

[0035] If, among the plurality of users, there is a user who meets a first preset condition, the user who meets the first preset condition is determined as a first candidate user, and the user with the highest task matching degree among the first candidate users is determined as the target user. The first preset condition is that the user's task matching degree is greater than or equal to a task threshold and the user's workload value is less than or equal to a workload threshold; or...

[0036] If there is a user among the multiple users who meets the second preset condition, the user who meets the second preset condition is determined as the second candidate user. Based on the task matching degree and workload of each user in the second candidate user, the allocation priority of each user in the second candidate user is determined. The user with the highest allocation priority among the second candidate users is determined as the target user. The second preset condition is that the user's task matching degree is less than the task threshold and the user's workload value is less than or equal to the workload threshold.

[0037] A second aspect of this application provides a task allocation apparatus, the apparatus comprising:

[0038] The acquisition module is used to acquire the user tags and task tags of the target tasks to be assigned for each user among multiple users. The user tags are obtained based on the user profile information of the user, and the task tags are obtained based on the task requirements of the target tasks.

[0039] The determining module is configured to: determine a first similarity for each user based on the user tag and the task tag, wherein the first similarity characterizes the similarity between the target task and each user; determine at least one second similarity corresponding to at least one historical task of each user, wherein the second similarity characterizes the similarity between the target task and the corresponding historical task; determine a task completion quality level for each user based on the at least one second similarity and at least one first accuracy corresponding to the at least one historical task of each user, wherein the first accuracy characterizes the completion quality of the corresponding historical task; and determine a task matching degree between each user and the target task based on the first similarity and the task completion quality level of each user, thereby obtaining multiple task matching degrees corresponding to the multiple users;

[0040] The allocation module is used to determine a target user from the multiple users based on the multiple task matching degrees; and to allocate the target task to the target user.

[0041] A third aspect of this application provides an electronic device, including a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods described in the first aspect of this application.

[0042] A fourth aspect of this application provides a storage medium that, when instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to perform a method as described in any one of the first aspects of this application.

[0043] A fifth aspect of this application provides a computer product that includes computer program code. When this computer program code is run on a computer, it causes the computer to implement the methods proposed in the embodiments of this application.

[0044] The technical solutions provided in this application have at least the following beneficial effects:

[0045] This application provides a task allocation method, apparatus, and electronic device. For any task to be allocated (such as a target task), user tags and target task task tags can be obtained through user profile information and task requirements of the target task, respectively. Since task tags can characterize the characteristics of the target task and user tags can characterize the characteristics of the user, the first similarity obtained from user tags and task tags, which characterizes the target task and each user, determines the task matching degree between each user and the target task. Then, the target user for the final allocation of the target task is determined by multiple task matching degrees corresponding to multiple users. This can allocate the target task to a relatively suitable target user to handle the target task, improving the efficiency and accuracy of task allocation. More importantly, since the user's task completion quality level can reflect the user's task completion status, the target task and each user's task matching degree can be used to determine the task matching degree between each user and the target task. The task matching degree between each user and the target task is determined by combining the first similarity between each user and the task completion quality level of each user. This not only assigns the target task to the most suitable user to handle the target task, thus improving the efficiency and accuracy of task allocation, but also ensures the completion quality of the task as much as possible when the task allocation is reasonable. In addition, since the second similarity is used to characterize the similarity between the target task and the corresponding historical task, and the first accuracy is used to characterize the completion quality of the corresponding historical task, the task completion quality level of each user is determined by at least one second similarity corresponding to at least one historical task of each user and at least one first accuracy corresponding to at least one historical task of each user. Therefore, it can more accurately reflect the task completion quality level of each user, and thus better ensure the completion quality of the task when the task allocation is reasonable. Attached Figure Description

[0046] Figure 1 This is a schematic diagram illustrating an application scenario for task allocation proposed in an embodiment of this application;

[0047] Figure 2 This is a flowchart illustrating a task allocation method proposed in an embodiment of this application;

[0048] Figure 3 This is a flowchart illustrating the process of determining the task completion quality level of each user in a task allocation method proposed in an embodiment of this application.

[0049] Figure 4 This is a flowchart illustrating the process of determining the second similarity between each user's historical tasks and target tasks in a task allocation method proposed in an embodiment of this application.

[0050] Figure 5 This is a schematic diagram of the process for determining the first similarity between each user and the target task in a task allocation method proposed in an embodiment of this application;

[0051] Figure 6 This is a flowchart illustrating the process of determining the task matching degree between each user and the target task in a task allocation method proposed in an embodiment of this application.

[0052] Figure 7 This is a flowchart illustrating how to determine a target user from multiple users in a task allocation method proposed in an embodiment of this application.

[0053] Figure 8 This is a schematic diagram of the structure of a task allocation device proposed in an embodiment of this application;

[0054] Figure 9 This is a schematic diagram of the structure of an electronic device proposed in an embodiment of this application. Detailed Implementation

[0055] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0056] To facilitate a clear description of the technical solutions in the embodiments of this application, in the embodiments of this application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes 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, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, and c can represent: a, or b, or c, or a and b, or a and c, or b and c, or a, b, and c, where a, b, and c can be single or multiple.

[0057] In the data annotation task allocation process, common allocation methods currently include manual allocation, automatic allocation, and dynamic retrieval. Manual allocation involves administrators assigning tasks one by one based on the annotators' expertise, experience, or available time. Automatic allocation utilizes preset rules, such as average distribution or allocation based on workload, to distribute tasks to annotators. Dynamic retrieval allows annotators to actively retrieve tasks from a task pool. However, both manual allocation and dynamic retrieval, along with automatic allocation using fixed rules, can lead to inaccurate task allocation. Therefore, improving the accuracy of task allocation is a pressing issue that needs to be addressed.

[0058] In view of this, embodiments of this application provide a task allocation method, apparatus, and electronic device. The method includes: obtaining a user tag and a task tag for a target task to be allocated for each of multiple users, wherein the user tag is obtained based on the user's profile information and the task tag is obtained based on the task requirements of the target task; determining a first similarity for each user based on the user tag and the task tag, wherein the first similarity is used to characterize the similarity between the target task and each user; determining at least one second similarity corresponding to at least one historical task of each user, wherein the second similarity is used to characterize the similarity between the target task and the corresponding historical task; determining a task completion quality level for each user based on at least one second similarity and at least one first accuracy corresponding to at least one historical task of each user, wherein the first accuracy is used to characterize the completion quality of the corresponding historical task; determining a task matching degree between each user and the target task based on the first similarity and the task completion quality level of each user, thereby obtaining multiple task matching degrees for multiple users; determining a target user from the multiple users based on the multiple task matching degrees; and allocating the target task to the target user. The above technical solution can ensure the completion quality of tasks when task allocation is reasonable.

[0059] The following describes an application scenario example of an embodiment of this application. For example, Figure 1 As shown, it includes: terminal device 101 and server 102.

[0060] For example, terminal device 101 can be a device with communication capabilities such as a smartphone, smartwatch, desktop computer, or laptop, capable of accessing the network and assigning tasks via wired or wireless means, and its form includes web applications, microprograms, or clients. For example, server 102 can be an independent physical server, server cluster, or cloud server, providing basic services including cloud storage, cloud computing, and content delivery networks for calculating task matching degree.

[0061] For example, in the embodiments of this application, task allocation and task matching degree calculation can both be performed on the terminal device, and the embodiments of this application do not limit this.

[0062] After understanding an application scenario example of the embodiments of this application, we will now describe in detail the execution steps of the task allocation method proposed in the embodiments of this application, such as... Figure 2 As shown, the method includes:

[0063] Step 201: Obtain the user tag and the task tag of the target task to be assigned for each user among multiple users. The user tag is obtained based on the user profile information, and the task tag is obtained based on the task requirements of the target task.

[0064] The data acquisition and processing module collects user and task information and completes the construction of user and task tags.

[0065] For example, user tags can be obtained based on each user's profile information or through resume information and information filled in by the system. By processing the user profile information, user tags for each user among multiple users are constructed from dimensions such as basic information, work experience, professional skills, and historical performance. For example, basic information may include information such as the user's age and gender, professional skills may include the user's relevant skills in handling tasks, and historical performance may include the user's response time to tasks and various aspects of performance in their work experience, such as sense of responsibility and efficiency.

[0066] For example, user tags could be: the first user is 23 years old, female, has been working for three years, has intermediate task processing ability, and responds to tasks quickly, etc.

[0067] For example, task tags can be obtained based on the task requirements of the target task, or they can be obtained through task requirements and information entered into the system. For example, in this step, task tags for the target task are constructed by processing the task requirements and considering dimensions such as task type, skill requirements, task difficulty, and task urgency. Here, for example, task urgency indicates whether the customer is in a hurry, which is different from the priority mentioned below. For example, urgent requests made by customers at the last minute or sudden system failures that need to be fixed immediately have a high urgency level, regardless of their importance.

[0068] For example, task tags could be: first task, marked task, requires more than 3 years of marking experience, task is difficult and urgent, etc.

[0069] By using the above multi-dimensional information, we obtained the user tags and target task tags for each user among multiple users. The following section describes how to obtain the matching degree between user and target task assignment.

[0070] Step 202: Determine the first similarity of each user based on the user tag and task tag. The first similarity is used to characterize the similarity between the target task and each user.

[0071] For example, the first similarity mentioned above can be the similarity between each user's user tags and task tags calculated by the cosine similarity algorithm, or it can be calculated by the user profile information of each user and the task requirements of the target task calculated by the cosine similarity algorithm, or it can be the average similarity between the similarity between each user's user tags and task tags and the user profile information of each user and the task requirements of the target task.

[0072] Calculate the first similarity between the target task and each user, which serves as an important parameter for subsequent calculations of the task matching degree between each user and the target task.

[0073] Step 203: Determine at least one second similarity for at least one historical task of each user, whereby the second similarity is used to characterize the similarity between the target task and the corresponding historical task.

[0074] At least one historical task corresponds to at least one second similarity, with each historical task corresponding to one second similarity. A second similarity can be obtained by comparing each historical task with the target task. A user has at least one historical task; calculating the similarity between each historical task and the target task yields at least one second similarity. For example, if a user has performed five historical tasks, calculating the similarity between each of those five tasks and the target task yields five second similarities corresponding to the five historical tasks.

[0075] The following section describes how to obtain the task completion quality level of each user through the second similarity.

[0076] Step 204: Determine the task completion quality level of each user based on at least one second similarity and at least one first accuracy corresponding to at least one historical task of each user, wherein the first accuracy is used to characterize the completion quality of the corresponding historical task.

[0077] For each user, at least one historical task corresponds to at least one first accuracy rate, with each historical task representing the completion quality of the corresponding historical task. Combining this with the content corresponding to at least one second similarity rate for each historical task, it can be understood that for each user, each historical task corresponds to one first accuracy rate and one second similarity rate. For example, the task completion quality level for each user can be obtained through the first accuracy rate and second similarity rate corresponding to each historical task.

[0078] For example, the first accuracy rate is used for the completion quality of the corresponding historical task. The completion quality of the historical task can be whether the user's historical task was completed, whether it was completed on time, the satisfaction with the quality of the completed task, etc. For example, if the user completes 60% of the task alone, then the first accuracy rate can be 60%. If the user's task completion satisfaction is 80% after system evaluation, then the first accuracy rate can be 80%. The first accuracy rate can also be obtained by combining multiple factors.

[0079] The task completion quality level of each user is determined based on at least one second similarity and at least one first accuracy corresponding to at least one historical task of each user.

[0080] Step 205: Based on the first similarity and the task completion quality level of each user, determine the task matching degree between each user and the target task, so as to obtain multiple task matching degrees corresponding to multiple users.

[0081] For each user, based on the first similarity between each user and the target task, and the user's task completion quality level, we can obtain the task matching degree between each user and the target task. Thus, for multiple users, we can obtain multiple task matching degrees corresponding to multiple users, with each user corresponding to one task matching degree.

[0082] For example, the multi-criteria compromise solution ranking (VIKOR) method can be used in this step to calculate the overall task matching degree.

[0083] For example, in addition to the VIKOR method, multi-criteria decision-making or machine learning methods can also be used to calculate task matching degree.

[0084] Step 206: Based on the matching degree of multiple tasks, determine the target user from multiple users; assign the target task to the target user.

[0085] For example, based on the matching degree of multiple tasks, the user with the highest task matching degree can be determined from multiple users as the target user, that is, the user who will handle the target task. Alternatively, the target user can be determined by combining multiple task matching degrees with each user's workload.

[0086] The above technical solution, for any task to be assigned (such as a target task), can obtain user tags and target task tags respectively through user profile information and task requirements. Since task tags can characterize the characteristics of the target task and user tags can characterize the characteristics of the user, the first similarity obtained from user tags and task tags, used to characterize the target task and each user, determines the task matching degree between each user and the target task. Furthermore, by using multiple task matching degrees corresponding to multiple users, the final target user for assigning the target task can be determined. This allows the target task to be assigned to a relatively suitable target user, improving the efficiency and accuracy of task allocation. More importantly, since the user's task completion quality level can reflect the user's task performance, the first similarity between the target task and each user... By combining similarity with each user's task completion quality level, the task matching degree between each user and the target task can be determined. This not only allows the target task to be assigned to the most suitable user to handle the target task, thus improving the efficiency and accuracy of task allocation, but also ensures the completion quality of the task as much as possible when the task allocation is reasonable. In addition, since the second similarity is used to characterize the similarity between the target task and the corresponding historical task, and the first accuracy is used to characterize the completion quality of the corresponding historical task, the task completion quality level of each user is determined by at least one second similarity corresponding to at least one historical task of each user and at least one first accuracy corresponding to at least one historical task of each user. Therefore, it can more accurately reflect the user's task completion quality level, and thus better ensure the completion quality of the task when the task allocation is reasonable.

[0087] The following describes how to determine the task completion quality level for each user.

[0088] In some embodiments, the task completion quality level of each user is determined based on at least one second similarity and at least one first accuracy corresponding to at least one historical task of each user, such as... Figure 3 As shown, it includes:

[0089] Step 301: Determine the average accuracy of each user's historical task completion quality based on at least one first accuracy for each user.

[0090] In some embodiments, a numerical value is obtained by summing at least one first accuracy rate corresponding to at least one historical task processed by each user, and dividing this numerical value by the number of at least one historical task processed by each user can yield the average accuracy rate of the completion quality of each user's historical tasks.

[0091] Step 302: Determine the difference between each of the at least one first accuracy rates and the average accuracy rate to obtain at least one difference rate corresponding to at least one first accuracy rate.

[0092] For each first accuracy rate, subtract the average accuracy rate from each first accuracy rate to obtain the difference between each first accuracy rate and the average accuracy rate. In this way, at least one difference rate can be obtained for at least one first accuracy rate, with one first accuracy rate corresponding to one difference rate.

[0093] Step 303: Determine the task completion quality level for each user based on the weighted sum of at least one second similarity and at least one difference, and the average accuracy.

[0094] The task completion quality level of each user is determined by multiplying the difference between the first accuracy and the average accuracy of each user by at least one second similarity corresponding to at least one historical task of each user, and then summing the results in a weighted sum, and adding the average accuracy of each user.

[0095] For example, the formula for the task completion quality level is as follows:

[0096] (1)

[0097] in, C To assess the quality level of task completion. This represents the average accuracy of k tasks. q k This represents the first accuracy rate for the k-th historical task. This represents the second similarity score corresponding to the k-th historical task.

[0098] For example, assuming a user has 3 historical tasks, the formula for calculating the task completion quality level is as follows:

[0099] (2)

[0100] in, C x To assess the quality level of task completion. The average accuracy of the representative data. q This represents the highest accuracy rate for each historical mission. These are the three second similarities corresponding to the three historical tasks.

[0101] Formula 1 above calculates the weighted sum of the differences between the three first accuracies and the average accuracy based on the three second similarities, and then sums this sum with the average accuracy to determine the user's task completion quality level.

[0102] The aforementioned technical solution incorporates the second similarity between each user's historical tasks and the target task, and simultaneously introduces a dynamic comparison between the user's first accuracy rate on historical tasks and their individual average accuracy rate. This allows for the accurate identification of users who are both proficient and well-suited for the task. On one hand, the second similarity filters out potential risks caused by experience mismatches, ensuring that only users truly familiar with this type of task are prioritized. On the other hand, the difference between the first accuracy rate and the average accuracy rate provides the system with real-time quality signals. When a user consistently outperforms their average level on similar tasks, their weight is increased, and vice versa, thus avoiding static scoring. This not only significantly improves the accuracy and quality of task allocation but also reduces rework and review costs caused by mismatches.

[0103] The following describes the process of calculating the second similarity between at least one historical task and the target task for each user.

[0104] In some embodiments, at least one second similarity is determined for at least one historical task of each user, such as Figure 4 As shown, it includes:

[0105] Step 401: Based on the task tag of the target task and the task tag of at least one historical task of each user, determine the similarity of at least one second tag corresponding to at least one historical task.

[0106] For example, the second label similarity can be calculated using a cosine similarity algorithm. The second label similarity is determined by the task label of the target task and the task labels of at least one historical task for each user. If the task label of the target task or the task label of at least one historical task consists of multiple labels, the second label similarity can be obtained by averaging the similarities calculated for each of the multiple labels. If the task label of the target task or the task label of at least one historical task is a single overall profile label, the second label similarity can be directly obtained using cosine similarity.

[0107] Step 402: Based on the task requirements of the target task and the task requirements of at least one historical task for each user, determine at least one second text similarity corresponding to at least one historical task.

[0108] For example, at least one second text similarity corresponding to at least one historical task can be determined using similarity algorithms such as RoBERTa and cosine similarity.

[0109] In this step, the task requirement text of the target task is paired one by one with the requirement texts of all historical tasks completed by each user, forming "target task and historical task" text pairs. Then, the pre-trained language model RoBERTa is used to vectorize each text pair, unifying requirement descriptions of varying lengths and wording into dense vectors within the same semantic space. Based on a bidirectional Transformer, RoBERTa accurately captures deep semantics, keywords, and implicit constraints in task requirements through contextual representations learned from massive corpora, thus avoiding the lexical mismatch problem of traditional bag-of-words models. After obtaining the vectors, cosine similarity is used as a metric to calculate the cosine value of the angle between the target requirement vector and each historical requirement vector. The closer this value is to 1, the higher the semantic overlap; the closer it is to 0, the more significant the difference in task requirements. This method generates a second set of text similarity scores for each user, intuitively quantifying the text-level matching degree between their past experience and the current target task, providing interpretable and ranking numerical evidence for subsequent quality assessment and task distribution.

[0110] Step 402: Determine at least one second similarity for each user based on at least one second tag similarity and at least one second text similarity.

[0111] For example, a second similarity can be determined based on the second tag similarity and the second text similarity corresponding to the same historical task for each user.

[0112] For example, the formula for calculating the second similarity corresponding to any historical task is as follows:

[0113] Second similarity = (Second label similarity + Second text similarity) / 2; (3)

[0114] The aforementioned technical solution calculates the second similarity between each user's historical tasks and target tasks through a two-dimensional cross-calculation of tags and text. This allows for consideration of both macro-level profiling and micro-level semantics when recalling historical samples: on one hand, the second tag similarity uses discrete, structured task tags to quickly measure whether there is overlap in "scenes, demographics, and skill points"; on the other hand, the second text similarity uses a deep semantic model to vectorize the entire task requirement text and calculate cosine similarity, capturing wording differences, implicit constraints, and technical terms, avoiding overlooking valuable experiences with sparse tags but highly relevant content. The second similarity generated after the weighted fusion of the two is not only more accurate and interpretable, but also improves the accuracy of task allocation.

[0115] In other embodiments, the second similarity may also be a second tag similarity, or it may simply be a second text similarity.

[0116] In some embodiments, a first similarity score for each user is determined based on each user's user tag and task tag, such as... Figure 5 As shown, it includes:

[0117] Step 501: Determine the first tag similarity between each user and the target task based on the user tag and the task tag.

[0118] Based on the user tag corresponding to each user and the task tag corresponding to the target task, determine the first tag similarity between each user and the target task.

[0119] For example, the similarity of the first label can be calculated using a cosine similarity algorithm, or it can be calculated using a set-based or statistical similarity algorithm. This application does not limit the specific algorithms used.

[0120] In this step, we can encode "user tags" and "task tags" into high-dimensional sparse or dense vectors, where each dimension corresponds to a tag or the embedding representation of a tag. Then, we use cosine similarity to calculate the cosine value of the angle between the target task's tag vector and each user tag vector. This value is between 0 and 1, and the closer it is to 1, the higher the semantic or functional overlap of the tag sets of the two.

[0121] Step 502: Based on the user profile information and task requirements, determine the first text similarity between each user and the target task.

[0122] Step 502 treats user profile information, such as personal description, areas of expertise, and past achievements, as well as task requirements, such as functional description, scenario limitations, and delivery standards, as free text. For example, a pre-trained RoBERTa model can be used to map two texts to dense vectors in the same semantic space. RoBERTa's bidirectional Transformer structure can fully capture contextual dependencies, implicit semantics, and technical terms. Then, cosine similarity is used to measure the angle between the two text vectors. The result is the first text similarity, which reflects the degree to which the user's text fits the task requirements and effectively makes up for any details that the labeling system may miss.

[0123] Step 503: Determine the first similarity of each user based on the first tag similarity and the first text similarity.

[0124] In step 503, the first label similarity output in step 501 and the first text similarity output in step 502 are weighted, concatenated and fused, or averaged, for example, by linear weighting, weighted geometric mean or gated network, so as to combine the structured semantics brought by discrete labels and the rich context brought by continuous text to generate the final first similarity.

[0125] For example, the formula for calculating the first similarity is as follows:

[0126] First similarity = (first tag similarity + first text similarity) / 2; (4)

[0127] The above technical solution calculates the second similarity by simultaneously fusing tag information and text content. This retains the high interpretability and fast filtering capabilities of discrete tags while fully utilizing the rich contextual semantics contained in user profile information and task requirements. This avoids missing potential high-quality users whose tags are incomplete but whose actual abilities match the tags. After the weighted fusion of the two, the second similarity is not only more accurate, but also provides the platform with an auditable and traceable basis for task distribution, ultimately significantly improving the accuracy of matching target tasks with target users.

[0128] In other embodiments, the first similarity may also be the first tag similarity, which is not limited in this application.

[0129] In some embodiments, the task matching degree between each user and the target task is determined based on a first similarity and each user's task completion quality level, such as... Figure 6 As shown, it includes:

[0130] Step 601: Determine the positive and negative ideal solutions for each evaluation index in the first similarity and task completion quality level index respectively.

[0131] Before determining the positive and negative ideal solutions, the first similarity and task completion quality level are standardized and processed. For example, the MinMax method can be used to standardize the indicators and scale the first similarity and task completion quality level to a uniform range, such as (0, 1).

[0132] The positive and negative ideal solutions for each evaluation index in the preprocessed first similarity and task completion quality level indicators are determined respectively. The positive ideal solution is the solution for all candidate solutions in the index... j The optimal value (maximizing the objective and minimizing the objective) is the optimal value on the target (maximizing the objective and minimizing the objective). A negative ideal solution is one where all candidate solutions achieve the optimal value on the target. j The worst value (maximizing the objective takes the minimum value, minimizing the objective takes the maximum value).

[0133] For example, the ideal solution =( , ): =max(first similarity), =max(task completion quality level); negative ideal solution =( , ): =min(first similarity), =min(task completion quality level).

[0134] Step 602: Calculate the group utility value and individual regret value using the positive ideal solution, negative ideal solution and weight of each evaluation indicator.

[0135] For example, the group utility value S is calculated. i and individual regret value R i The formulas are as follows:

[0136] (5)

[0137] in, S i Indicates the first i The overall score of each alternative solution is used to determine the closest possible solution to the ideal solution. w j Indicates the first j The weight of each evaluation indicator reflects its importance in decision-making. In this embodiment, the sum of the weights is 1. : No. j The ideal solution (optimal value) for an indicator may be the maximum or minimum value (depending on the indicator type).

[0138] f ij For the first i The first scheme is in the j The actual value of each indicator. For the first j The negative ideal solution (worst value) of each indicator. The denominator is the difference between the ideal solution and the negative ideal solution, used for standardization to convert indicators with different dimensions into dimensionless relative values. The weights of the first similarity and task completion quality level indicators can both be 0.5.

[0139] For example, the individual regret value R i The formula is as follows:

[0140] (6)

[0141] R i Indicates the first i The maximum disadvantage score of an individual across all indicators reflects the degree of deviation from their weakest link. The smaller the value, the better (close to 0 indicates that even the worst indicator is close to the ideal solution; a larger value indicates that at least one indicator performs poorly).

[0142] Step 603: Determine the task matching degree between each user and the target task based on the group utility value and the individual regret value.

[0143] (7)

[0144] in, Q i As a compromise decision indicator value, v This is a tradeoff factor, which can be 0.5. Because... Q i The smaller the value, the higher the priority. Calculate 1- Q i The value serves as the overall score for each solution, representing the task matching degree (value range: 0-1).

[0145] The above technical solution obtains the task matching degree of each user for the target task by using the second similarity between the user and the target task and the task completion quality level of each user. By comprehensively considering the above multiple parameters, a more accurate task matching degree value can be obtained, thereby improving the accuracy of task allocation in the subsequent task allocation process.

[0146] The following describes how to assign target tasks to the most suitable users.

[0147] In some embodiments, such as Figure 7 As shown, the method also includes:

[0148] Step 701: Determine the task order of the multiple tasks to be assigned based on the task priority of each task to be assigned and / or the task start time of each task to be assigned. The multiple tasks to be assigned include the target task.

[0149] In some embodiments, the task order of multiple tasks to be assigned can be determined based on the task priority of each task among multiple tasks to be assigned. The higher the priority, the earlier the task appears in the order.

[0150] In other embodiments, the task order of multiple tasks to be assigned can be determined based on the task start time of each task. The earlier the task start time, the higher its priority.

[0151] In other embodiments, the task order of the multiple tasks to be assigned can be determined based on the task priority and start time of each task. In this case, a first allocation order of the at least one task can be determined based on its task priority, with higher priority tasks being assigned earlier; a second allocation order of at least one task with the same task priority can be determined based on its start time; and the allocation order of the multiple tasks to be assigned can be determined based on the second allocation order.

[0152] In some embodiments, task priority can be determined based on the importance of the task, with higher priority tasks being more important.

[0153] After determining the allocation order of the multiple tasks to be assigned, the target user is determined from multiple users based on the matching degree of multiple tasks, including:

[0154] Step 702: Based on the order of the target tasks in the task sequence, determine the target user from multiple users according to the matching degree of multiple tasks.

[0155] In this embodiment, following the steps outlined above, the matching degree between each of the multiple tasks, including the target task, and the multiple task matching degrees of the target task with the multiple task matching degrees of the target task is calculated. After sorting the multiple tasks, for the target task, the target user is determined from the multiple users according to the target task's order and the multiple task matching degrees of the target task. For example, if the multiple tasks include 5 tasks, and the target task is the second task in the order of the 5 tasks, the target user is determined from the multiple users based on the multiple task matching degrees after the first task is completed.

[0156] In other embodiments, the task order among multiple tasks can be determined. For a target task, the matching degree between the target task and multiple tasks of multiple users is determined according to the order of the target task in the task order. Then, the target user is determined from the multiple users based on the multiple task matching degrees. The specific method for determining the matching degree between the target task and multiple users can be referred to the above embodiments and will not be repeated here. For example, if the multiple tasks include 5 tasks, and the target task is the second task in the order of the 5 tasks, after the task matching degree of the first task is calculated, the task matching degree of the target task is determined according to the second similarity and the task completion quality level. Then, the target user is determined from the multiple users based on the calculated multiple task matching degrees.

[0157] In the above technical solution, the allocation order of tasks to be assigned is first determined based on their priority and start time. Then, the target user is selected from multiple users based on the order of the target task within that task sequence. This avoids high-priority tasks not being allocated and processed in a timely manner, automatically adjusts task allocation, and improves resource utilization.

[0158] The following explains how to identify target users from multiple users.

[0159] In some implementations, the user with the highest task match among multiple task match scores can be identified as the target user.

[0160] In other embodiments, instead of focusing solely on task matching, the target user can be determined by referencing the real-time workload value of each user after obtaining the task matching scores of all users.

[0161] In some implementations, target users are identified from multiple users based on the matching degree of multiple tasks, including:

[0162] Target users are identified from multiple users based on the matching degree of multiple tasks and the workload value of each user.

[0163] For example, by assigning different weights to task matching and workload, the scores of heavily loaded users can be significantly lowered. This way, even if a user has a matching score as high as 0.95, if their workload is nearing its limit, they will be ranked after a user with a matching score of 0.85 who is almost idle. This ensures accurate task allocation while avoiding assigning tasks to already overloaded users, thus significantly reducing the risk of delays and improving task quality.

[0164] In some embodiments, determining a target user from multiple users based on multiple task matching degrees and each user's workload value includes:

[0165] If there is a user among multiple users who meets the first preset condition, the user who meets the first preset condition is determined as the first candidate user, and the user with the highest task matching degree among the first candidate users is determined as the target user. The first preset condition is that the user's task matching degree is greater than or equal to the task threshold and the user's workload value is less than or equal to the workload threshold.

[0166] Alternatively, if there are users among multiple users who meet the second preset condition, the users who meet the second preset condition are identified as second candidate users. Based on the task matching degree and workload of each user in the second candidate users, the allocation priority of each user in the second candidate users is determined. The user with the highest allocation priority among the second candidate users is identified as the target user. The second preset condition is that the user's task matching degree is less than the task threshold and the user's workload value is less than or equal to the workload threshold.

[0167] In this embodiment of the application, candidate users are determined based on whether the first preset condition or the second preset condition is met, and then they are allocated according to different allocation strategies.

[0168] For example, it can be further determined whether there are any users among the multiple users who meet the first preset condition. If there are users among the multiple users who meet the first preset condition, a first matching is performed, and these users who meet the first preset condition are the first candidate users. The target user is determined based on the task matching degree, and the first candidate user with the highest task matching degree is the target user. For example, the first candidate user includes one or more users.

[0169] If none of the multiple users meet the first preset condition, then a second assessment is made regarding whether the workload value is less than or equal to the workload threshold and whether the user's task matching degree is less than the task threshold (i.e., the second candidate condition). This is used to determine the second candidate user, and then the target user is determined based on the allocation priority. The target user is the user with the highest allocation priority. For example, the second candidate user may include one or more users. In this case, the second candidate users are sorted according to allocation priority, and a greedy strategy is used to prioritize assigning tasks to users with lower current workloads and matching abilities. The formula for calculating allocation priority is as follows:

[0170] Assignment priority = Task matching degree * (1 - Workload value / Workload threshold) (8)

[0171] For example, the workload value can be the user's current real-time workload or the workload updated according to a preset frequency. The workload threshold can be the maximum load threshold corresponding to each user or the average maximum load threshold calculated from multiple users; this embodiment of the application does not impose any limitations.

[0172] The above technical solution employs different allocation strategies under various circumstances, comprehensively considering user workload and task matching, and flexibly assigns target tasks to the users most suitable for handling the tasks, thereby improving task allocation efficiency and further enhancing the quality of user completion of the target task.

[0173] In some embodiments, such as Figure 8 As shown, a task allocation device is provided, comprising: an acquisition module 801, a determination module 802, and an allocation module 803, wherein:

[0174] The acquisition module 801 is used to acquire the user tag and the task tag of the target task to be assigned for each user among multiple users. The user tag is obtained based on the user's profile information, and the task tag is obtained based on the task requirements of the target task.

[0175] The determination module 802 is used to determine a first similarity for each user based on the user tag and task tag of each user, wherein the first similarity is used to characterize the similarity between the target task and each user; determine at least one second similarity corresponding to at least one historical task of each user, wherein the second similarity is used to characterize the similarity between the target task and the corresponding historical task; determine the task completion quality level of each user based on at least one second similarity and at least one first accuracy corresponding to at least one historical task of each user, wherein the first accuracy is used to characterize the completion quality of the corresponding historical task; and determine the task matching degree between each user and the target task based on the first similarity and the task completion quality level of each user, so as to obtain multiple task matching degrees corresponding to multiple users;

[0176] The allocation module 803 is used to determine the target user from multiple users based on the matching degree of multiple tasks, and to allocate the target task to the target user.

[0177] Further limitations regarding the task allocation device can be found in the limitations of the task allocation method described above, and will not be repeated here. Each module in the aforementioned task allocation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware or independently of the processor in the terminal device, or stored in software in the memory of the terminal device, so that the processor can call and execute the operations corresponding to each module.

[0178] Another embodiment provides a computer-readable storage medium for storing a computer program. This computer program contains instructions for implementing the methods described in the embodiments of this application. By installing this computer program on a computer, the computer can execute the corresponding methods.

[0179] Another embodiment proposes a computer program product that includes computer program code. When this computer program code is run on a computer, it causes the computer to implement the methods proposed in the embodiments of this application. Thus, a user can implement these methods by using this computer program product.

[0180] In some embodiments, Figure 9 This is a schematic block diagram of the electronic device provided in the embodiments of this application.

[0181] Electronic device 900 may include: a memory 901 storing executable program code and a processor 902 coupled to the memory 901.

[0182] The processor 902 calls the executable program code stored in the memory to execute any of the methods disclosed in the embodiments of this application. The processor 902 is the control center of the electronic device, connecting various parts of the entire electronic device through various interfaces and lines. By running or executing software programs and / or modules stored in the memory, and calling data stored in the memory, it performs various functions of the electronic device and processes data, thereby performing overall monitoring of the electronic device.

[0183] The memory 901 can be used to store software programs and modules. The processor executes various functional applications and data processing of the electronic device by running the software programs and modules stored in the memory. It should be understood that in this embodiment, the processor can be a Central Processing Unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0184] In implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly manifested as execution by a hardware processor, or as a combination of hardware and software modules within the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor executes the instructions in the memory, combining them with its hardware to complete the steps of the above method. To avoid repetition, detailed descriptions are omitted here.

[0185] 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 implementation should not be considered beyond the scope of the embodiments of this application.

[0186] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0187] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0188] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0189] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0190] If the aforementioned function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application embodiment, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0191] The above description is merely a specific implementation of the embodiments of this application, but the protection scope of the embodiments of this application 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 embodiments of this application should be included within the protection scope of the embodiments of this application. Therefore, the protection scope of the embodiments of this application should be determined by the protection scope of the claims.

Claims

1. A task allocation method, characterized in that, The method includes: Obtain the user tags and task tags of the target tasks to be assigned for each user among multiple users. The user tags are obtained based on the user profile information, and the task tags are obtained based on the task requirements of the target tasks. Based on the user tag and the task tag of each user, a first similarity is determined for each user, and the first similarity is used to characterize the similarity between the target task and each user; Determine at least one second similarity for at least one historical task of each user, wherein the second similarity is used to characterize the similarity between the target task and the corresponding historical task; Based on the at least one second similarity and at least one first accuracy corresponding to the at least one historical task of each user, the task completion quality level of each user is determined, wherein the first accuracy is used to characterize the completion quality of the corresponding historical task; Based on the first similarity and the task completion quality level of each user, the task matching degree between each user and the target task is determined to obtain multiple task matching degrees corresponding to the multiple users; Based on the multiple task matching degrees, the target user is determined from the multiple users; Assign the target task to the target user; The step of determining the task completion quality level of each user based on the at least one second similarity and the at least one first accuracy corresponding to the at least one historical task of each user includes: Based on the at least one first accuracy rate of each user, determine the average accuracy rate of the completion quality of the historical tasks of each user; Determine the difference between each of the at least one first accuracy rates and the average accuracy rate to obtain at least one difference corresponding to the at least one first accuracy rate; The task completion quality level of each user is determined by the weighted sum of the at least one difference obtained by the at least one second similarity and the average accuracy. Determining at least one second similarity corresponding to at least one historical task of each user includes: Based on the task tag of the target task and the task tags of at least one historical task of each user, determine the similarity of at least one second tag corresponding to the at least one historical task; Based on the task requirements of the target task and the task requirements of each user's at least one historical task, at least one second text similarity is determined for the at least one historical task. The at least one second similarity of each user is determined based on the at least one second tag similarity and the at least one second text similarity.

2. The method according to claim 1, characterized in that, The step of determining the first similarity of each user based on the user tag and the task tag includes: Based on the user tags and the task tags, determine the first tag similarity between each user and the target task; Based on the user profile information and the task requirements, determine the first text similarity between each user and the target task; The first similarity of each user is determined based on the first tag similarity and the first text similarity.

3. The method according to claim 1, characterized in that, The step of determining the task matching degree between each user and the target task based on the first similarity and the task completion quality level of each user includes: Determine the positive and negative ideal solutions for each evaluation index in the first similarity and the task completion quality level index, respectively; The group utility value and individual regret value are calculated using the positive ideal solution, negative ideal solution, and weight of each evaluation index. Based on the group utility value and the individual regret value, the task matching degree between each user and the target task is determined.

4. The method according to any one of claims 1-3, characterized in that, The method further includes: The task order of the multiple tasks to be assigned is determined based on the task priority of each task to be assigned and / or the task start time of each task to be assigned, wherein the multiple tasks to be assigned include the target task. The step of determining the target user from the multiple users based on the multiple task matching degrees includes: Based on the order of the target tasks in the task sequence, and according to the matching degree of the multiple tasks, the target user is determined from the multiple users.

5. The method according to any one of claims 1-3, characterized in that, The step of determining the target user from the multiple users based on the multiple task matching degrees includes: The target user is determined from the multiple users based on the multiple task matching degrees and the workload value of each user.

6. The method according to claim 5, characterized in that, The step of determining the target user from the multiple users based on the multiple task matching degrees and the workload value of each user includes: If, among the plurality of users, there is a user who meets a first preset condition, the user who meets the first preset condition is determined as a first candidate user, and the user with the highest task matching degree among the first candidate users is determined as the target user. The first preset condition is that the user's task matching degree is greater than or equal to a task threshold and the user's workload value is less than or equal to a workload threshold; or... If there is a user among the multiple users who meets the second preset condition, the user who meets the second preset condition is determined as the second candidate user. Based on the task matching degree and workload of each user in the second candidate user, the allocation priority of each user in the second candidate user is determined. The user with the highest allocation priority among the second candidate users is determined as the target user. The second preset condition is that the user's task matching degree is less than the task threshold and the user's workload value is less than or equal to the workload threshold.

7. A task allocation device, characterized in that, The device includes: The acquisition module is used to acquire the user tags and task tags of the target tasks to be assigned for each user among multiple users. The user tags are obtained based on the user profile information of the user, and the task tags are obtained based on the task requirements of the target tasks. The determining module is configured to: determine a first similarity for each user based on the user tag and the task tag, wherein the first similarity characterizes the similarity between the target task and each user; determine at least one second similarity corresponding to at least one historical task of each user, wherein the second similarity characterizes the similarity between the target task and the corresponding historical task; determine a task completion quality level for each user based on the at least one second similarity and at least one first accuracy corresponding to the at least one historical task of each user, wherein the first accuracy characterizes the completion quality of the corresponding historical task; and determine a task matching degree between each user and the target task based on the first similarity and the task completion quality level of each user, thereby obtaining multiple task matching degrees corresponding to the multiple users; The allocation module is used to determine a target user from the multiple users based on the multiple task matching degrees; and to allocate the target task to the target user. The determining module is further configured to: determine the average accuracy of the completion quality of each user's historical tasks based on the at least one first accuracy of each user; determine the difference between each of the at least one first accuracy and the average accuracy to obtain at least one difference corresponding to the at least one first accuracy; determine the task completion quality level of each user by weighted summation of the at least one difference using the at least one second similarity and the average accuracy; determine at least one second tag similarity corresponding to the at least one historical task based on the task tag of the target task and the task tag of each user's at least one historical task; determine at least one second text similarity corresponding to the at least one historical task based on the task requirements of the target task and the task requirements of each user's at least one historical task; and determine at least one second similarity of each user based on the at least one second tag similarity and the at least one second text similarity.

8. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method according to any one of claims 1-6.

Images