Task allocation method based on current demand and historical demand

Abstract

The invention relates to the technical field of data processing, in particular to a current demand and historical demand-based task allocation method, which comprises the following steps of: collecting current and historical demand document data and personnel-historical demand score data; respectively calculating theme distribution in the current demand document data and the historical demand document data; calculating theme similarity between the current demand document data and the historical demand document data according to the two theme distributions; predicting personnel-current demand score data based on the obtained topic similarity; the personnel-current demand score data and the personnel-historical demand score data are combined, comprehensive scores of the personnel are obtained, and a task allocation list is generated according to the comprehensive scores; according to the method, the most suitable task is recommended for the personnel on the basis of the score of the personnel on the historical demand, and the fair, reasonable and efficient task allocation is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular relates to a task allocation method based on current demand and historical demand. Background Art

[0002] Software engineering is a systematic approach that applies engineering principles and methods to the software development process, aiming to improve the efficiency, quality, and maintainability of software development to adapt to rapidly changing technology and business needs. In the software engineering process, documentation is an important part of project development, used to standardize the project development process, improve team collaboration efficiency, and ensure software quality. Documents include project management documents, technical specification documents, requirement documents, design documents, architecture documents, test documents, user manuals, development manuals, operation and maintenance documents, and code documents. In recent years, with the rapid development of big data and artificial intelligence technologies, automated and intelligent demand analysis and task allocation methods have gradually attracted attention.

[0003] Among them, LDA topic model and collaborative filtering algorithm, as two important text analysis and recommendation system technologies, have shown great potential in the field of software engineering. The topic extraction of text information such as task description of the requirement document is carried out through the LDA topic model, and the complex text information is converted into a concise topic distribution representation. This method can effectively reveal the intrinsic connection between texts, help developers quickly understand project requirements, and lay the foundation for subsequent task allocation. For example, a corresponding task set can be built for each requirement topic, so that developers can choose appropriate tasks according to their skills and interests. The collaborative filtering algorithm is used to analyze the similarity between developers, including skill similarity, historical task completion status and other factors, to achieve intelligent task allocation. This method can not only improve the accuracy of task allocation, but also promote collaboration and communication within the team. For example, tasks with high similarity can be recommended to developers based on their skill level and historical task completion status, thereby improving development efficiency and task completion quality. Summary of the invention

[0004] In view of this, the present invention aims to provide a task allocation method based on current and historical requirements, combining historical requirement documents and current requirement documents, estimating personnel's evaluation scores for current requirements, and then generating a task allocation list that is more in line with software engineering requirements, thereby achieving more intelligent and precise task allocation for software engineering requirements, thereby improving the efficiency and quality of the software development process.

[0005] To achieve the above object, the technical solution created by the present invention is implemented as follows: A task allocation method based on current demand and historical demand, comprising: S1: Collect current requirement document data, historical requirement document data, and personnel-historical requirement score data; S2: Calculate the topic distribution in the current demand document data and the historical demand document data in step S1 respectively; S3: Calculate the topic similarity between the current demand document data and the historical demand document data in step S1 according to the two topic distributions obtained in step S2; S4: Based on the topic similarity obtained in step S3, predict the personnel-current demand score data; S5: Combine the personnel-current demand score data obtained in step S4 and the personnel-historical demand score data in step S1 to obtain the comprehensive score of the personnel, and generate a task allocation list based on the comprehensive score.

[0006] Furthermore, step S1 also includes: preprocessing the current demand document data and the historical demand document data; wherein the preprocessing includes: text cleaning, noise removal, word segmentation, and stop word removal of the current demand document data and the historical demand document data.

[0007] Furthermore, in step S2, two topic distributions are calculated respectively using the LDA model.

[0008] Furthermore, in step S2, the training of the LDA model is also included, including: taking the requirement document set and the predefined number of topics as input, and taking the topic distribution of each requirement document and the vocabulary distribution of each topic as output, to train the LDA model; during the training process, Gibbs sampling is used to update the topic allocation of each word.

[0009] Furthermore, in step S3, the JS divergence of the two topic distributions obtained in step S2 is calculated; and the topic similarity is obtained based on the JS divergence.

[0010] Furthermore, in step S4, based on the topic similarity and the person-historical demand score data, the collaborative filtering algorithm is used to predict the person-current demand score data: ; in, Indicates the predicted completion status of personnel u on the current demand document data a - the current demand score, The personnel-historical requirement score that represents the completion of the historical requirement document data b by personnel u, Indicates the topic similarity between the current requirement document data a and the historical requirement document data b. Indicates the highest ranking of topic similarity between the historical requirement document data b and the current requirement document data a in the collection A collection of historical requirements document data.

[0011] Furthermore, in step S5, the comprehensive score is calculated by the following formula: ; in, It represents the comprehensive score of personnel u on the completion of the current requirement document data a. The personnel-historical requirement score that represents the completion of the historical requirement document data b by personnel u, Indicates personnel-historical demand score The indicator setting weights, Indicates the predicted completion status of personnel u on the current demand document data a - the current demand score, Indicates personnel-current demand score The indicator setting weights, Indicates the time that person u participated in the development of the current requirements document The normalized coefficient of Represents the normalization coefficient Set the weights of the indicators.

[0012] Furthermore, the process of generating a task assignment list according to the comprehensive score in step S5 includes: Generate a list of personnel by comprehensive score; Based on the estimated workload, obtain the specified number of personnel in the personnel list, and the specified number of personnel meets the following requirements:

[0013] Where C represents the number of designated personnel, Indicates the time when the cth designated person participated in the development of the current requirement document data, Indicates the deadline for the requirement. Indicates workload; Match the personnel list with the specified number of personnel to obtain the task assignment list.

[0014] Compared with the prior art, the invention can achieve the following beneficial effects: (1) The task allocation method based on current and historical needs created by the present invention comprehensively considers multi-dimensional information such as the skill level, historical work performance, and current workload of team members, and intelligently recommends the most suitable task for each member, thereby achieving fair, reasonable, and efficient task allocation. Through algorithm optimization, it reduces the task allocation deviation caused by human factors and improves team collaboration efficiency. Through the implementation of the present invention, the field of software engineering is promoted to develop in a more intelligent and automated direction, improving software development efficiency and quality, reducing development costs, and creating greater commercial value. (2) The task allocation method based on current and historical needs created by the present invention uses a probabilistic topic model such as LDA to identify large-scale document sets, automatically extract key demand features from multi-source data such as project documents and user feedback, reduce manual intervention, and improve the accuracy and efficiency of demand analysis. At the same time, through the continuous learning and optimization of the machine learning model, the system's ability to identify complex and fuzzy needs is improved; (3) The task allocation method based on current demand and historical demand created by the present invention establishes a flexible demand response mechanism. When new demand occurs, the system can quickly evaluate and provide demand analysis and task allocation solutions to ensure that the project proceeds smoothly as planned; through data-driven decision support, demand can be handled in a timely and effective manner, reducing the impact on project schedule and cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings constituting part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments and descriptions of the present invention are used to explain the present invention and do not constitute an improper limitation on the present invention. In the drawings: Figure 1 A flowchart of a method for allocating tasks based on current and historical demands according to an embodiment of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the invention more clear, the invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described here are only used to explain the invention and do not constitute a limitation of the invention.

[0017] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0018] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

[0019] like Figure 1 As shown, the task allocation method based on current demand and historical demand described in the embodiment of the present invention includes: S1: Collect current requirement document data, historical requirement document data, and personnel-historical requirement score data.

[0020] In some embodiments, step S1 also includes preprocessing the current demand document data and the historical demand document data. Specifically, the preprocessing includes: removing noise, segmenting, and removing stop words from the current demand document data and the historical demand document data. In a certain embodiment, for the analysis of Chinese demand documents, unnecessary punctuation, spaces, and line breaks are matched by regular expressions, and the matched content is replaced with an empty string to remove noise; Chinese text objects are segmented by ICTCLAS (Institute of Computing Technology, Chinese Lexical Analysis System); and stop words that have no practical meaning in the document are matched using a Chinese stop word library to remove stop words.

[0021] S2: Calculate the topic distribution in the current demand document data and the historical demand document data in step S1 respectively.

[0022] In some embodiments, in step S2, two topic distributions are calculated respectively using LDA (Latent Dirichlet Allocation). In a specific embodiment, the topic distributions of the current demand document data and the historical demand document data satisfy the following joint distribution. For any type of demand document data, the joint distribution can be expressed as:

[0023] in: Represents the set of words in all requirement document data including current requirements and historical requirements. Represents the requirements document data Middle words; represents the topic assignment of each word in all requirement document data, It is the requirement document data Middle The subject of the word; Represents the topic distribution of all requirement document data, It is the requirement document data The distribution of topics; represents the word distribution of all topics, It is The word distribution of each topic; is the hyperparameter of the Dirichlet distribution, controlling the topic distribution The sparsity of is the hyperparameter of the Dirichlet distribution, controlling the word distribution The sparsity of Represents the topic distribution of the demand document data. This parameter is prior information and can be expressed as: ; in, is the Dirichlet distribution, Represents the collection of all requirement document data; Represents the word distribution of the topic. This parameter is prior information and can be expressed as: ; in, is a Dirichlet distribution, Indicates the number of predefined topics in the LDA model, which is set by the user; represents the topic assignment distribution, which can be expressed as: ; in, is a discrete distribution, It is the requirement document data The total number of words; Represents the word generation distribution, which can be expressed as: ; in, is a discrete distribution.

[0024] In summary, the topic distribution of the demand document data also satisfies: .

[0025] Before using the LDA model to calculate the two topic distributions respectively, the LDA model needs to be trained, including: A collection of all requirements documents including current and historical requirements and a predefined number of topics The LDA model is trained with the topic distribution of each requirement document and the vocabulary distribution of each topic as input and output. During the training process, Gibbs sampling is used to update the topic assignment of each word. At each update, the conditional probability of assigning the word to different topics in the requirement document is calculated for each word, and the topic is reassigned to the word based on these conditional probabilities. Iterate multiple times until convergence.

[0026] The Gibbs sampling formula is as follows: ; in, Indicates that the current word is not included The set of all words of Indicates that the current word is not included The set of topic assignments for all words in Indicates the topic number, represents all possible topic numbers, Represents the vocabulary in all requirements documents, represents an index in the vocabulary, Representation Requirements Document (excluding the current word ) is assigned to the topic The number of words, Indicates the subject (excluding the current word ) contains words The above formula is about the sampling rule of index, which is updated through iteration Gradually converge to the posterior distribution of the joint distribution.

[0027] S3: Based on the two topic distributions obtained in step S2, calculate the topic similarity between the current demand document data and the historical demand document data in step S1.

[0028] In some embodiments, the JS divergence of the two topic distributions is calculated, and the topic similarity is obtained based on the JS divergence.

[0029] Among them, JS divergence is a symmetric extension of KL divergence (Kullback-Leibler Divergence), which can handle zero probability problems and is sensitive to small changes in probability distribution. It also has the advantages of robustness, clear range of results, and easy interpretation. The topic similarity based on JS divergence is expressed as follows: ; in, Indicates the topic similarity between the current requirement document data a and the historical requirement document data b. Represents the topic distribution of the current demand document data a The topic distribution of historical demand document data b The JS divergence can be further expressed as: ; in, and Respectively represent the current demand document data a and the historical demand document data b topics, KL(·) represents KL divergence calculation; m represents the topic distribution of the current demand document data a The topic distribution of historical demand document data b The mean distribution of is: ; From the calculated similarity results, select the one that is most similar to the current requirement document data a Historical requirements document data , forming a set : ; in, represents the set of all candidate items that have overlapping scores with the current requirement document data a. Indicates the pre-set similarity threshold. Different similarity thresholds will be set for different task conditions. .

[0030] S4: Based on the topic similarity obtained in step S3, predict the person-current demand score data.

[0031] In step S4, based on the topic similarity and the person-historical demand score data, the collaborative filtering algorithm is used to predict the person-current demand score data: ; in, Indicates the predicted completion status of personnel u on the current demand document data a - the current demand score, The person-historical requirement score that represents the completion of the historical requirement document data b by person u.

[0032] S5: Combine the personnel-current demand score data obtained in step S4 and the personnel-historical demand score data in step S1 to obtain the comprehensive score of the personnel, and generate a task allocation list based on the comprehensive score.

[0033] In step S5, in view of the differences in historical demand performance of personnel, the different adaptability of each personnel to different subject requirements, and their current workload conditions, in order to maximize work efficiency and quality of results, the above evaluation indicators are comprehensively considered and a comprehensive evaluation method is adopted. The purpose is to accurately quantify the degree of matching of each person to the new requirements, and then reasonably allocate the new demand tasks to the most suitable candidate personnel, so as to achieve optimal allocation of resources and improve work efficiency. The comprehensive score is calculated by the following formula: ; in, It represents the comprehensive score of personnel u on the completion of the current requirement document data a. The personnel-historical requirement score that represents the completion of the historical requirement document data b by personnel u, Indicates personnel-historical demand score The indicator setting weights, Indicates the predicted completion status of personnel u on the current demand document data a - the current demand score, Indicates personnel-current demand score The indicator setting weights, Indicates the time that person u participated in the development of the current requirements document The normalized coefficient of Represents the normalization coefficient Set the weights of the indicators.

[0034] The process of generating a task assignment list according to the comprehensive score in step S5 includes: Generate a list of personnel by comprehensive score; Based on the estimated workload, obtain the specified number of personnel in the personnel list, and the specified number of personnel meets the following requirements:

[0035] Where C represents the number of designated personnel, Indicates the time when the cth designated person participated in the development of the current requirement document data, Indicates the deadline for the requirement. Indicates workload; Assign the current new demand task to the C personnel with the highest scores in the personnel list, and their development time satisfies the above formula, that is, they can cover the workload requirements of the new demand task, and obtain the final task assignment list, thereby ensuring that the personnel in the current candidate list are suitable for handling the current demand and can complete the workload involved in the current demand on time. It should be noted that the task assignment list contains the new demand task and the list of C selected personnel, that is, after selecting the C personnel who participated in the development of the current demand document data, the task assignment list can be obtained.

[0036] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the disclosure of the present invention can be performed in parallel, sequentially or in different orders, as long as the desired results of the technical solution disclosed in the present invention can be achieved, and this document does not limit this.

[0037] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A task allocation method based on current demand and historical demand, characterized in that: include: S1: Collect current requirement document data, historical requirement document data, and personnel-historical requirement score data; S2: Calculate the topic distribution in the current demand document data and the historical demand document data in step S1 respectively; S3: Calculate the topic similarity between the current demand document data and the historical demand document data in step S1 according to the two topic distributions obtained in step S2; S4: Based on the topic similarity obtained in step S3, predict the personnel-current demand score data; S5: Combine the personnel-current demand score data obtained in step S4 and the personnel-historical demand score data in step S1 to obtain the comprehensive score of the personnel, and generate a task allocation list based on the comprehensive score.

2. The task allocation method based on current demand and historical demand according to claim 1 is characterized in that: Step S1 also includes: preprocessing the current demand document data and the historical demand document data; wherein the preprocessing includes: text cleaning, noise removal, word segmentation and stop word removal of the current demand document data and the historical demand document data.

3. The task allocation method based on current demand and historical demand according to claim 1 is characterized in that: In step S2, the two topic distributions are calculated respectively using the LDA model.

4. The task allocation method based on current demand and historical demand according to claim 3 is characterized in that: In step S2, the LDA model is also trained, including: taking the requirement document set and a predefined number of topics as input, and taking the topic distribution of each requirement document and the vocabulary distribution of each topic as output, to train the LDA model; during the training process, Gibbs sampling is used to update the topic allocation of each word.

5. The task allocation method based on current demand and historical demand according to claim 1 is characterized in that: In step S3, the JS divergence of the two topic distributions obtained in step S2 is calculated; and the topic similarity is obtained based on the JS divergence.

6. The task allocation method based on current demand and historical demand according to claim 1 is characterized in that: In step S4, based on the topic similarity and the person-historical demand score data, a collaborative filtering algorithm is used to predict the person-current demand score data: ； in, Indicates the predicted completion status of personnel u on the current demand document data a - the current demand score, The personnel-historical requirement score that represents the completion of the historical requirement document data b by personnel u, Indicates the topic similarity between the current requirement document data a and the historical requirement document data b. Indicates the highest ranking of topic similarity between the historical requirement document data b and the current requirement document data a in the collection A collection of historical requirements document data.

7. The task allocation method based on current demand and historical demand according to claim 1 is characterized in that: In step S5, the comprehensive score is calculated by the following formula: ； in, It represents the comprehensive score of personnel u on the completion of the current requirement document data a. The personnel-historical requirement score that represents the completion of the historical requirement document data b by personnel u, Indicates personnel-historical demand score The indicator setting weights, Indicates the predicted completion status of personnel u on the current demand document data a - the current demand score, Indicates personnel-current demand score The indicator setting weights, Indicates the time that person u participated in the development of the current requirements document The normalized coefficient of Represents the normalization coefficient Set the weights of the indicators.

8. The task allocation method based on current demand and historical demand according to claim 1 is characterized in that: The process of generating a task assignment list according to the comprehensive score in step S5 includes: Generate a list of personnel according to the comprehensive scores; Based on the assessed workload, a specified number of personnel is obtained from the personnel list, where the specified number of personnel satisfies: Where C represents the number of designated personnel, Indicates the time when the cth designated person participated in the development of the current requirement document data, represents the required deadline, and W represents the workload; The personnel list is matched with the designated number of personnel to obtain the task allocation list.

Images