CN113191023B - Crowd-sourcing-aware task allocation and user recruitment model cross-validation method and system - Google Patents
Crowd-sourcing-aware task allocation and user recruitment model cross-validation method and system Download PDFInfo
- Publication number
- CN113191023B CN113191023B CN202110593667.7A CN202110593667A CN113191023B CN 113191023 B CN113191023 B CN 113191023B CN 202110593667 A CN202110593667 A CN 202110593667A CN 113191023 B CN113191023 B CN 113191023B
- Authority
- CN
- China
- Prior art keywords
- task
- model
- user
- agent
- crowd
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06311—Scheduling, planning or task assignment for a person or group
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
Abstract
The invention discloses a cross validation method and a system of a crowd sensing task allocation and user recruitment model, wherein the method comprises the steps of inputting a plurality of space-time subtasks obtained by dividing a crowd sensing task model according to regions and time by using a task allocation model to be cross validated and parameters of an agent determined by using the user recruitment model; importing the space-time subtask into an artificial social platform which is pre-established with a user intelligent agent model, an artificial social environment model and an intelligent agent interaction model, and starting simulation calculation to generate an experimental sample; and performing data statistics on the generated experimental sample aiming at the performed specified evaluation index to obtain an experimental result containing the specified evaluation index. The method can realize the cross validation of the task allocation model and the user recruitment model, and has the advantages of strong parallel computing capability, good user customization capability, simple and efficient development process, strong practicability and wide application prospect.
Description
Technical Field
The invention relates to a city crowd sensing computing technology, in particular to a crowd sensing task allocation and user recruitment model cross validation method and system.
Background
In crowd-sourcing aware applications, an activity initiator may publish some spatial tasks in the real world, invite a user carrying a mobile smart device to move to a target location specified by the task to perform the task, and request data related to a particular location. The mobility of the user provides unprecedented opportunities for data collection and transmission through mobile devices, such as: in an air quality monitoring task initiated by a city environment management department, a responding participant needs to move to a place specified by the task, and acquires air quality data by using a relevant sensor and uploads the data to a central server through a smart phone. The life cycle of crowd sensing applications is generally divided into four phases: the method comprises the steps of task generation, task distribution, user execution and data aggregation and uploading, wherein the most important problem in research is to design a reasonable task distribution and user recruitment model so as to achieve the aim of a platform side or participants (users).
In crowd sensing applications, there are generally two task allocation modes, namely, user selection mode (word selection) and server allocation mode (server allocation). In the user selection mode, the participant may select a task to be completed by browsing published tasks on the server. For example, there is a task of monitoring city air quality on a bidding platform, which can be performed by any user registered on the platform and upload the collected data via a smartphone. The server distribution model collects some information (e.g., location, preferences) and task requirements (spatio-temporal context, domain) of all registered users and distributes the appropriate tasks directly to the appropriate users. In this mode, the server allocates the crowd sensing task to the optimal set of users according to different strategies, for example, maximizing the quality of the completed task, minimizing the system cost, and the like. Therefore, the server allocation mode is more mainstream in the crowd sensing application, because the user resources can be utilized to the maximum extent, and the task completion quality is improved.
With the intensive research of more scholars, the crowd-sourcing aware task allocation and user recruitment model develops from the allocation considering only a single task to the allocation considering multi-task coupling, from the goal of optimizing only the overall utility to simultaneously optimizing a single task and the overall multi-task, and from the allocation and recruitment only on a dedicated crowd-sourcing aware application to the social network-based participant recruitment and task allocation. The researches break through some original fixed-form assumptions to a certain extent, but when the provided crowd sensing user recruitment and task allocation model is verified, simulation experiments of real data sets can be carried out only on the basis of an induced optimization model, and the influence brought by the problems of diversity among users, temporary exit of the users, unqualified data quality and the like is not really considered. Because the cost for carrying out large-scale real experiments is huge, and the real experiments on all newly proposed crowd sensing user recruitment and task allocation models are repetitive work. Therefore, how to implement the generalized verification of the crowd sensing model and improve the efficiency of the verification of the crowd sensing model becomes a key technical problem to be solved urgently.
Disclosure of Invention
The technical problems to be solved by the invention are as follows: aiming at the problems in the prior art, the invention provides a method and a system for cross-verifying the task allocation model and the user recruitment model through crowd sensing.
In order to solve the technical problems, the invention adopts the technical scheme that:
a crowd-sourcing-aware task allocation and user recruitment model cross-validation method comprises the following steps:
1) inputting a plurality of space-time subtasks obtained by dividing the crowd sensing task model according to regions and time by using a task allocation model to be cross-verified and parameters of an agent of the agent determined by using a user recruitment model;
2) the space-time subtask is led into an artificial social platform which is pre-established with a user agent model, an artificial social environment model and an agent interaction model, the user agent model comprises a plurality of agent agents, the agent interaction model is an interaction model among agents, model parameters of the user agent model are set according to the determined parameters of the agents, and model parameters of the artificial social environment model and the agent interaction model are set;
3) setting interference of random operation by using an artificial society calculation experiment scheme design tool in an artificial society platform, and starting an artificial society parallel calculation engine to run and generate an experiment sample; and performing data statistics on the generated experimental sample aiming at the performed specified evaluation index to obtain an experimental result containing the specified evaluation index.
Optionally, the attributes of the crowd sensing task model in step 1) include type, spatio-temporal background information, domain knowledge, pricing, task execution mode, and task decomposition constraints, where: the types comprise type information provided for a crowd sensing task initiator when the task is issued, and comprise a type considering time emergency degree requirement, a type considering space range requirement and a type considering space position relation at different times; the space-time background information is time and space information provided by a crowd sensing task initiator when the task is issued; the domain knowledge is task domain information provided by a crowd sensing task initiator when the task is issued; the pricing is task price information provided by a crowd sensing task initiator when the task is issued; the task execution mode is a task execution mode provided by a crowd sensing task initiator when the task is issued, and comprises user independent completion, user cooperation completion and completion through a shared service; the task decomposition constraint is a task decomposition constraint condition provided for the crowd sensing task initiator when the task initiator issues the task, and comprises a space constraint and a time constraint.
Optionally, the types of requirements regarding time urgency include urgent tasks and general tasks; the types considering the space range requirement comprise a point task, a region task and a complex task, wherein the complex task is the combination of the point task and the region task; the types of the position relationship of the considered space at different time comprise a static task and a dynamic task, wherein the static task refers to a task with a fixed space position at different time, and the dynamic task refers to a task with a changed space position at different time.
Optionally, the attributes of the agent in step 1) include: the system comprises the following components of space-time information, credibility, skills, preference, population identification, gender, age and part or all of social roles, wherein the space-time information comprises space information and time information, the space information comprises the current position, historical track and predicted motion track and interest points of a user, and the time information comprises the working time and available participation time of the user; the credibility comprises the probability of correctly completing the task by the user, and the probability of correctly completing the task by the user is calculated according to the correctly marked proportion in the historical data of different task types completed by the user; the skills comprise knowledge skill values of the user in a plurality of specified skill areas; the preferences include a user's preference for task type, preference for location, and preference for time.
Optionally, the artificial social environment model in step 2) includes an agent and an environment entity, the environment entity includes a community, a factory, a school and a company, the agent forms a family relationship, a neighbor relationship and an environment entity related interpersonal relationship of the agent with a family as a unit and with the environment entity and the family as a hub, and the environment entity related interpersonal relationship includes a friend relationship in the school or a fellow of a workplace.
Optionally, step 2) is preceded by the step of establishing an artificial social environment model: inputting the demographic data of a selected area, generating population identification of an agent, and determining the position of environment entities, wherein the environment entities comprise schools, workplaces, consumption places and entertainment places; setting an administrative region of an agent to generate a residential building; assigning the agent to a family, generating a family structure, determining the family role of the agent, and the age and gender attributes of the agent; distributing addresses for families in residential buildings, and generating a family relation and a neighbor relation of an intelligent agent; according to the age and gender attributes of the intelligent agent, corresponding consumption places and entertainment places are allocated to the intelligent agent, corresponding schools or working places are allocated to the intelligent agent, friend relations in the schools or friend relations in the working places are established for the intelligent agent, and finally a complete artificial social environment model is obtained.
Optionally, the agent interaction model in step 2) is an agent space mobile network G formed by connecting agent agents and space position nodes through edgesI(VI,EI,WI) Contact network G with intelligent agentm(VI,Vm,Em,Wm) Formed bimodal network G2(GI,Gm) In which V isI,VmRepresenting the set of agent individuals and geospatial locations of the agent, respectively, EI,EmSets of contiguous edges, W, representing a personal contact network and an individual space mobile network, respectivelyI，WmRespectively represent EI,EmCorresponding set of weight coefficients, modal network G2(GI,Gm) The connection edges only exist between the agents and the spatial position, and the movement of the agents is subject to the spatial mobile network GI(VI,EI,WI) Is connected with the edge weight coefficient WIThe agent selects the contact object to be contacted by the agent contact network Gm(VI,Vm,Em,Wm) Is connected with the edge weight coefficient WmAnd (4) controlling.
Optionally, the step 3) is followed by a step of comparing the experimental result containing the specified evaluation index with the input detection result of the conventional optimization model, and a step of performing a multi-disc display on part or all of the experimental samples in the experimental result containing the specified evaluation index by using an artificial social state visualization tool in the artificial social platform.
In addition, the invention also provides a crowd sensing task allocation and user recruitment model cross validation device, which comprises a microprocessor and a memory which are connected with each other, wherein the microprocessor is programmed or configured to execute the steps of the crowd sensing task allocation and user recruitment model cross validation method.
Additionally, the present invention also provides a computer readable storage medium having stored therein a computer program programmed or configured to perform the crowd-sourced task allocation and user recruitment model cross-validation method.
Compared with the prior art, the invention mainly has the following advantages:
1. the invention utilizes the thought of artificial society, establishes user intelligent agent models of various intelligent agent agents of participants in the artificial society based on a data-driven mode of modeling of various intelligent agent agents, and carries out large-scale simulation experiments of a task allocation model and a user recruitment model, which is a brand-new solution idea.
2. The invention adopts a data-driven user agent model mode of various agent agents, constructs a user agent model, an artificial social environment model and an agent interaction model of an artificial city, configures relevant parameters of the model based on actual data and experimental requirements, and can develop large-scale cross validation experiments of a crowd sensing user recruitment and task allocation model; the invention can carry out experimental verification on any user recruitment and task allocation model of crowd sensing application, and the data-driven mode is more real, thereby increasing the effectiveness of the invention in actual scenes.
3. The method is suitable for verifying the crowd sensing model with various task types, various constraints on participants and various optimization targets, and has the advantage of good universality.
Drawings
FIG. 1 is a schematic diagram of a basic flow of a method according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating the decomposition of the attributes and tasks of the crowd sensing task model according to an embodiment of the present invention.
Fig. 3 is a flow chart of user recruitment and task assignment in an embodiment of the invention.
FIG. 4 is a diagram of an agent model according to an embodiment of the invention.
FIG. 5 is a flowchart illustrating a process of generating a model of a demographic environment in an embodiment of the present invention.
FIG. 6 is a flowchart of performing a large-scale cross-validation experiment in an embodiment of the present invention.
Detailed Description
In order to better describe the flow and the action of the present invention, the crowd sensing task in a specific scene of air quality monitoring will be taken as an example, and the crowd sensing task allocation and user recruitment model cross-validation method and system of the present invention will be further described in detail. However, it should be noted that the method and system for cross-validation of crowd sensing task allocation and user recruitment model according to the present invention do not depend on a specific crowd sensing task, but can implement crowd sensing tasks of various task types, various constraints on participants, and various optimization objectives, which are not illustrated herein.
Referring to fig. 1, the crowd-sourcing aware task allocation and user recruitment model cross-validation method of the present embodiment includes:
1) inputting a plurality of space-time subtasks obtained by dividing the crowd sensing task model according to regions and time by using a task allocation model to be cross-verified and parameters of an agent of the agent determined by using a user recruitment model;
2) the method comprises the steps that a space-time subtask is led into an artificial social platform which is pre-established with a user agent model, an artificial social environment model and an agent interaction model, the user agent model comprises a plurality of agent agents (agents), the agent interaction model is an interaction model among agents, model parameters of the user agent model are set according to the determined parameters of the agents, and model parameters of the artificial social environment model and the agent interaction model are set;
3) setting interference of random operation by using an artificial society calculation experiment scheme design tool in an artificial society platform, and starting an artificial society parallel calculation engine to run and generate an experiment sample; and performing data statistics on the generated experimental sample aiming at the performed specified evaluation indexes to obtain an experimental result containing the specified evaluation indexes.
In the real world, the mobile crowd-sourcing perception task is a series of physical processes that can be cooperatively performed by multiple independent users. In general, there are several important aspects for modeling tasks in crowd sensing activities, which are described in detail below. As shown in particular in fig. 2. Referring to fig. 2, the attributes of the crowd sensing task model in step 1) of this embodiment include types, spatio-temporal background information, domain knowledge, pricing, task execution modes, and task decomposition constraints, where:
the types comprise type information provided for a crowd sensing task initiator when the task is issued, and comprise a type considering time emergency degree requirement, a type considering space range requirement and a type considering space position relation at different times; in the embodiment, different task types of crowd sensing are defined for the spatio-temporal information in the crowd sensing activity. First, according to the time information, there are two task types: emergency tasks or general tasks. For an emergency task, a user is required to complete the task as soon as possible within a limited time to acquire timely and useful information, such as collecting traffic dynamic information, monitoring a drainage state and the like. Unlike the timeliness requirements of emergency tasks for users, generic tasks do not need to be completed immediately. For example, a user may collect utility information over a period of time (such as a week). Then, the crowd sensing task can be divided into three categories according to the spatial information: point tasks, area tasks, and complex tasks. Point tasks, such as observing the flow of people at an intersection, are site specific. Thus, the user must visit a particular location to complete the task. Regional tasks require that the task be completed in a region, such as collecting the air quality of a street. The complex task also comprises a plurality of different subtasks, namely, one task can be divided into a plurality of subtasks according to different requirements, and different users are assigned to complete different subtasks. Finally, the crowd sensing task can be divided into a static task and a dynamic task considering temporal and spatial information. Static tasks consist of spatial information with fixed positions, such as air quality monitoring, noise information mapping, etc. The spatial information of the dynamic task includes an uncertain position sequence corresponding to the time information, such as mobile car sharing service, parcel delivery, suspicious vehicle tracking, and the like.
The space-time background information is time and space information provided by a crowd sensing task initiator when the task is issued; it is necessary to define spatiotemporal context information for crowd sensing. Generally, the originator of the activity must provide spatiotemporal information such as road names in traffic monitoring, restaurant locations in take-away, etc. when issuing the task. For spatial information, point locations and regional locations are often used to describe tasks such as recording the number of passengers at a bus stop, measuring the air quality in a large park, etc. Two important times for the time information that the user needs to pay attention to are the start time and the end time. For example, a user may need to collect information for a concert (e.g., 18:00-20:00) during a performance.
The domain knowledge is task domain information provided by a crowd sensing task initiator when the task is issued; in this embodiment, the crowd sensing task may involve different knowledge domains as any work/task. Thus, with D ═ D1,...,dmRepresents the required knowledge domain. For example, organizing a common activity (task t) by different users may require a lot of domain knowledge, such as activity planning (d)1) Advertisement promotion (d)2) Personal arrangement (d)3) And the like.
Pricing is task price information provided by a crowd sensing task initiator when a task is issued; since the user gains revenue by completing the task, the price of the task plays an important role in crowd sensing activities, which affects the aggressiveness of user participation. Generally, the originator of the activity has the right to price the task, which relates to the size of the task, the difficulty of the user performing the task, and so on.
The task execution mode is a task execution mode provided by the crowd sensing task initiator when the task is issued, and comprises user independent completion, user cooperation completion and completion through a shared service; the crowd sensing task should explain how users are required to complete the task, for example, some tasks are completed by different users independently; some tasks need to be completed by cooperation of several users; some tasks are accomplished through shared services.
And the task decomposition constraint is a task decomposition constraint condition provided for the crowd sensing task initiator when the task initiator issues the task, and comprises a space constraint and a time constraint. Often a complex task requires numerous users to perform sensing tasks at different locations, such as sensing the air quality of every street in a city. The perceptual task is reasonably divided into space and time (space constraint and time constraint) according to the needs of the activity initiator. Taking an urban air quality monitoring task as an example, spatial constraint refers to dividing a sensing area into 2 km-2 km grids, and a user executes a sensing task in each different grid; the time constraint means that a user needs to collect data once every hour and upload the data to the central server within a specified time.
As an optional implementation manner, the types of requirements considering the time urgency level in this embodiment include an urgent task and a general task; considering the types of space range requirements, including point tasks, area tasks and complex tasks, wherein the complex tasks are the combination of the point tasks and the area tasks; the types of the position relationship of the space at different times include a static task and a dynamic task, the static task refers to a task whose space position is fixed at different times, and the dynamic task refers to a task whose space position changes at different times. After the purpose and the type of the crowd sensing task are determined, the crowd sensing task can be classified, such as an emergency task or a general task, if data of air quality in a city, real-time road condition data and the like are collected; point tasks, area tasks, or complex tasks. In general, crowd-sourcing aware tasks are complex tasks that need to be broken down into multiple subtasks. According to the data granularity requirement and the time urgency requirement of an activity initiator, a task is divided into a plurality of space-time subtasks according to regions and time, and meanwhile, equipment, skills, domain knowledge and the like required by a user to complete the subtasks are determined.
The task allocation model and the user recruitment model are cross-validation targets of the method of the embodiment. According to the requirements of the activity initiator, the user recruitment and task allocation modes can be determined, and a corresponding task allocation model and a corresponding user recruitment model are established. Firstly, determining the total budget of the perception task, and adopting a user selection mode or a server distribution mode or a mixed mode, determining the total target and constraint of the perception task, such as the expenditure of the optimization total task, the completion quality of the optimization total task or the total reward obtained by the optimization user, and the like, and establishing a corresponding single-target or multi-target optimization model. Taking an urban air quality monitoring task as an example, a user recruitment and task allocation model of a crowd sensing task is established based on a server allocation mode. Assuming that the crowd sensing task is decomposed into m sub-regions and n sensing periods, a time-space sub-task set is formed:
S×C＝{(si,cj)|si∈S,cj∈C}
in the above formula, S represents a set of spatial subregions, C represents a set of sensing periods, and SiDenotes the i-th element in S, cjRepresenting the jth element in C. The set of mobile users on the social network is U ═ U1,u2,...,ukIn which u1～ukRespectively representing the 1 st to k th users in the mobile user set U. Selecting a part of registered users of social network as a seed user
Based on the existing community propagation model IC or LT model,
is the affected user (the recruited user) and further, Covered (f (U ')) represents the set of spatio-temporal tasks that the selected set of users f (U') can cover, then the optimization problem can be described as finding a certain set of seeds
Maximizing the following goal (maximizing the number of tasks completed in the spatio-temporal task set):
Subject to|U′|≤p and f(U′)≤q
in the above equation, maximise represents maximizing the objective function, p is the maximum number of seeded users, and q is the maximum number of recruited users. Referring to fig. 3, the specific steps are based on the seed user set
As input, calculating the probability of each user becoming the final recruited participant by using a propagation method with maximized influence, and forming an acceptance probability vector by using the results of thousands of Monte Carlo simulations; and then, establishing a user mobility prediction model based on the mobile information data set of the user, and predicting the coverage of the task. And finally, calculating the probability of each time-space subtask being Covered by the participation of the user, and further calculating the total coverage rate Covered (U') of the task as output.
In order to describe a user agent model, a multi-element model is required to be established to describe various attributes of the agent, such as: a skill level; interest; sex; age; reputation value, etc. Referring to fig. 4, the attributes of the agent in step 1) of the present embodiment include: some or all of the spatiotemporal information, reputation, skills, preferences, demographic identification, gender, age, and social roles,
the space-time information comprises space information and time information, the space information comprises the current position, the historical track, the predicted motion track and the interest point of the user, and the time information comprises the working time and the participatable time of the user; user temporal and spatial information refers to temporal and spatial information used to select appropriate participants to efficiently complete a task. Firstly, the available time of the user needs to be confirmed, for example, white-collar workers can only participate in the crowd sensing task in the off-duty time; for the spatial information of the user, some positioning technologies are used to identify the current location, including indoor and outdoor positioning technologies. The spatial information of the user includes not only the current position thereof but also the movement trajectory of the user. By researching the historical movement pattern of the user, more information can be mined and obtained, so that effective task distribution can be carried out, such as the path of future movement, interest points and the like.
The credibility comprises the probability of the user for correctly completing the task, and the probability of the user for correctly completing the task is calculated according to the correctly marked proportion in the historical data of different task types completed by the user; the reputation value of a user reflects the likelihood that the user will properly complete a task. In general, a reputation value for a user may be calculated based on historical data for completing different task types. For example, user u1The task of marking 10 images was completed, 8 of which were correctly marked. Therefore, we can conclude that user u1There is an 80% probability of correctly performing the label image task, with a reputation value of 0.8.
Skills include the user's knowledge skill value in a variety of specified skill areas; a user may possess multiple skills. The user's skill corresponds to the knowledge in a particular area of skill, and a continuous scale (e.g., scale [0,1]) may be used to represent the user's level of expertise in a particular area. For example, a skill value of 0 reflects that the user has no expertise in the corresponding domain. Note that only users with skill levels not below the minimum knowledge requirements for the task have an opportunity to complete the task.
Preferences include user preferences for task type, location, and time. In crowd-sourcing tasks, a user's preference for completing a task is generally expressed in three areas: a preference for task type, a preference for location, and a preference for time. Some users are willing to do normal work without extra mobile burden on task types, but some users tend to complete urgent tasks to get more incentives. With regard to the preference of place and time, different people have different favorite places at the corresponding times. For example, younger people may prefer to obtain information in a mall at night, while older people may prefer to perform tasks in a park in the morning.
In the embodiment, the factors such as diversity among users, temporary exit of the users, and ineligibility of data quality collected by the users to conditions are considered, and the users which are possibly recruited in an actual scene are described by establishing the agent of the agent in the artificial society. Firstly, the spatiotemporal attributes of the user refer to the time and the area in which the user can participate in the perception task; then the skill attributes of the user, specifically referring to the user's familiarity with operating certain areas of sensory tasks and the proficiency level indicating his/her skill level (below a certain threshold indicates that a sensory task may fail); the interest and preference of the user follow, and the user has different preference and tendency to different task types; again, the historical trustworthiness of the user, the accumulated reputation value from which past tasks were completed. If the user always submits valid data at the appointed time and place, the corresponding reputation value is high; finally, each participant has a certain probability of exiting execution of the subtask due to a personal transaction. Other attributes such as demographic identity, gender, age, social role, etc. In the present embodiment, such an agent is described in the form of an agent tuple, and the actual attribute values of each agent are largely different.
The artificial social environment model in this embodiment refers to a model that is modeled by a demographic environment, combines a family structure with demographic data (macro data) and maps the data to individual microscopic attributes, and ensures logical correctness inherent in a population. The agent interaction model refers to a model which can describe the contact and interaction behaviors generated by agent agents of agents in the same geographic space as the simulation time advances. In this embodiment, the artificial social environment model in step 2) includes an agent and an environment entity, the environment entity includes a community, a factory, a school and a company, the agent forms a family relationship, a neighbor relationship and an environment entity related interpersonal relationship of the agent with a family as a unit and the environment entity and the family as a hub, and the environment entity related interpersonal relationship includes a friend relationship in the school or a fellow-in-work relationship. A certain city is selected, census data and annual data published by the national statistical office of the people's republic of China are utilized, family roles played by intelligent agents of each user are described through a family structure design algorithm based on family households, family generations and family roles, and mapping from macroscopic data to microscopic individual attributes is completed. The artificial social environment model consists of 2 parts of geographic environment and population generation. The generation of the geographic environment is carried out according to a hierarchical geographic environment model and a geographic space discretization method, and the population generation is completed by two processes of family structure generation and population attribute matching. Furthermore, different algorithms may be employed to generate the geographic context entity depending on how rich the context information data is. Specifically, four cases are classified: generating environment entities with spatially distributed location information and population accommodating quantities; the method has spatial distribution position information and no environment entity for accommodating population quantity; the environment entity generation method comprises the steps that the environment entity distribution quantity in a space area is generated, but the space distribution position information and the environment entity accommodation population quantity are not generated; and generating environment entities without the distribution quantity of the environment entities in the space region and the space distribution position information. The population geographic matching algorithm establishes a matching relationship between the individual and the environmental entity, including assigning a home address, assigning a school, assigning a work place, and the like.
And generating the geographical environment of the artificial society by a geographical space discretization method according to the hierarchical geographical environment model. According to the richness degree of the acquired geographic information data, different methods can be adopted to generate the geographic environment entity. And then, establishing a matching relationship between the individual and the environment entity by utilizing a population and geographic environment matching algorithm, wherein the matching relationship comprises the steps of allocating a work address, allocating a school, allocating a family and the like. In addition, according to different social characteristics of the agent of. Referring to fig. 5, step 2) of the present embodiment includes the steps of establishing an artificial social environment model: inputting the demographic data of a selected area, generating population identification of the intelligent agent, and determining the position of environment entities, wherein the environment entities comprise schools, workplaces, consumption places and entertainment places; setting an administrative region of an agent to generate a residential building; assigning the agent to a family, generating a family structure, determining the family role of the agent, and the age and gender attributes of the agent; distributing addresses for families in residential buildings, and generating a family relation and a neighbor relation of an intelligent agent; according to the age and gender attributes of the intelligent agent, corresponding consumption places and entertainment places are allocated to the intelligent agent, corresponding schools or working places are allocated to the intelligent agent, friend relations in the schools or friend relations in the working places are established for the intelligent agent, and finally a complete artificial social environment model is obtained. The selected demographic data of a certain area can be the demographic data of a selected city, the macroscopic data is mapped to the attributes of the microscopic intelligent agents, the internal logic of the population is ensured to be correct, and meanwhile, a corresponding demographic environment model is generated. In addition, some other attributes can determine a rough value range through a mode of investigation or empirical collection, and different intelligent agent parameters are given through a random generation mode. In addition, different parameter values are designed according to experimental requirements to observe the influence of different parameters on results.
In this embodiment, the agent interaction model in step 2) is an agent space mobile network G formed by connecting agent agents and space position nodes through edgesI(VI,EI,WI) Contact network G with intelligent agentm(VI,Vm,Em,Wm) Formed bimodal network G2(GI,Gm) In which V isI,VmRepresenting sets of agent individuals and geospatial locations of the agent, respectively, EI,EmSets of edges, W, representing personal contact networks and individual space mobile networks, respectivelyI，WmRespectively represent EI,EmCorresponding sets of weight coefficients, bimodal network G2(GI,Gm) The connecting edges in (3) only exist between the agents and the spatial positions, and the movement of the agents is influenced by the spatial mobile network GI(VI,EI,WI) Is connected with the edge weight coefficient WIThe agent selects the contact object to be contacted by the agent contact network Gm(VI,Vm,Em,Wm) Is connected with the edge weight coefficient WmAnd (4) controlling. The spatial movement laws of agent agents in the artificial society can be described in the form of activity logs.
In the embodiment, in the step 2), when the model parameters of the user agent model are set according to the determined parameters of the agent, and the model parameters of the artificial social environment model and the agent interaction model are set, the set relevant parameters are such as the spatiotemporal background information (starting time), the domain knowledge requirement level and the like in the crowd sensing task model; skill level, spatio-temporal location, preferences, reputation values, etc. in the user agent model; the position of the interest point in the artificial social environment model, the position of the agent, etc.; contact probabilities of agent agents in the agent interaction model, and the like. The specific process is as follows: 1) firstly, selecting demographic data of a certain city, mapping macroscopic data to attributes of a microscopic intelligent agent, ensuring correct internal logic of population, and generating a corresponding population geographic environment model; s2) initializing the configuration file. In this embodiment, the configuration files include a set of situation symbols, a set of custom entity classes, a set of custom entities, a set of complex data types, a set of data displays, and a set of data package configurations. The entity is an object which exists in a real world in a guest and can be distinguished from each other, is a main body for executing the simulation behavior, and the customized entity is an entity related to situation display in the simulation process, such as an intelligent entity and an environmental entity; s3) determining a rough value range of other attributes through a mode of investigation or empirical collection, and endowing different intelligent agent parameters through a random generation mode; s4), designing different parameter values according to experiment requirements, and observing the influence of different parameters on results.
The basis for developing the large-scale cross validation experiment in the step 3) of the embodiment is an artificial social Parallel computing Engine, which is a Parallel computing Engine based on Message paging Interface, supports multi-process distributed Parallel simulation and adopts a conservative discrete event synchronization algorithm. The specific flow of experimental development is shown in fig. 6, and is detailed as follows: s1) based on the configured model parameters, further carrying out experiment setting and some random operation interference according to the requirement of an activity initiator by utilizing an artificial social computing experiment scheme design tool, and starting a simulation engine to generate a large number of experiment samples; s2) based on the samples, carrying out statistics of some key data on the data types defined by the data display centralized diagram of the configuration file by utilizing an artificial social operation support tool, such as the overall income of a user, the overall expenditure of a platform side, the completion quality of tasks and the like; s3) the profits of the platform side and the user side and the task completion quality are used as evaluation indexes, and the results of the simple optimization model and the results of the large-scale calculation experiment of the embodiment are compared to judge the advantages and disadvantages of the crowd sensing user recruitment and task allocation model. In addition, referring to fig. 1 and fig. 6, the step 3) of the present embodiment further includes a step of comparing the experimental result containing the specified evaluation index with the input detection result of the conventional optimization model, and a step of performing a review display on part or all of the experimental samples in the experimental result containing the specified evaluation index by using an artificial social status visualization tool in the artificial social platform. After configured models and parameters are set for all models including a user intelligent agent model, an artificial social environment model and an intelligent agent interaction model, an artificial social computing experiment scheme design tool is utilized to further carry out experiment setting according to the requirements of an activity initiator and carry out interference of random operations, and a simulation engine is started to generate a large number of experiment samples. Then, based on the samples, the artificial social operation support tool is used for carrying out statistics on some key data, such as the overall income of a user, the overall expenditure of a platform side, the completion quality of tasks and the like, judging the advantages and disadvantages of a crowd sensing user recruitment and task allocation model, and meanwhile, the artificial social state visualization tool can be used for carrying out repeated display on a certain experimental sample.
To sum up, the cross-validation method for the crowd sensing task allocation and user recruitment model in the embodiment includes defining the crowd sensing task and decomposing the task; defining and constructing a specific user recruitment model; constructing an agent (virtual user) model and determining agent parameter attributes; constructing an artificial social environment model and an interaction model; configuring relevant parameters of the model according to actual data and experimental requirements; the method has the advantages of strong parallel computing capability, good user customization capability, simple and efficient development process, strong practicability, wide application prospect and the like. The key means for realizing the method of the embodiment comprises crowd sensing task modeling, agent modeling in artificial society, population geographic environment modeling and artificial society experiment operation supporting tools, and can realize cross validation and evaluation of any crowd sensing task allocation and user recruitment model. Compared with the conventional method, the method has the advantages that various attributes and factors of the user are considered, the method is closer to the real situation, and the experimental development calculation efficiency based on the parallel experimental engine is higher, so that the method has higher practical value. The embodiment method applies a data-driven agent modeling mode for the first time, considers the diversity problems of users, tasks, equipment and the like, establishes a user agent model of participants in the artificial society, provides a general cross validation method of a crowd sensing task allocation model, can be used for simulating the task allocation and user recruitment processes in crowd sensing application, and can validate the rationality of the crowd sensing model by developing large-scale simulation experiments on a computing platform. Compared with the traditional planning model solving mode, the method considers a plurality of factors (aspects such as users and tasks) closer to reality, and reduces the cost required by a real experiment in a large-scale simulation mode.
In addition, the present embodiment also provides a crowd sensing task assignment and user recruitment model cross-validation apparatus, comprising a microprocessor and a memory connected to each other, wherein the microprocessor is programmed or configured to perform the steps of the aforementioned crowd sensing task assignment and user recruitment model cross-validation method.
Further, the present embodiments also provide a computer readable storage medium having stored therein a computer program programmed or configured to perform the aforementioned crowd-sourced aware task assignment and user recruitment model cross-validation methods.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein. The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks. These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the invention may occur to those skilled in the art without departing from the principle of the invention, and are considered to be within the scope of the invention.
Claims (9)
1. A crowd-sourcing-aware task allocation and user recruitment model cross-validation method is characterized by comprising the following steps:
1) inputting a plurality of space-time subtasks obtained by dividing the crowd sensing task model according to regions and time by using a task allocation model to be cross-verified and parameters of an agent of the agent determined by using a user recruitment model;
2) the space-time subtask is led into an artificial social platform which is pre-established with a user agent model, an artificial social environment model and an agent interaction model, the user agent model comprises a plurality of agent agents, the agent interaction model is an interaction model among agents, model parameters of the user agent model are set according to the determined parameters of the agents, and model parameters of the artificial social environment model and the agent interaction model are set;
3) setting interference of random operation by using an artificial society calculation experiment scheme design tool in an artificial society platform, and starting an artificial society parallel calculation engine to run and generate an experiment sample; performing data statistics on the generated experimental samples aiming at the performed specified evaluation indexes to obtain an experimental result containing the specified evaluation indexes;
attributes of the crowd sensing task model in the step 1) comprise types, space-time background information, domain knowledge, pricing, task execution modes and task decomposition constraints, wherein: the types comprise type information provided for a crowd sensing task initiator when the task is issued, and comprise a type considering time emergency degree requirement, a type considering space range requirement and a type considering space position relation at different times; the space-time background information is time and space information provided by a crowd sensing task initiator when the task is issued; the domain knowledge is task domain information provided by a crowd sensing task initiator when the task is issued; the pricing is task price information provided by a crowd sensing task initiator when the task is issued; the task executing mode is a task executing mode provided by a crowd sensing task initiator when the task is issued, and comprises user independent completion, user cooperation completion and completion through a shared service; the task decomposition constraint is a task decomposition constraint condition provided for the crowd sensing task initiator when the task initiator issues the task, and comprises a space constraint and a time constraint.
2. The crowd-sourcing aware task allocation and user recruitment model cross validation method of claim 1, wherein the types of consideration temporal urgency requirements include urgent tasks and general tasks; the types considering the space range requirement comprise a point task, a region task and a complex task, wherein the complex task is the combination of the point task and the region task; the types of the position relationship of the considered space at different time comprise a static task and a dynamic task, wherein the static task refers to a task with a fixed space position at different time, and the dynamic task refers to a task with a changed space position at different time.
3. The crowd-sourcing aware task allocation and user recruitment model cross-validation method of claim 1, wherein the attributes of agent agents in step 1) comprise: the system comprises the following steps of partial or all of spatiotemporal information, credibility, skills, preference, population identification, gender, age and social roles, wherein the spatiotemporal information comprises spatial information and time information, the spatial information comprises the current position, historical track and predicted motion track and interest points of a user, and the time information comprises the working time and participatable time of the user; the credibility comprises the probability of correctly completing the task by the user, and the probability of correctly completing the task by the user is calculated according to the correctly marked proportion in the historical data of different task types completed by the user; the skills comprise knowledge skill values of the user in a plurality of specified skill fields; the preferences include a user's preference for task type, preference for location, and preference for time.
4. The crowd-sourcing-aware task allocation and user recruitment model cross-validation method as claimed in claim 3, wherein the artificial social environment model in step 2) comprises an agent and an environment entity, the environment entity comprises a community, a factory, a school and a company, the agent forms a family relationship, a neighbor relationship and an environment entity related personal relationship of the agent in a family unit and in an environment entity and a family hub, and the environment entity related personal relationship comprises a friend relationship in a school or a friend relationship in a workplace.
5. The crowd-sourcing aware task allocation and user recruitment model cross-validation method of claim 4, wherein step 2) is preceded by the step of building an artificial social environment model: inputting the demographic data of a selected area, generating population identification of an agent, and determining the position of environment entities, wherein the environment entities comprise schools, workplaces, consumption places and entertainment places; setting an administrative region of an intelligent agent to generate a residential building; distributing the intelligent agent to a family, generating a family structure, and determining the family role of the intelligent agent and the age and gender attributes of the intelligent agent; distributing addresses for families in residential buildings, and generating a family relation and a neighbor relation of an intelligent agent; according to the age and gender attributes of the intelligent agent, corresponding consumption places and entertainment places are allocated to the intelligent agent, corresponding schools or working places are allocated to the intelligent agent, friend relations in the schools or friend relations in the working places are established for the intelligent agent, and finally a complete artificial social environment model is obtained.
6. The crowd-sourcing-aware task allocation and user recruitment model cross-validation method as claimed in claim 1, wherein the agent interaction model in step 2) is an agent-based spatial mobile network formed by connecting agent agents and spatial nodes via edgesG I (V I ,E I ,W I ) Contact network with intelligent agentG m (V I ,V m ,E m ,W m ) Formed bimodal networkG 2(G I , G m ) WhereinV I ,V m Respectively representing a set of agent individuals and geospatial locations,E I ,E m respectively representing a set of contiguous edges of a personal contact network and an individual spatial mobile network,W I ，W m respectively representE I ,E m Corresponding sets of weight coefficients, a bimodal networkG 2(G I , G m ) The connection edges only exist between the agents and the spatial location, and the movement of the agents is subject to the spatial mobile networkG I (V I ,E I ,W I ) Is connected with the edge weight coefficientW I The agent selects the contact object to be contacted by the agent contact networkG m (V I ,V m ,E m ,W m ) Connected edge weight coefficient ofW m And (4) controlling.
7. The crowd sensing task assignment and user recruitment model cross-validation method according to claim 1, further comprising a step of comparing the experimental results containing the specified evaluation indexes with the input detection results of the conventional optimization model after the step 3), and a step of displaying a part or all of the experimental samples containing the specified evaluation indexes by using an artificial social state visualization tool in the artificial social platform.
8. A crowd-sourcing aware task allocation and user recruitment model cross-validation apparatus comprising a microprocessor and a memory interconnected, wherein the microprocessor is programmed or configured to perform the steps of the crowd-sourcing aware task allocation and user recruitment model cross-validation method of any one of claims 1 to 7.
9. A computer-readable storage medium having stored thereon a computer program programmed or configured to perform the crowd-sourced task allocation and user recruitment model cross-validation method of any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110593667.7A CN113191023B (en) | 2021-05-28 | 2021-05-28 | Crowd-sourcing-aware task allocation and user recruitment model cross-validation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110593667.7A CN113191023B (en) | 2021-05-28 | 2021-05-28 | Crowd-sourcing-aware task allocation and user recruitment model cross-validation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113191023A CN113191023A (en) | 2021-07-30 |
| CN113191023B true CN113191023B (en) | 2022-07-05 |
Family
ID=76985934
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110593667.7A Active CN113191023B (en) | 2021-05-28 | 2021-05-28 | Crowd-sourcing-aware task allocation and user recruitment model cross-validation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113191023B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113780955A (en) * | 2021-09-17 | 2021-12-10 | 拉扎斯网络科技（上海）有限公司 | Position checking method and device, electronic equipment and storage medium |
| CN114339655B (en) * | 2022-03-14 | 2022-05-24 | 中国人民解放军国防科技大学 | Sparse mobile crowd-sourcing sensing method with cooperative human-vehicle participation |
| CN114943340A (en) * | 2022-06-03 | 2022-08-26 | 哈尔滨理工大学 | Social relationship reasoning and task allocation method for crowd sensing system |
| CN115002159B (en) * | 2022-06-06 | 2023-10-03 | 哈尔滨理工大学 | Community classification and user selection method for sparse group intelligent perception system |
| CN115099535B (en) * | 2022-08-24 | 2022-12-20 | 东南大学 | Dual-target crowd sensing excitation mechanism method |
| CN117151914B (en) * | 2023-11-01 | 2024-01-30 | 中国人民解放军国防科技大学 | Crowd sensing user selection method and device based on comprehensive influence evaluation |
| CN117474509B (en) * | 2023-12-27 | 2024-04-02 | 烟台大学 | Worker recruitment method and system based on trust evaluation framework and tabu search |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108197889A (en) * | 2017-12-28 | 2018-06-22 | 吉林大学 | A kind of intelligent perception user recruitment method and system based on mobility prediction |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150242798A1 (en) * | 2014-02-26 | 2015-08-27 | Xerox Corporation | Methods and systems for creating a simulator for a crowdsourcing platform |
| CN106529776B (en) * | 2016-10-31 | 2017-11-24 | 合肥工业大学 | The autonomous cotasking distribution method of multiple agent |
| CN107688828B (en) * | 2017-08-23 | 2020-11-03 | 武汉大学 | Bus crowding degree estimation method based on mobile phone sensor |
| US20190244099A1 (en) * | 2018-02-05 | 2019-08-08 | Deepmind Technologies Limited | Continual reinforcement learning with a multi-task agent |
| CN109885944A (en) * | 2019-02-25 | 2019-06-14 | 天津工业大学 | A kind of blending space Mechanism of Task Allocation in mobile intelligent perception |
| CN112381454B (en) * | 2020-12-04 | 2023-10-31 | 中国科学院自动化研究所 | Role-based multi-agent task collaboration system |
-
2021
- 2021-05-28 CN CN202110593667.7A patent/CN113191023B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108197889A (en) * | 2017-12-28 | 2018-06-22 | 吉林大学 | A kind of intelligent perception user recruitment method and system based on mobility prediction |
Non-Patent Citations (2)
| Title |
|---|
| KD-ACP: A Software Framework for Social Computing in Emergency Management;Bin Chen，etc;《Artificial Intelligence and Its Applications 2014》;20151005;1-27 * |
| Mobile crowdsensing: A survey on privacy-preservation, task management, assignment models, and incentives mechanisms;FazlullahKhan，etc;《Future Generation Computer Systems》;20190514;第100卷;456-472 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113191023A (en) | 2021-07-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113191023B (en) | Crowd-sourcing-aware task allocation and user recruitment model cross-validation method and system | |
| Zhao et al. | Spatial crowdsourcing: current state and future directions | |
| Long et al. | Mapping block-level urban areas for all Chinese cities | |
| Guo et al. | ActiveCrowd: A framework for optimized multitask allocation in mobile crowdsensing systems | |
| Auld et al. | Activity planning processes in the Agent-based Dynamic Activity Planning and Travel Scheduling (ADAPTS) model | |
| CN107111794B (en) | Predicting and exploiting variability of travel time in a mapping service | |
| US20200273347A1 (en) | Joint order dispatching and fleet management for online ride-sharing platforms | |
| Malleson et al. | Analysis of crime patterns through the integration of an agent-based model and a population microsimulation | |
| WO2017062684A1 (en) | Systems and methods for dynamically generating patrol schedules based on historic demand event data | |
| Dahal et al. | An agent-integrated irregular automata model of urban land-use dynamics | |
| Ullah et al. | A study of information technology adoption for real-estate management: A system dynamic model | |
| CN115049159B (en) | Population distribution prediction method and device, storage medium and electronic equipment | |
| Fernández et al. | Bike3S: A tool for bike sharing systems simulation | |
| Lihu et al. | A multi-agent model of changes in urban safety livability | |
| CN115829124A (en) | Charging pile address selection method, device, equipment and storage medium | |
| Jiang et al. | Matching model between private idle parking slots and demanders for parking slot sharing | |
| Li et al. | Impact of the pre-simulation process of occupant behaviour modelling for residential energy demand simulations | |
| Gong et al. | Agent-based modelling with geographically weighted calibration for intra-urban activities simulation using taxi GPS trajectories | |
| Nayak et al. | A critical review of activity participation decision: a key component of activity-based travel demand models | |
| Hunter et al. | Where should police forces target their residential burglary reduction efforts? Using official victimisation data to predict burglary incidences at the neighbourhood level | |
| Quan et al. | An optimized task assignment framework based on crowdsourcing knowledge graph and prediction | |
| Li et al. | Modeling of occupant behavior considering spatial variation: geostatistical analysis and application based on American time use survey data | |
| Hackney | Integration of social networks in a large-scale travel behavior microsimulation | |
| Wei | Energy Footprinting and Human-Centric Building Co-Optimization with Multi-Task Deep Reinforcement Learning | |
| Farmer | Commuting flows & local labour markets: Spatial interaction modelling of travel-to-work |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |