Methods and systems for modeling crowdsourcing platform

摘要

Methods and systems for creating statistical model. Values of at least one output parameter corresponding to each combination of values of a set of input parameters are collected. The at least one output parameter and the set of input parameters are associated with one or more crowdsourcing tasks. Each combination of the values of the set of input parameters is determined based on at least one of one or more design of experiments or a historical data. Further, a distribution of the values of the at least one output parameter is determined. The statistical model is created based on the values of the set of input parameters and one or more characteristics associated with the distribution of the values of the at least one output parameter.

主权项

1. A method for creating a performance statistical model for an image-analysis task on a crowdsourcing platform, the method comprising:
collecting, by one or more processors, values of at least one output parameter corresponding to each combination of values of a set of input parameters, wherein the at least one output parameter corresponds to at least one of a task accuracy, a task completion time, or a task acceptance rate, and the set of input parameters corresponds to at least one of a task-incentive associated with the image-analysis task, a playback speed associated with the image-analysis task, a country in which the image-analysis task is posted, a lifetime of the image-analysis task, time available to complete the image-analysis task, days of a week on which the image-analysis task is attempted, a time of the day when the image-analysis task is attempted, a recurrence of the image-analysis task, skill set associated with a crowdworker, availability of the crowdworker to perform the image-analysis task, motivation of the crowdworker in performing the image-analysis task, intentions of the crowdworker in performing the image-analysis task, and difficulty level of the image-analysis task; determining, by the one or more processors, a distribution of the values of the at least one output parameter; creating, by the one or more processors, the performance statistical model, based on the set of input parameters and one or more characteristics associated with the distribution of the values of the at least one output parameter; and validating, by the one or more processors, the performance statistical model, wherein the validation comprises:
estimating one or more second characteristics associated with a second distribution of values of the at least one output parameter, based on a real-time data associated with the image-analysis task;comparing at least one of the distribution of the values of the at least one output parameter and the second distribution, or the one or more characteristics and the one or more second characteristics;validating the performance statistical model based on the comparison; andcreating a new performance statistical model utilizing the one or more second characteristics, based on the validation.