Fine-grained recommendation mechanism for inhibiting water army problem in crowdsourcing system

Abstract

The invention discloses a fine-grained recommendation mechanism for inhibiting a water army problem in a crowdsourcing system. The fine-grained recommendation mechanism for inhibiting the water army problem in the crowdsourcing system comprises the following steps that Q problems and K topics are input into a topic model, Q and K are both positive integers, the topic k of a question q is obtained,q is one of the Q problems, and k is one of the K topics; the professional level and interest degree of users in the user list with respect to the theme k is calculated, and a user portrait is generated; and the ranked user list is output for the task with the theme of k by combining the interestingness and professional level collaborative perception. According to the invention, the interestingness and the professional level are utilized to play a role together; in addition, by means of the fact that when interest exists, reacting force can be applied to the professional level, meanwhile, itis considered that the interest degree of the user changes along with time in the real world, the water army problem in the crowdsourcing system is restrained, the completion quality of the crowdsourcing task is improved, and the user experience degree is improved.

Description

technical field [0001] The invention relates to the technical field of data sorting, in particular to a fine-grained recommendation method and a recommendation system for suppressing troll problems in a crowdsourcing system. Background technique [0002] With the development of online services on the Internet, the working mode of using group intelligence to perform large-scale tasks has been more and more widely used. Crowdsourcing is a work model in which task requesters recruit workers with different abilities to handle tasks through monetary rewards. The benefits brought by the crowdsourcing model have been widely recognized today, and there are more than one crowdsourcing platform (such as Amazon Mechanical Turk) started running. The main components of the crowdsourcing system include: service providers who operate the platform, task requesters who initiate tasks, and users who participate in tasks and give answers. [0003] Although crowdsourcing has made great progre...

AI Technical Summary

Problems solved by technology

Although these methods are relatively more reasonable than methods that only consider reliability, they are still lacking in granularity compared with IntexCrowd

First, these methods do not take into account that interest is time-varying. Second, almost all of these methods only use professional knowledge or interest as the main recommendation criterion and ignore the synergistic effect of interest and professional level.

and ignores the fact that level of interest can sometimes exert counterforce to level of expertise

Images