面向众包任务的个性化搜索技术的研究与实现

发布时间：2018-05-15 11:39

本文选题：众包 + 任务　；参考：《扬州大学》2017年硕士论文

【摘要】：随着互联网技术迅速发展和“互联网+”时代的到来,越来越多的企业开始使用众包平台,来寻找解决方案。由于众包平台上的企业种类越来越多,每天发布任务的数量急剧增加,使得众包平台上的工人难以快速找到自己喜欢和擅长的任务。因此工人需要花费大量的时间挑选任务,使得工人在平台上的工作效率不高。为了帮助工人在众包平台挑选任务,提高其工作效率,本文提出了一个新的众包任务搜索技术。在搜索众包任务时,该技术主要利用了探索性搜索技术,考虑了工人的历史任务完成情况以及他们的搜索偏好等行为特征。其主要研究工作如下:(1)建立了工人兴趣偏好模型。本文将工人的兴趣偏好主要划分为长期兴趣偏好和短期兴趣偏好两部分。长期兴趣偏好一部分提取于工人在注册众包平台时填写相关表格内容,另一部分主要来自于分析工人历史任务的完成情况。本文在分析工人历史任务时,首先提取工人的相关工作文档,然后使用基于LDA模型的主题词提取方法提取相关主题词作为工人长期兴趣的特征表示。实验结果表明使用本模型的搜索结果精度更高。(2)研究了探索性众包任务搜索技术。探索性搜索技术在搜索众包任务时结合了工人的历史任务完成情况和工人自身的搜索特点,将搜索结果个性化的展现给工人。通过分析工人在众包平台上的多次交互行为,可以引导工人的搜索偏向,从而进一步挖掘出他们的搜索偏好,直到工人找到感兴趣的任务。实验结果表明探索性搜索技术的应用可以有效的缩短工人们的搜索时间,提高工人搜索任务的效率。(3)众包平台的工人搜索任务原型系统实现。首先介绍了系统的需求分析,并进行功能分析,然后提出系统框架的总体设计目标,并且将总体框架设计成多模块层次结构,登陆注册模块、任务模块、任务搜索模块、后台管理模块,最后实现了各个功能。经过测试,系统各模块可以稳定运行,实现了工人个性化任务搜索,缩短了工人搜索任务的时间。
[Abstract]:With the rapid development of Internet technology and the arrival of "Internet" era, more and more enterprises begin to use crowdsourcing platform to find solutions. Due to more and more types of enterprises in crowdsourcing platform, the number of tasks released every day increases dramatically, which makes it difficult for workers on crowdsourcing platform to find their favorite and good tasks quickly. As a result, workers spend a lot of time selecting tasks, making them less efficient on the platform. In order to help workers select tasks in crowdsourcing platform and improve their working efficiency, a new crowdsourcing task search technique is proposed in this paper. When searching crowdsourcing tasks, this technique mainly uses exploratory search technology, taking into account the behavior characteristics of workers' historical task completion and their search preferences. The main research work is as follows: 1) the model of interest preference of workers is established. In this paper, workers' interest preference is divided into two parts: long term interest preference and short term interest preference. One part of the long-term interest preference is extracted from the worker filling out the relevant form content when registering the crowdsourcing platform, the other part mainly comes from the analysis of the completion of the worker's historical task. In this paper, when analyzing the historical task of workers, we first extract the relevant work documents of workers, and then use the LDA model-based method to extract the relevant subject words as the feature representation of workers' long-term interest. The experimental results show that the precision of the model is higher than that of the others. (2) the exploratory crowdsourcing task search technique is studied. When searching crowdsourcing tasks, exploratory search technology combines the completion of workers' historical tasks and the characteristics of workers' own search, and presents the search results to workers individualized. By analyzing workers' interactions on crowdsourcing platforms, the search bias of workers can be guided, and their search preferences can be further excavated until the workers find the task of interest. The experimental results show that the application of exploratory search technology can effectively shorten the search time of workers and improve the efficiency of workers' search tasks. Firstly, the requirement analysis of the system is introduced, and the function analysis is carried out, then the overall design goal of the system framework is put forward, and the overall framework is designed into multi-module hierarchy, login registration module, task search module. Background management module, the final realization of each function. After testing, each module of the system can run stably, realize the worker individualized task search, shorten the time of the worker search task.
【学位授予单位】：扬州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3

【相似文献】