scholarly journals Complex crowdsourcing task allocation strategies employing supervised and reinforcement learning

2017 ◽  
Vol 1 (2) ◽  
pp. 146-160
Author(s):  
Lizhen Cui ◽  
Xudong Zhao ◽  
Lei Liu ◽  
Han Yu ◽  
Yuan Miao
Keyword(s):  
Reinforcement Learning ◽  
Task Allocation ◽  
Large Scale ◽  
Expert Knowledge ◽  
Policy Network ◽  
Allocation Strategy ◽  
Theoretical Frameworks ◽  
Data Set ◽  
Content Type ◽  
Allocation Strategies

Purpose Allocation of complex crowdsourcing tasks, which typically include heterogeneous attributes such as value, difficulty, skill required, effort required and deadline, is still a challenging open problem. In recent years, agent-based crowdsourcing approaches focusing on recommendations or incentives have emerged to dynamically match workers with diverse characteristics to tasks to achieve high collective productivity. However, existing approaches are mostly designed based on expert knowledge grounded in well-established theoretical frameworks. They often fail to leverage on user-generated data to capture the complex interaction of crowdsourcing participants’ behaviours. This paper aims to address this challenge. Design/methodology/approach The paper proposes a policy network plus reputation network (PNRN) approach which combines supervised learning and reinforcement learning to imitate human task allocation strategies which beat artificial intelligence strategies in this large-scale empirical study. The proposed approach incorporates a policy network for the selection of task allocation strategies and a reputation network for calculating the trends of worker reputation fluctuations. Then, by iteratively applying the policy network and reputation network, a multi-round allocation strategy is proposed. Findings PNRN has been trained and evaluated using a large-scale real human task allocation strategy data set derived from the Agile Manager game with close to 500,000 decision records from 1,144 players in over 9,000 game sessions. Extensive experiments demonstrate the validity and efficiency of computational complex crowdsourcing task allocation strategy learned from human participants. Originality/value The paper can give a better task allocation strategy in the crowdsourcing systems.

Optimization of hierarchical reinforcement learning relationship extraction model

2020 ◽  
Vol 48 (3) ◽  
pp. 129-136
Author(s):  
Qihang Wu ◽  
Daifeng Li ◽  
Lu Huang ◽  
Biyun Ye
Keyword(s):  
Reinforcement Learning ◽  
Large Scale ◽  
Relation Extraction ◽  
Unified Framework ◽  
Data Set ◽  
Joint Learning ◽  
Content Type ◽  
Hierarchical Reinforcement Learning ◽  
Learning Framework ◽  
Unstructured Text

Purpose Entity relation extraction is an important research direction to obtain structured information. However, most of the current methods are to determine the relations between entities in a given sentence based on a stepwise method, seldom considering entities and relations into a unified framework. The joint learning method is an optimal solution that combines relations and entities. This paper aims to optimize hierarchical reinforcement learning framework and provide an efficient model to extract entity relation. Design/methodology/approach This paper is based on the hierarchical reinforcement learning framework of joint learning and combines the model with BERT, the best language representation model, to optimize the word embedding and encoding process. Besides, this paper adjusts some punctuation marks to make the data set more standardized, and introduces positional information to improve the performance of the model. Findings Experiments show that the model proposed in this paper outperforms the baseline model with a 13% improvement, and achieve 0.742 in F1 score in NYT10 data set. This model can effectively extract entities and relations in large-scale unstructured text and can be applied to the fields of multi-domain information retrieval, intelligent understanding and intelligent interaction. Originality/value The research provides an efficient solution for researchers in a different domain to make use of artificial intelligence (AI) technologies to process their unstructured text more accurately.

Financial distress determinants among SMEs: empirical evidence from Sweden

2020 ◽  
Vol 47 (3) ◽  
pp. 547-560 ◽  
Author(s):  
Darush Yazdanfar ◽  
Peter Öhman
Keyword(s):  
Financial Crisis ◽  
Financial Distress ◽  
Large Scale ◽  
Global Financial Crisis ◽  
Binary Logistic Regression ◽  
Data Availability ◽  
Cross Sectional ◽  
Data Set ◽  
Content Type ◽  
The Global Financial Crisis

PurposeThe purpose of this study is to empirically investigate determinants of financial distress among small and medium-sized enterprises (SMEs) during the global financial crisis and post-crisis periods.Design/methodology/approachSeveral statistical methods, including multiple binary logistic regression, were used to analyse a longitudinal cross-sectional panel data set of 3,865 Swedish SMEs operating in five industries over the 2008–2015 period.FindingsThe results suggest that financial distress is influenced by macroeconomic conditions (i.e. the global financial crisis) and, in particular, by various firm-specific characteristics (i.e. performance, financial leverage and financial distress in previous year). However, firm size and industry affiliation have no significant relationship with financial distress.Research limitationsDue to data availability, this study is limited to a sample of Swedish SMEs in five industries covering eight years. Further research could examine the generalizability of these findings by investigating other firms operating in other industries and other countries.Originality/valueThis study is the first to examine determinants of financial distress among SMEs operating in Sweden using data from a large-scale longitudinal cross-sectional database.

Different firm responses to the COVID-19 pandemic shocks: machine-learning evidence on the Vietnamese labor market

2021 ◽  
Vol ahead-of-print (ahead-of-print) ◽  
Author(s):  
Lam Hoang Viet Le ◽  
Toan Luu Duc Huynh ◽  
Bryan S. Weber ◽  
Bao Khac Quoc Nguyen
Keyword(s):  
Machine Learning ◽  
Labor Market ◽  
Large Scale ◽  
Government Support ◽  
Policy Implications ◽  
Machine Learning Techniques ◽  
Firm Characteristics ◽  
Data Set ◽  
Content Type ◽  
Firm Responses

PurposeThis paper aims to identify the disproportionate impacts of the COVID-19 pandemic on labor markets.Design/methodology/approachThe authors conduct a large-scale survey on 16,000 firms from 82 industries in Ho Chi Minh City, Vietnam, and analyze the data set by using different machine-learning methods.FindingsFirst, job loss and reduction in state-owned enterprises have been significantly larger than in other types of organizations. Second, employees of foreign direct investment enterprises suffer a significantly lower labor income than those of other groups. Third, the adverse effects of the COVID-19 pandemic on the labor market are heterogeneous across industries and geographies. Finally, firms with high revenue in 2019 are more likely to adopt preventive measures, including the reduction of labor forces. The authors also find a significant correlation between firms' revenue and labor reduction as traditional econometrics and machine-learning techniques suggest.Originality/valueThis study has two main policy implications. First, although government support through taxes has been provided, the authors highlight evidence that there may be some additional benefit from targeting firms that have characteristics associated with layoffs or other negative labor responses. Second, the authors provide information that shows which firm characteristics are associated with particular labor market responses such as layoffs, which may help target stimulus packages. Although the COVID-19 pandemic affects most industries and occupations, heterogeneous firm responses suggest that there could be several varieties of targeted policies-targeting firms that are likely to reduce labor forces or firms likely to face reduced revenue. In this paper, the authors outline several industries and firm characteristics which appear to more directly be reducing employee counts or having negative labor responses which may lead to more cost–effect stimulus.

Computational modeling of Weibo user influence based on information interactive network

2016 ◽  
Vol 40 (7) ◽  
pp. 867-881 ◽  
Author(s):  
Dingguo Yu ◽  
Nan Chen ◽  
Xu Ran
Keyword(s):  
Social Networks ◽  
Computational Model ◽  
Large Scale ◽  
Mobile Internet ◽  
Activity Level ◽  
Message Content ◽  
Data Set ◽  
Sina Weibo ◽  
Content Type ◽  
User Influence

Purpose With the development and application of mobile internet access, social media represented by Weibo, WeChat, etc. has become the main channel for information release and sharing. High-impact users in social networks are key factors stimulating the large-scale propagation of information within social networks. User influence is usually related to the user’s attention rate, activity level, and message content. The paper aims to discuss these issues. Design/methodology/approach In this paper, the authors focused on Sina Weibo users, centered on users’ behavior and interactive information, and formulated a weighted interactive information network model, then present a novel computational model for Weibo user influence, which combined multiple indexes such as the user’s attention rate, activity level, and message content influence, etc., the model incorporated the time dimension, through the calculation of users’ attribute influence and interactive influence, to comprehensively measure the user influence of Sina Weibo users. Findings Compared with other models, the model reflected the dynamics and timeliness of the user influence in a more accurate way. Extensive experiments are conducted on the real-world data set, and the results validate the performance of the approach, and demonstrate the effectiveness of the dynamics and timeliness. Due to the similarity in platform architecture and user behavior between Sina Weibo and Twitter, the calculation model is also applicable to Twitter. Originality/value This paper presents a novel computational model for Weibo user influence, which combined multiple indexes such as the user’s attention rate, activity level, and message content influence, etc.

Czech civil sector face-to-face with freelance activism

2016 ◽  
Vol 36 (11/12) ◽  
pp. 774-791
Author(s):  
Pavol Frič ◽  
Martin Vávra
Keyword(s):  
Large Scale ◽  
Civil Society Organizations ◽  
Political Consumerism ◽  
Data Set ◽  
Content Type ◽  
Individual Level ◽  
Face To Face ◽  
Online Activism ◽  
Civil Sector ◽  
Relationship Of

Purpose The purpose of this paper is to answer following question: what is the relationship between member activism performed through civil society organizations (CSOs) and individualized freelance activism (in form of online activism, everyday making, political consumerism or checkbook activism) independent of organizational framework? Is it a relationship of mutual competition or support? Design/methodology/approach Analysis is carried out on data from 2009 questionnaire-based survey on volunteering, representative for adult Czech population. The data set allowed the authors to relate member activism with freelance activism and in case of member activism distinguish the type of organization and the level of its professionalization. Findings Dominant pattern the authors identified in data is mutual support of both types of volunteering documented by significant overlap of these forms of public engagement. The most striking is the overlap for active members of new advocacy NGOs and the weakest for traditional clubs. Regression analysis shows that on an individual level “mixed activism” (compared with “pure freelance activism”) is linked with higher education and higher confidence in civic organizations. Originality/value The civil practice of individualized freelance activism was described and analysed by various authors as an activity of specific types of activist, but there has not yet been any research giving reflection on such a large scale of freelance activism types as in the analysis. The authors set them together in contrast to the member (collective, organized) form of civic activism and also took into account the influence of professionalization and type of CSOs.

Explore Deep Neural Network and Reinforcement Learning to Large-scale Tasks Processing in Big Data

2019 ◽  
Vol 33 (13) ◽  
pp. 1951010 ◽  
Author(s):  
Chunyi Wu ◽  
Gaochao Xu ◽  
Yan Ding ◽  
Jia Zhao
Keyword(s):  
Neural Network ◽  
Big Data ◽  
Reinforcement Learning ◽  
Large Scale ◽  
Policy Network ◽  
Text Learning ◽  
Mapping Algorithm ◽  
Task Requirements ◽  
Network Mapping

Large-scale tasks processing based on cloud computing has become crucial to big data analysis and disposal in recent years. Most previous work, generally, utilize the conventional methods and architectures for general scale tasks to achieve tons of tasks disposing, which is limited by the issues of computing capability, data transmission, etc. Based on this argument, a fat-tree structure-based approach called LTDR (Large-scale Tasks processing using Deep network model and Reinforcement learning) has been proposed in this work. Aiming at exploring the optimal task allocation scheme, a virtual network mapping algorithm based on deep convolutional neural network and [Formula: see text]-learning is presented herein. After feature extraction, we design and implement a policy network to make node mapping decisions. The link mapping scheme can be attained by the designed distributed value-function based reinforcement learning model. Eventually, tasks are allocated onto proper physical nodes and processed efficiently. Experimental results show that LTDR can significantly improve the utilization of physical resources and long-term revenue while satisfying task requirements in big data.

Review on state-of-the-art dynamic task allocation strategies for multiple-robot systems

2020 ◽  
Vol 47 (6) ◽  
pp. 929-942
Author(s):  
Seenu N. ◽  
Kuppan Chetty R.M. ◽  
Ramya M.M. ◽  
Mukund Nilakantan Janardhanan
Keyword(s):  
Task Allocation ◽  
State Of The Art ◽  
Objective Functions ◽  
Research Directions ◽  
Content Type ◽  
Dynamic Task ◽  
Dynamic Task Allocation ◽  
Robot Systems ◽  
Allocation Strategies ◽  
Multi Robot

Purpose This paper aims to present a concise review on the variant state-of-the-art dynamic task allocation strategies. It presents a thorough discussion about the existing dynamic task allocation strategies mainly with respect to the problem application, constraints, objective functions and uncertainty handling methods. Design/methodology/approach This paper briefs the introduction of multi-robot dynamic task allocation problem and discloses the challenges that exist in real-world dynamic task allocation problems. Numerous task allocation strategies are discussed in this paper, and it establishes the characteristics features between them in a qualitative manner. This paper also exhibits the existing research gaps and conducive future research directions in dynamic task allocation for multiple mobile robot systems. Findings This paper concerns the objective functions, robustness, task allocation time, completion time, and task reallocation feature for performance analysis of different task allocation strategies. It prescribes suitable real-world applications for variant task allocation strategies and identifies the challenges to be resolved in multi-robot task allocation strategies. Originality/value This paper provides a comprehensive review of dynamic task allocation strategies and incites the salient research directions to the researchers in multi-robot dynamic task allocation problems. This paper aims to summarize the latest approaches in the application of exploration problems.

Maternal incarceration penalty: an examination of the effect of maternal conviction and incarceration on childhood delinquency

2017 ◽  
Vol 7 (1) ◽  
pp. 29-46 ◽  
Author(s):  
Kristen N. Sobba ◽  
Brenda Prochaska ◽  
Emily Berthelot
Keyword(s):  
Criminal Behavior ◽  
Delinquent Behavior ◽  
Binary Logistic Regression ◽  
Self Control ◽  
Theoretical Frameworks ◽  
Data Set ◽  
Content Type ◽  
Low Self Control ◽  
Property Destruction ◽  
The Impact

Purpose Several studies have reported the impact of paternal incarceration and criminal behavior on childhood delinquency; however, fewer studies have addressed the influence of maternal criminality on children’s behavioral outcomes. Integrating self-control and attachment theoretical frameworks, the purpose of this paper is to address the impact of mothers who have been stopped, arrested, convicted, and incarcerated in relation to their children’s delinquent behavior. Design/methodology/approach The Fragile Families and Child Wellbeing data set was used to better understand this relationship. By using binary logistic regression, two types of delinquent behavior were assessed: destroying property and fighting. Findings The results revealed that mothers’ criminal behavior affected children’s fighting tendencies but did not significantly impact children’s tendency to destroy property. Furthermore, certain childhood antisocial traits and demographic characteristics revealed to also impact children’s delinquent behavior. From the results, implications and prevention strategies were drawn describing techniques to combat delinquency. Originality/value This research lays a foundation for future researchers to explore mother-child attachment and the transmission of low self-control from mother to child in relation to criminality. The current research is one of the first studies to specifically address how maternal criminal behavior affects their children’s tendency to engage in delinquency, specifically examining property destruction and fighting.

scholarly journals Learning to rank with click-through features in a reinforcement learning framework

2016 ◽  
Vol 12 (4) ◽  
pp. 448-476 ◽  
Author(s):  
Amir Hosein Keyhanipour ◽  
Behzad Moshiri ◽  
Maryam Piroozmand ◽  
Farhad Oroumchian ◽  
Ali Moeini
Keyword(s):  
Reinforcement Learning ◽  
Learning To Rank ◽  
Training Data ◽  
High Dimensionality ◽  
Compact Representation ◽  
Second Phase ◽  
Data Sets ◽  
Data Set ◽  
Content Type ◽  
Benchmark Data

Purpose Learning to rank algorithms inherently faces many challenges. The most important challenges could be listed as high-dimensionality of the training data, the dynamic nature of Web information resources and lack of click-through data. High dimensionality of the training data affects effectiveness and efficiency of learning algorithms. Besides, most of learning to rank benchmark datasets do not include click-through data as a very rich source of information about the search behavior of users while dealing with the ranked lists of search results. To deal with these limitations, this paper aims to introduce a novel learning to rank algorithm by using a set of complex click-through features in a reinforcement learning (RL) model. These features are calculated from the existing click-through information in the data set or even from data sets without any explicit click-through information. Design/methodology/approach The proposed ranking algorithm (QRC-Rank) applies RL techniques on a set of calculated click-through features. QRC-Rank is as a two-steps process. In the first step, Transformation phase, a compact benchmark data set is created which contains a set of click-through features. These feature are calculated from the original click-through information available in the data set and constitute a compact representation of click-through information. To find most effective click-through feature, a number of scenarios are investigated. The second phase is Model-Generation, in which a RL model is built to rank the documents. This model is created by applying temporal difference learning methods such as Q-Learning and SARSA. Findings The proposed learning to rank method, QRC-rank, is evaluated on WCL2R and LETOR4.0 data sets. Experimental results demonstrate that QRC-Rank outperforms the state-of-the-art learning to rank methods such as SVMRank, RankBoost, ListNet and AdaRank based on the precision and normalized discount cumulative gain evaluation criteria. The use of the click-through features calculated from the training data set is a major contributor to the performance of the system. Originality/value In this paper, we have demonstrated the viability of the proposed features that provide a compact representation for the click through data in a learning to rank application. These compact click-through features are calculated from the original features of the learning to rank benchmark data set. In addition, a Markov Decision Process model is proposed for the learning to rank problem using RL, including the sets of states, actions, rewarding strategy and the transition function.

Farm-level evidence on risk balancing behavior in the EU-15

2014 ◽  
Vol 74 (1) ◽  
pp. 17-37 ◽  
Author(s):  
Yann de Mey ◽  
Frankwin van Winsen ◽  
Erwin Wauters ◽  
Mark Vancauteren ◽  
Ludwig Lauwers ◽  
...  
Keyword(s):  
Fixed Effects ◽  
Large Scale ◽  
The European Union ◽  
Data Set ◽  
Correlation Relationship ◽  
Content Type ◽  
Farm Level ◽  
Relationship Analysis ◽  
Trade Offs ◽  
Fixed Effects Regression

Purpose – The purpose of this paper is to present empirical evidence of risk balancing behavior by European farmers. More specifically, the authors investigate strategic adjustments in the level of financial risk (FR) in response to changes in the level of business risk (BR). Design/methodology/approach – The authors conducted a correlation relationship analysis and run several linear fixed effects regression models using the European Union (EU)-15 FADN panel data set for the period 1995-2008. Findings – Overall, the paper finds EU evidence of risk balancing. The correlation relationship analysis suggests that just over half of the farm observations are risk balancers whereas the other (smaller) half are not. The coefficient in our fixed effects regression suggests that a 1 percent increase in BR reduces FR by 0.043 percent and has a standard error so low that the existence of non-risk balancers is doubtful. The results reject evidence of strong-form risk balancing – inverse trade-offs between FR and BR keeping total risk (TR) constant – but cannot reject weak-form risk balancing – inverse trade-offs between FR and BR with some observed changes in TR. Furthermore, the extent of risk balancing behavior is found to differ between different European countries and across farm typologies. Practical implications – This study provides European policy makers a first insight into risk balancing behavior of EU farmers. When risk balancing occurs, BR-reducing agricultural policies induce strategic upwards leverage adjustments that unintentionally reestablish or even increase total farm-level risk. Originality/value – Making use of the large and unique FADN database, to the best of the authors knowledge, this study is the first that provides European (EU-15) evidence on risk balancing behavior, is conducted at an unprecedented large scale, and presents the first risk balancing evidence across countries and farming systems.

Sign in / Sign up

Export Citation Format

Share Document