Ioanna Lykourentzou

Task assignment optimization in knowledge-intensive crowdsourcing

We present SmartCrowd, a framework for optimizing task assignment in knowledge-intensive crowdsourcing (KI-C). SmartCrowd distinguishes itself by formulating, for the first time, the problem of worker-to-task assignment in KI-C as an optimization problem, by proposing efficient adaptive algorithms to solve it and by accounting for human factors, such as worker expertise, wage requirements, and availability inside the optimization process. We present rigorous theoretical analyses of the task assignment optimization problem and propose optimal and approximation algorithms with guarantees, which rely on index pre-computation and adaptive maintenance. We perform extensive performance and quality experiments using real and synthetic data to demonstrate that the SmartCrowd approach is necessary to achieve efficient task assignments of high-quality under guaranteed cost budget.

A semantic approach to concept lattice-based information retrieval

The volume of available information is growing, especially on the web, and in parallel the questions of the users are changing and becoming harder to satisfy. Thus there is a need for organizing the available information in a meaningful way in order to guide and improve document indexing for information retrieval applications taking into account more complex data such as semantic relations. In this paper we show that Formal Concept Analysis (FCA) and concept lattices provide a suitable and powerful support for such a task. Accordingly, we use FCA to compute a concept lattice, which is considered both a semantic index to organize documents and a search space to model terms. We introduce the notions of cousin concepts and classification-based reasoning for navigating the concept lattice and retrieve relevant information based on the content of concepts. Finally, we detail a real-world experiment and show that the present approach has very good capabilities for semantic indexing and document retrieval.

Collective intelligence-based resource allocation to optimise knowledge and innovation harnessing in corporate environments

A typical problem that large enterprises face is how to effectively harness the intangible knowledge, expertise, skills and lessons learnt of their staff members. Intangible knowledge harnessing and its codification into usable document formats are vital for the enterprise since they directly affect its ability to innovate and solve complex new problems. However, the identification of the individuals whose contribution can solve each knowledge-demanding problem may be difficult. In this paper, we propose a novel mechanism that uses the collective intelligence of the corporate crowd to identify the tacit knowledge competencies of each employee and coordinate their contributions, inside a wiki-like system, so that each individual may contribute in the most efficient way. Experimental results show that, compared to the fully self-coordinated pattern used by current collaborative knowledge harnessing approaches, the proposed mechanism can help the corporate community allocate its intangible skill resources more efficiently, and thus produce more qualitative knowledge in a timelier manner.

Optimizing Crowd Performance through Collective Intelligence-Based System Design: An Application on Social Network Information Diffusion

In this paper we propose a novel system design approach to improve the performance of large-scale interacting user communities towards a collective goal. The proposed approach has been applied on the example case of social network information diffusion and results show that it helped the participating community achieve less spam and improved information coverage.