Brain-Computer Interface (BCI) technology has created solutions to assist and augment human cognitive or sensory motor functions through several communication channels between the brain and external devices. BCI operates by processing temporal features extracted from neurological data such as EEG due to its relatively inexpensive and non-intrusive nature. However, EEG data can be very difficult to classify with machine learning and statistical methods due to its high signal-to-noise ratio as well as non-linear and non-stationary characteristics of data. Therefore, this paper proposes a novel approach to algorithm selection for complicated data such as EEG by creating a voting algorithm, `ML Democracy’. ML Democracy combines multiple machine learning algorithms through bootstrap aggregating, which is also known as bagging and model stacking. Three weighting methods will be presented in this paper including (i) distance from randomness, (ii) distance from perfect accuracy, and (iii) learned weights by a neural network. Several algorithms including neural network and ensemble learning methods are considered. The efficacy of proposed approach is demonstrated using a Physionet motor-imagery EEG data set [17].

Abstract Machine clients are increasingly making use of the Web to perform tasks. While Web services traditionally mimic remote procedure calling interfaces, a new generation of so-called hypermedia APIs works through hyperlinks and forms, in a way similar to how people browse the Web. This means that existing composition techniques, which determine a procedural plan upfront, are not sufficient to consume hypermedia APIs, which need to be navigated at runtime. Clients instead need a more dynamic plan that allows them to follow hyperlinks and use forms with a preset goal. Therefore, in this paper, we show how compositions of hypermedia APIs can be created by generic Semantic Web reasoners. This is achieved through the generation of a proof based on semantic descriptions of the APIs' functionality. To pragmatically verify the applicability of compositions, we introduce the notion of pre-execution and post-execution proofs. The runtime interaction between a client and a server is guided by proofs but driven by hypermedia, allowing the client to react to the application's actual state indicated by the server's response. We describe how to generate compositions from descriptions, discuss a computer-assisted process to generate descriptions, and verify reasoner performance on various composition tasks using a benchmark suite. The experimental results lead to the conclusion that proof-based consumption of hypermedia APIs is a feasible strategy at Web scale.

Topic Modelling (TM) has gained momentum over the last few years within the humanities to analyze topics represented in large volumes of full text. This paper proposes an experiment with the usage of TM based on a large subset of digitized archival holdings of the European Commission (EC). Currently, millions of scanned and OCRed files are available and hold the potential to significantly change the way historians of the construction and evolution of the European Union can perform their research. However, due to a lack of resources, only minimal metadata are available on a file and document level, seriously undermining the accessibility of this archival collection. The article explores in an empirical manner the possibilities and limits of TM to automatically extract key concepts from a large body of documents spanning multiple decades. By mapping the topics to headings of the EUROVOC thesaurus, the proof of concept described in this paper offers the future possibility to represent the identified topics with the help of a hierarchical search interface for end-users.

Collaborative knowledge bases that make their data freely available in a machine-readable form are central for the data strategy of many projects and organizations. The two major collaborative knowledge bases are Wikimedia's Wikidata and Google's Freebase. Due to the success of Wikidata, Google decided in 2014 to offer the content of Freebase to the Wikidata community. In this paper, we report on the ongoing transfer efforts and data mapping challenges, and provide an analysis of the effort so far. We describe the Primary Sources Tool, which aims to facilitate this and future data migrations. Throughout the migration, we have gained deep insights into both Wikidata and Freebase, and share and discuss detailed statistics on both knowledge bases.

We have created and evaluated an algorithm capable of deduplicating and clustering exact- and near-duplicate media items of type photo and video that get shared on multiple social networks in the context of events. This algorithm works in an entirely ad hoc manner without requiring any pre-calculation. When people attend events, they more and more share event-related media items publicly on social networks to let their social network contacts relive and witness the attended events. In the past, we have worked on methods to accumulate such public user-generated multimedia content in order to summarize events visually, for example, in the form of media galleries or slideshows. In this paper, first, we introduce social-network-specific reasons and challenges that cause near-duplicate media items. Second, we detail an algorithm for the task of deduplicating and clustering exact- and near-duplicate media items stemming from multiple social networks. Finally, we evaluate the algorithm's strengths and weaknesses and thoroughly compare its performance with the state-of-the-art feature detection algorithms SIFT, ASIFT and SURF and show that for the given use case it performs almost equally well accuracy-wise, but strongly outperforms speed-wise.

Wikipedia is a global crowdsourced encyclopedia that at time of writing is available in 287 languages. Wikidata is a likewise global crowdsourced knowledge base that provides shared facts to be used by Wikipedias. In the context of this research, we have developed an application and an underlying Application Programming Interface (API) capable of monitoring realtime edit activity of all language versions of Wikipedia and Wikidata. This application allows us to easily analyze edits in order to answer questions such as "Bots vs. Wikipedians, who edits more?", "Which is the most anonymously edited Wikipedia?", or "Who are the bots and what do they edit?". To the best of our knowledge, this is the first time such an analysis was done for Wikidata and for really all Wikipedias---large and small. According to our results, all Wikipedias and Wikidata together are edited by about 50% bots and by about 23% anonymous users. Wikidata alone accounts for about 48% of the totally observed edits. If we do not consider Wikidata, i.e., if we only look at all Wikipedias, about 15% of all edits are made by bots and 26% of all edits are made by anonymous users. Overall, we found a stabilizing number of 274 active bots during our observation period. Our application is available publicly online at the URL http://wikipedia-edits.herokuapp.com/, its code has been open-sourced under the Apache 2.0 license.

In this paper, we report on work around the I-SEARCH EU (FP7 ICT STREP) project whose objective is the development of a multimodal search engine targeted at mobile and desktop devices. Each of these device classes has its specific hardware capabilities and set of supported features. In order to provide a common multimodal search experience across device classes, one size does not fit all. We highlight ways to achieve the same functionality agnostic of the device class being used for the search, and present concrete use cases.