Modern Greek is normally written in the Greek alphabet. In informal online messages, however, Greek is often written using characters available on Latin-character keyboards, a form known as Greeklish. Originally used to bypass the lack of support for the Greek alphabet in older computers, Greeklish is now also used to avoid switching languages on multilingual keyboards, hide spelling mistakes, or as a form of slang. There is no consensus mapping, hence the same Greek word can be written in numerous different ways in Greeklish. Even native Greek speakers may struggle to understand (or be annoyed by) Greeklish, which requires paying careful attention to context to decipher. Greeklish may also be a problem for NLP models trained on Greek datasets written in the Greek alphabet. Experimenting with a range of statistical and deep learning models on both artificial and real-life Greeklish data, we find that: (i) prompting large language models (e.g., GPT-4) performs impressively well with few- or even zero-shot training, outperforming several fine-tuned encoder-decoder models; however (ii) a twenty years old statistical Greeklish transliteration model is still very competitive; and (iii) the problem is still far from having been solved; (iv) nevertheless, downstream Greek NLP systems that need to cope with Greeklish, such as moderation classifiers, can benefit significantly even with the current non-perfect transliteration systems. We make all our code, models, and data available and suggest future improvements, based on an analysis of our experimental results.

Can a chatbot enable us to change our conceptions, to be critically reflective? To what extent can interaction with a technologically 'minimal' medium such as a chatbot evoke emotional engagement in ways that can challenge us to act on the world? In this paper, we discuss the design of a provocative bot, a 'bot of conviction', aimed at triggering conversations on complex topics (e.g. death, wealth distribution, gender equality, privacy) and, ultimately, soliciting specific actions from the user it converses with. We instantiate our design with a use case in the cultural sector, specifically a Neolithic archaeological site that acts as a stage of conversation on such hard themes. Our larger contributions include an interaction framework for bots of conviction, insights gained from an iterative process of participatory design and evaluation, and a vision for bot interaction mechanisms that can apply to the HCI community more widely.

In this work we present some findings on how gesture-based interfaces can stimulate and motivate children into learning the basics of music notation. We report on a serious game we developed that relies on a hand detection interface using motion-sensing camera technology. Our initial experiments indicate the accuracy, effectiveness, and user acceptance of the gesture-based interface among children, thereby validating the idea that such interfaces can be used to facilitate the teaching of music concepts.

We look at agents reasoning about actions from a first-person perspective. The agent has a representation of world as situation calculus action theory. It can perform sensing actions to acquire information. The agent acts “online”, i.e., it performs an action only if it is certain that the action can be executed, and collects sensing results from the actual world. When the agent reasons about its future actions, it indeed considers that it is acting online; however only possible sensing values are available. The kind of reasoning about actions we consider for the agent is verifying a first-order (FO) variant (without quantification across situations) of linear time temporal logic (LTL) . We mainly focus on bounded action theories, where the number of facts that are true in any situation is bounded. The main results of this paper are: (i) possible sensing values can be based on consistency if the initial situation description is FO; (ii) for bounded action theories, progression over histories that include sensing results is always FO; (iii) for bounded theories, verifying our FO LTL against online executions with sensing is decidable.

In this paper we present the idea that by using AI planning in concert with formal task modeling, the overhead associated with plan creation for complex tasks can be reduced. The proposed approach uses a SysML taxonomy to model the system capabilities and the process specification, and the PDDL planning language to determine acceptable objective solutions. This idea is applied to the manufacturing domain, and examples are shown modeling a multi-robot system in an automobile manufacturing environment. A discussion is given regarding the merits of the demonstrated approach.

TELEIOS is a recent European project that addresses the need for scalable access to petabytes of Earth Observation data and the discovery and exploitation of knowledge that is hidden in them. TELEIOS builds on scientific database technologies (array databases, SciQL, data vaults) and Semantic Web technologies (stRDF and stSPARQL) implemented on top of a state of the art column store database system (MonetDB). We demonstrate a first prototype of the TELEIOS Virtual Earth Observatory (VEO) architecture, using a forest fire monitoring application as example.