The quality of output from large language models (LLMs), particularly in machine translation (MT), is closely tied to the quality of in-context examples (ICEs) provided along with the query, i.e., the text to translate. The effectiveness of these ICEs is influenced by various factors, such as the domain of the source text, the order in which the ICEs are presented, the number of these examples, and the prompt templates used. Naturally, selecting the most impactful ICEs depends on understanding how these affect the resulting translation quality, which ultimately relies on translation references or human judgment. This paper presents a novel methodology for in-context learning (ICL) that relies on a search algorithm guided by domain-specific quality estimation (QE). Leveraging the XGLM model, our methodology estimates the resulting translation quality without the need for translation references, selecting effective ICEs for MT to maximize translation quality. Our results demonstrate significant improvements over existing ICL methods and higher translation performance compared to fine-tuning a pre-trained language model (PLM), specifically mBART-50.

While quality estimation (QE) can play an important role in the translation process, its effectiveness relies on the availability and quality of training data. For QE in particular, high-quality labeled data is often lacking due to the high cost and effort associated with labeling such data. Aside from the data scarcity challenge, QE models should also be generalizable, i.e., they should be able to handle data from different domains, both generic and specific. To alleviate these two main issues -- data scarcity and domain mismatch -- this paper combines domain adaptation and data augmentation within a robust QE system. Our method first trains a generic QE model and then fine-tunes it on a specific domain while retaining generic knowledge. Our results show a significant improvement for all the language pairs investigated, better cross-lingual inference, and a superior performance in zero-shot learning scenarios as compared to state-of-the-art baselines.

Continuously-growing data volumes lead to larger generic models. Specific use-cases are usually left out, since generic models tend to perform poorly in domain-specific cases. Our work addresses this gap with a method for selecting in-domain data from generic-domain (parallel text) corpora, for the task of machine translation. The proposed method ranks sentences in parallel general-domain data according to their cosine similarity with a monolingual domain-specific data set. We then select the top K sentences with the highest similarity score to train a new machine translation system tuned to the specific in-domain data. Our experimental results show that models trained on this in-domain data outperform models trained on generic or a mixture of generic and domain data. That is, our method selects high-quality domain-specific training instances at low computational cost and data size.

Live Action Role-Playing (LARP) games and similar experiences are becoming a popular game genre. Here, we discuss how artificial intelligence techniques, particularly those commonly used in AI for Games, could be applied to LARP. We discuss the specific properties of LARP that make it a surprisingly suitable application field, and provide a brief overview of some existing approaches. We then outline several directions where utilizing AI seems beneficial, by both making LARPs easier to organize, and by enhancing the player experience with elements not possible without AI.

In previous research we concluded that a personality profile, based on the Five Factor Model, can be constructed from observations of a player’s behavior in a module that we designed for Neverwinter Nights (Lankveld et al. 2011a). In the present research, we investigate whether we can do the same thing in an actual modern commercial video game, in this case the game Fallout 3. We stored automatic observations on 36 participants who played the introductory stages of Fallout 3. We then correlated these observations with the participants’ personality profiles, expressed by values for five personality traits as measured by the standard NEO-FFI questionnaire. Our analysis shows correlations between all five personality traits and the game observations. These results validate and generalize the results from our previous research (Lankveld et al. 2011a). We may conclude that Fallout 3, and by extension other modern video games, allows players to express their personality, and can therefore be used to create personality profiles.

This research aims at building a preference-based player model of Civilization IV players. Our model incorporates attributes which are defined for AI players. We use a sequential minimal optimization (SMO) classifier to build the player model based on a training set with observations of a large number of games between six AI players. The model was validated on a test set of games between the same six AI players. While it did not seem to generalize well to the preferences of different AI players, it did manage to accurately predict some of the preferences for a veteran human player. Further tests showed that AI players with the same play styles but different preference values were often confused by the model. We conclude that for a complex game such as Civilization IV a model that attempts to accurately predict specific preference values is hard to construct. A model that focusses on play styles might succeed better.

Dynamic scripting is a reinforcement learning approach to adaptive game AI that learns, during gameplay, which game tactics an opponent should select to play effectively. We introduce the evolutionary state-based tactics generator (ESTG), which uses an evolutionary algorithm to generate tactics automatically. Experimental results show that ESTG improves dynamic scripting's performance in a real-time strategy game. We conclude that high-quality domain knowledge can be automatically generated for strong adaptive game AI opponents.

The decision-making process of computer-controlled opponents in video games is called game AI. Adaptive game AI can improve the entertainment value of games by allowing computer-controlled opponents to ix weaknesses automatically in the game AI and to respond to changes in human-player tactics. Dynamic scripting is a reinforcement learning approach to adaptive game AI that learns, during gameplay, which game tactics an opponent should select to play effectively. In previous work, the tactics used by dynamic scripting were designed manually. We introduce the evolutionary state-based tactics generator (ESTG), which uses an evolutionary algorithm to generate tactics automatically. Experimental results show that ESTG improves dynamic scripting's performance in a real-time strategy game. We conclude that high-quality domain knowledge can be automatically generated for strong adaptive game AI opponents. Game developers can bene it from applying ESTG, as it considerably reduces the time and effort needed to create adaptive game AI.