Dialogue Systems (DS) are increasingly in demand for automating tasks through natural language interactions. However, the core techniques for user comprehension in DS depend heavily on large amounts of labeled data, limiting their applicability in data-scarce environments common to many companies. This paper identifies best practices for data-efficient development and cost-effective deployment of DS in real-world application scenarios. We evaluate whether fine-tuning a medium-sized Large Language Model (LLM) for joint Intent Classification (IC) and Slot Filling (SF), with moderate hardware resource requirements still affordable by SMEs, can achieve competitive performance using less data compared to current state-of-the-art models. Experiments on the Spanish and English portions of the MASSIVE corpus demonstrate that the Llama-3-8B-Instruct model fine-tuned with only 10% of the data outperforms the JointBERT architecture and GPT-4o in a zero-shot prompting setup in monolingual settings. In cross-lingual scenarios, Llama-3-8B-Instruct drastically outperforms multilingual JointBERT demonstrating a vastly superior performance when fine-tuned in a language and evaluated in the other.

Virtual Reality (VR) training provides safe, cost-effective engagement with lifelike scenarios but lacks intuitive communication between users and the virtual environment. This study investigates the use of Large Language Models (LLMs) as conversational tutors in VR health and safety training, examining the impact of game context and state variables on LLM-generated answers in zero- and few-shot settings. Results demonstrate that incorporating both game context and state information significantly improves answer accuracy, with human evaluations showing gains of up to 0.26 points in zero-shot and 0.18 points in few-shot settings on a 0-1 scale.

Code-switching (CS) is a very common phenomenon in regions with various co-existing languages. Since CS is such a frequent habit in informal communications, both spoken and written, it also arises naturally in Human-Machine Interactions. Therefore, in order for natural language understanding (NLU) not to be degraded, CS must be taken into account when developing chatbots. The co-existence of multiple languages in a single NLU model has become feasible with multilingual language representation models such as mBERT. In this paper, the efficacy of zero-shot cross-lingual transfer learning with mBERT for NLU is evaluated on a Basque-Spanish CS chatbot corpus, comparing the performance of NLU models trained using in-domain chatbot utterances in Basque and/or Spanish without CS. The results obtained indicate that training joint multi-intent classification and entity recognition models on both languages simultaneously achieves best performance, better capturing the CS patterns.

The main objective of this work is the elaboration and public release of BaSCo, the first corpus with annotated linguistic resources encompassing Basque-Spanish code-switching. The mixture of Basque and Spanish languages within the same utterance is popularly referred to as Euskañol, a widespread phenomenon among bilingual speakers in the Basque Country. Thus, this corpus has been created to meet the demand of annotated linguistic resources in Euskañol in research areas such as multilingual dialogue systems. The presented resource is the result of translating to Euskañol a compilation of texts in Basque and Spanish that were used for training the Natural Language Understanding (NLU) models of several task-oriented bilingual chatbots. Those chatbots were meant to answer specific questions associated with the administration, fiscal, and transport domains. In addition, they had the transverse potential to answer to greetings, requests for help, and chit-chat questions asked to chatbots. BaSCo is a compendium of 1377 tagged utterances with every sample annotated at three levels: (i) NLU semantic labels, considering intents and entities, (ii) code-switching proportion, and (iii) domain of origin.