Tenders are powerful means of investment of public funds and represent a strategic development resource. Despite the efforts made so far by governments at national and international levels to digitalise documents related to the Public Administration sector, most of the information is still available in an unstructured format only. With the aim of bridging this gap, we present OIE4PA, our latest study on extracting and classifying relations from tenders of the Public Administration. Our work focuses on the Italian language, where the availability of linguistic resources to perform Natural Language Processing tasks is considerably limited. Nevertheless, OIE4PA adopts a multilingual approach so it can be applied to several languages by providing appropriate training data. Rather than purely training a classifier on a portion of the extracted relations, the backbone idea of our learning strategy is to put a supervised method based on self-training to the proof and to assess whether or not it improves the performance of the classifier. For evaluation purposes, we built a dataset composed of 2,000 triples which have been manually annotated by two human experts. The in-vitro evaluation shows that OIE4PA achieves a MacroF $$_1$$1 equal to 0.89 and a 91 $$\%$$% accuracy. In addition, OIE4PA was used as the pillar of a prototype search engine, which has been evaluated through an in-vivo experiment with positive feedback from 32 final users, obtaining a SUS score equal to 83.98.

Scientists in the marine domain process satellite images in order to extract information that can be used for monitoring, understanding, and forecasting of marine phenomena, such as turbidity, algal blooms and oil spills. The growing need for effective retrieval of related information has motivated the adoption of semantically aware strategies on satellite images with different spatio-temporal and spectral characteristics. A big issue of these approaches is the lack of coincidence between the information that can be extracted from the visual data and the interpretation that the same data have for a user in a given situation. In this work, we bridge this semantic gap by connecting the quantitative elements of the Earth Observation satellite images with the qualitative information, modelling this knowledge in a marine phenomena ontology and developing a question answering mechanism based on natural language that enables the retrieval of the most appropriate data for each user’s needs. The main objective of the presented methodology is to realize the content-based search of Earth Observation images related to the marine application domain on an application-specific basis that can answer queries such as “Find oil spills that occurred this year in the Adriatic Sea”.

Evaluating Artificial Players for the Language Game “La Ghigliottina” (Ghigliottin-AI) task is one of the tasks organized in the context of the 2020 EVALITA edition, a periodic evaluation campaign of Natural Language Processing (NLP) and speech tools for the Italian language. Ghigliottin-AI participants are asked to build an artificial player able to solve “La Ghigliottina”, namely the final game of an Italian TV show called “L’Eredità”. The game involves a single player who is given a set of five words unrelated to each other, but related with a sixth word that represents the solution to the game. Fourteen teams registered to Ghigliottin-AI. Nevertheless, only two teams submitted their run. In order to evaluate the submitted systems, we rely on an API base methodology, via a Remote Evaluation Server (RES). In this report we describe the Ghigliottin-AI task, the data, the evaluation and we discuss results.

Word Sense Disambiguation (WSD) is a historical task in computational linguistics that has received much attention over the years. However, with the advent of Large Language Models (LLMs), interest in this task (in its classical definition) has decreased. In this study, we evaluate the performance of various LLMs on the WSD task. We extend a previous benchmark (XL-WSD) to re-design two subtasks suitable for LLM: 1) given a word in a sentence, the LLM must generate the correct definition; 2) given a word in a sentence and a set of predefined meanings, the LLM must select the correct one. The extended benchmark is built using the XL-WSD and BabelNet. The results indicate that LLMs perform well in zero-shot learning but cannot surpass current state-of-the-art methods. However, a fine-tuned model with a medium number of parameters outperforms all other models, including the state-of-the-art.