Semantic Enrichment of Scientific Documents with Semantic Lenses – Developing methodologies, tools and prototypes for their concrete use

Con questa dissertazione di tesi miro ad illustrare i risultati della mia ricerca nel campo del Semantic Publishing, consistenti nello sviluppo di un insieme di metodologie, strumenti e prototipi, uniti allo studio di un caso d‟uso concreto, finalizzati all‟applicazione ed alla focalizzazione di Lenti Semantiche (Semantic Lenses).

NLG-Metricverse: An End-to-End Library for Evaluating Natural Language Generation

Driven by recent deep learning breakthroughs, natural language generation (NLG) models have been at the center of steady progress in the last few years. However, since our ability to generate human-indistinguishable artificial text lags behind our capacity to assess it, it is paramount to develop and apply even better automatic evaluation metrics. To facilitate researchers to judge the effectiveness of their models broadly, we suggest NLG-Metricverse—an end-to-end open-source library for NLG evaluation based on Python. This framework provides a living collection of NLG metrics in a unified and easy- to-use environment, supplying tools to efficiently apply, analyze, compare, and visualize them. This includes (i) the extensive support of heterogeneous automatic metrics with n-arity management, (ii) the meta-evaluation upon individual performance, metric-metric and metric-human correlations, (iii) graphical interpretations for helping humans better gain score intuitions, (iv) formal categorization and convenient documentation to accelerate metrics understanding. NLG-Metricverse aims to increase the comparability and replicability of NLG research, hopefully stimulating new contributions in the area.

Sviluppo di Retrieval-based Chatbot per l'Italiano con Transformer

L’Intelligenza Artificiale negli ultimi anni sta plasmando il futuro dell’umanità in quasi tutti i settori. È già il motore principale di diverse tecnologie emergenti come i big data, la robotica e l’IoT e continuerà ad agire come innovatore tecnologico nel futuro prossimo. Le recenti scoperte e migliorie sia nel campo dell’hardware che in quello matematico hanno migliorato l’efficienza e ridotto i tempi di esecuzione dei software. È in questo contesto che sta evolvendo anche il Natural Language Processing (NLP), un ramo dell’Intelligenza Artificiale che studia il modo in cui fornire ai computer l'abilità di comprendere un testo scritto o parlato allo stesso modo in cui lo farebbe un essere umano. Le ambiguità che distinguono la lingua naturale dalle altre rendono ardui gli studi in questo settore. Molti dei recenti sviluppi algoritmici su NLP si basano su tecnologie inventate decenni fa. La ricerca in questo settore è quindi in continua evoluzione. Questa tesi si pone l'obiettivo di sviluppare la logica di una chatbot help-desk per un'azienda privata. Lo scopo è, sottoposta una domanda da parte di un utente, restituire la risposta associata presente in una collezione domande-risposte. Il problema che questa tesi affronta è sviluppare un modello di NLP in grado di comprendere il significato semantico delle domande in input, poiché esse possono essere formulate in molteplici modi, preservando il contenuto semantico a discapito della sintassi. A causa delle ridotte dimensioni del dataset italiano proprietario su cui testare il modello chatbot, sono state eseguite molteplici sperimentazioni su un ulteriore dataset italiano con task affine. Attraverso diversi approcci di addestramento, tra cui apprendimento metrico, sono state raggiunte alte accuratezze sulle più comuni metriche di valutazione, confermando le capacità del modello proposto e sviluppato.

Vocabulary Transfer and Knowledge Distillation for Language Model Compression

Natural Language Processing (NLP) has seen tremendous improvements over the last few years. Transformer architectures achieved impressive results in almost any NLP task, such as Text Classification, Machine Translation, and Language Generation. As time went by, transformers continued to improve thanks to larger corpora and bigger networks, reaching hundreds of billions of parameters. Training and deploying such large models has become prohibitively expensive, such that only big high tech companies can afford to train those models. Therefore, a lot of research has been dedicated to reducing a model’s size. In this thesis, we investigate the effects of Vocabulary Transfer and Knowledge Distillation for compressing large Language Models. The goal is to combine these two methodologies to further compress models without significant loss of performance. In particular, we designed different combination strategies and conducted a series of experiments on different vertical domains (medical, legal, news) and downstream tasks (Text Classification and Named Entity Recognition). Four different methods involving Vocabulary Transfer (VIPI) with and without a Masked Language Modelling (MLM) step and with and without Knowledge Distillation are compared against a baseline that assigns random vectors to new elements of the vocabulary. Results indicate that VIPI effectively transfers information of the original vocabulary and that MLM is beneficial. It is also noted that both vocabulary transfer and knowledge distillation are orthogonal to one another and may be applied jointly. The application of knowledge distillation first before subsequently applying vocabulary transfer is recommended. Finally, model performance due to vocabulary transfer does not always show a consistent trend as the vocabulary size is reduced. Hence, the choice of vocabulary size should be empirically selected by evaluation on the downstream task similar to hyperparameter tuning.

Bio-QA-GNN: Reasoning with Language Models and Knowledge Graphs for Interpretable Biomedical Question Answering

Nowadays the idea of injecting world or domain-specific structured knowledge into pre-trained language models (PLMs) is becoming an increasingly popular approach for solving problems such as biases, hallucinations, huge architectural sizes, and explainability lack—critical for real-world natural language processing applications in sensitive fields like bioinformatics. One recent work that has garnered much attention in Neuro-symbolic AI is QA-GNN, an end-to-end model for multiple-choice open-domain question answering (MCOQA) tasks via interpretable text-graph reasoning. Unlike previous publications, QA-GNN mutually informs PLMs and graph neural networks (GNNs) on top of relevant facts retrieved from knowledge graphs (KGs). However, taking a more holistic view, existing PLM+KG contributions mainly consider commonsense benchmarks and ignore or shallowly analyze performances on biomedical datasets. This thesis start from a propose of a deep investigation of QA-GNN for biomedicine, comparing existing or brand-new PLMs, KGs, edge-aware GNNs, preprocessing techniques, and initialization strategies. By combining the insights emerged in DISI's research, we introduce Bio-QA-GNN that include a KG. Working with this part has led to an improvement in state-of-the-art of MCOQA model on biomedical/clinical text, largely outperforming the original one (+3.63\% accuracy on MedQA). Our findings also contribute to a better understanding of the explanation degree allowed by joint text-graph reasoning architectures and their effectiveness on different medical subjects and reasoning types. Codes, models, datasets, and demos to reproduce the results are freely available at: \url{https://github.com/disi-unibo-nlp/bio-qagnn}.

Frame-driven Extraction of Linked Data and Ontologies from Text

Ontology design and population -core aspects of semantic technologies- re- cently have become fields of great interest due to the increasing need of domain-specific knowledge bases that can boost the use of Semantic Web. For building such knowledge resources, the state of the art tools for ontology design require a lot of human work. Producing meaningful schemas and populating them with domain-specific data is in fact a very difficult and time-consuming task. Even more if the task consists in modelling knowledge at a web scale. The primary aim of this work is to investigate a novel and flexible method- ology for automatically learning ontology from textual data, lightening the human workload required for conceptualizing domain-specific knowledge and populating an extracted schema with real data, speeding up the whole ontology production process. Here computational linguistics plays a fundamental role, from automati- cally identifying facts from natural language and extracting frame of relations among recognized entities, to producing linked data with which extending existing knowledge bases or creating new ones. In the state of the art, automatic ontology learning systems are mainly based on plain-pipelined linguistics classifiers performing tasks such as Named Entity recognition, Entity resolution, Taxonomy and Relation extraction [11]. These approaches present some weaknesses, specially in capturing struc- tures through which the meaning of complex concepts is expressed [24]. Humans, in fact, tend to organize knowledge in well-defined patterns, which include participant entities and meaningful relations linking entities with each other. In literature, these structures have been called Semantic Frames by Fill- 6 Introduction more [20], or more recently as Knowledge Patterns [23]. Some NLP studies has recently shown the possibility of performing more accurate deep parsing with the ability of logically understanding the structure of discourse [7]. In this work, some of these technologies have been investigated and em- ployed to produce accurate ontology schemas. The long-term goal is to collect large amounts of semantically structured information from the web of crowds, through an automated process, in order to identify and investigate the cognitive patterns used by human to organize their knowledge.

Deep learning for computer vision: a comparison between convolutional neural networks and hierarchical temporal memories on object recognition tasks

In recent years, Deep Learning techniques have shown to perform well on a large variety of problems both in Computer Vision and Natural Language Processing, reaching and often surpassing the state of the art on many tasks. The rise of deep learning is also revolutionizing the entire field of Machine Learning and Pattern Recognition pushing forward the concepts of automatic feature extraction and unsupervised learning in general. However, despite the strong success both in science and business, deep learning has its own limitations. It is often questioned if such techniques are only some kind of brute-force statistical approaches and if they can only work in the context of High Performance Computing with tons of data. Another important question is whether they are really biologically inspired, as claimed in certain cases, and if they can scale well in terms of "intelligence". The dissertation is focused on trying to answer these key questions in the context of Computer Vision and, in particular, Object Recognition, a task that has been heavily revolutionized by recent advances in the field. Practically speaking, these answers are based on an exhaustive comparison between two, very different, deep learning techniques on the aforementioned task: Convolutional Neural Network (CNN) and Hierarchical Temporal memory (HTM). They stand for two different approaches and points of view within the big hat of deep learning and are the best choices to understand and point out strengths and weaknesses of each of them. CNN is considered one of the most classic and powerful supervised methods used today in machine learning and pattern recognition, especially in object recognition. CNNs are well received and accepted by the scientific community and are already deployed in large corporation like Google and Facebook for solving face recognition and image auto-tagging problems. HTM, on the other hand, is known as a new emerging paradigm and a new meanly-unsupervised method, that is more biologically inspired. It tries to gain more insights from the computational neuroscience community in order to incorporate concepts like time, context and attention during the learning process which are typical of the human brain. In the end, the thesis is supposed to prove that in certain cases, with a lower quantity of data, HTM can outperform CNN.

Lightweight Internet Traffic Classification - A Subject Based Solution with Word Embeddings

Internet traffic classification is a relevant and mature research field, anyway of growing importance and with still open technical challenges, also due to the pervasive presence of Internet-connected devices into everyday life. We claim the need for innovative traffic classification solutions capable of being lightweight, of adopting a domain-based approach, of not only concentrating on application-level protocol categorization but also classifying Internet traffic by subject. To this purpose, this paper originally proposes a classification solution that leverages domain name information extracted from IPFIX summaries, DNS logs, and DHCP leases, with the possibility to be applied to any kind of traffic. Our proposed solution is based on an extension of Word2vec unsupervised learning techniques running on a specialized Apache Spark cluster. In particular, learning techniques are leveraged to generate word-embeddings from a mixed dataset composed by domain names and natural language corpuses in a lightweight way and with general applicability. The paper also reports lessons learnt from our implementation and deployment experience that demonstrates that our solution can process 5500 IPFIX summaries per second on an Apache Spark cluster with 1 slave instance in Amazon EC2 at a cost of $ 3860 year. Reported experimental results about Precision, Recall, F-Measure, Accuracy, and Cohen's Kappa show the feasibility and effectiveness of the proposal. The experiments prove that words contained in domain names do have a relation with the kind of traffic directed towards them, therefore using specifically trained word embeddings we are able to classify them in customizable categories. We also show that training word embeddings on larger natural language corpuses leads improvements in terms of precision up to 180%.

Predire la borsa valori con language model neurali

Nonostante lo scetticismo di molti studiosi circa la possibilità di prevedere l'andamento della borsa valori, esistono svariate teorie ipotizzanti la possibilità di utilizzare le informazioni conosciute per predirne i movimenti futuri. L’avvento dell’intelligenza artificiale nella seconda parte dello scorso secolo ha permesso di ottenere risultati rivoluzionari in svariati ambiti, tanto che oggi tale disciplina trova ampio impiego nella nostra vita quotidiana in molteplici forme. In particolare, grazie al machine learning, è stato possibile sviluppare sistemi intelligenti che apprendono grazie ai dati, riuscendo a modellare problemi complessi. Visto il successo di questi sistemi, essi sono stati applicati anche all’arduo compito di predire la borsa valori, dapprima utilizzando i dati storici finanziari della borsa come fonte di conoscenza, e poi, con la messa a punto di tecniche di elaborazione del linguaggio naturale umano (NLP), anche utilizzando dati in linguaggio naturale, come il testo di notizie finanziarie o l’opinione degli investitori. Questo elaborato ha l’obiettivo di fornire una panoramica sull’utilizzo delle tecniche di machine learning nel campo della predizione del mercato azionario, partendo dalle tecniche più elementari per arrivare ai complessi modelli neurali che oggi rappresentano lo stato dell’arte. Vengono inoltre formalizzati il funzionamento e le tecniche che si utilizzano per addestrare e valutare i modelli di machine learning, per poi effettuare un esperimento in cui a partire da dati finanziari e soprattutto testuali si tenterà di predire correttamente la variazione del valore dell’indice di borsa S&P 500 utilizzando un language model basato su una rete neurale.

Quality and Aspect based Argument Generation

Natural Language Processing has always been one of the most popular topics in Artificial Intelligence. Argument-related research in NLP, such as argument detection, argument mining and argument generation, has been popular, especially in recent years. In our daily lives, we use arguments to express ourselves. The quality of arguments heavily impacts the effectiveness of our communications with others. In professional fields, such as legislation and academic areas, arguments of good quality play an even more critical role. Therefore, argument generation with good quality is a challenging research task that is also of great importance in NLP. The aim of this work is to investigate the automatic generation of arguments with good quality, according to the given topic, stance and aspect (control codes). To achieve this goal, a module based on BERT [17] which could judge an argument's quality is constructed. This module is used to assess the quality of the generated arguments. Another module based on GPT-2 [19] is implemented to generate arguments. Stances and aspects are also used as guidance when generating arguments. After combining all these models and techniques, the ranks of the generated arguments could be acquired to evaluate the final performance. This dissertation describes the architecture and experimental setup, analyzes the results of our experimentation, and discusses future directions.

SynBA: A contextualized Synonim-Based adversarial Attack for text classification

With the advent of high-performance computing devices, deep neural networks have gained a lot of popularity in solving many Natural Language Processing tasks. However, they are also vulnerable to adversarial attacks, which are able to modify the input text in order to mislead the target model. Adversarial attacks are a serious threat to the security of deep neural networks, and they can be used to craft adversarial examples that steer the model towards a wrong decision. In this dissertation, we propose SynBA, a novel contextualized synonym-based adversarial attack for text classification. SynBA is based on the idea of replacing words in the input text with their synonyms, which are selected according to the context of the sentence. We show that SynBA successfully generates adversarial examples that are able to fool the target model with a high success rate. We demonstrate three advantages of this proposed approach: (1) effective - it outperforms state-of-the-art attacks by semantic similarity and perturbation rate, (2) utility-preserving - it preserves semantic content, grammaticality, and correct types classified by humans, and (3) efficient - it performs attacks faster than other methods.

Hierarchical Multi-Label Text Classification in a Low-Resource Setting

In this thesis we address a multi-label hierarchical text classification problem in a low-resource setting and explore different approaches to identify the best one for our case. The goal is to train a model that classifies English school exercises according to a hierarchical taxonomy with few labeled data. The experiments made in this work employ different machine learning models and text representation techniques: CatBoost with tf-idf features, classifiers based on pre-trained models (mBERT, LASER), and SetFit, a framework for few-shot text classification. SetFit proved to be the most promising approach, achieving better performance when during training only a few labeled examples per class are available. However, this thesis does not consider all the hierarchical taxonomy, but only the first two levels: to address classification with the classes at the third level further experiments should be carried out, exploring methods for zero-shot text classification, data augmentation, and strategies to exploit the hierarchical structure of the taxonomy during training.

Vantaggi e limiti dell'integrazione tra sistemi di traduzione assistita e sistemi di traduzione automatica

Computer-assisted translation (or computer-aided translation or CAT) is a form of language translation in which a human translator uses computer software in order to facilitate the translation process. Machine translation (MT) is the automated process by which a computerized system produces a translated text or speech from one natural language to another. Both of them are leading and promising technologies in the translation industry; it therefore seems important that translation students and professional translators become familiar with this relatively new types of technology. Whether used together, not only might these two different types of systems reduce translation time, but also lead to a further improvement in the field of translation technologies. The dissertation consists of four chapters. The first one surveys the chronological development of MT and CAT tools, the emergence of pre-editing, post-editing and controlled language and the very last frontiers in this sector. The second one provide a general overview on the four main CAT tools that are used nowadays and tested hereto. The third chapter is dedicated to the experimentations that have been conducted in order to analyze and evaluate the performance of the four integrated systems that are the core subject of this dissertation. Finally, the fourth chapter deals with the issue of terminological equivalence in interlinguistic translation. The purpose of this dissertation is not to provide an objective and definitive solution to the complex issues that arise at any time in the field of translation technologies, this aim being well away from being achieved, but to supply information about the limits and potentiality that are typical of those instruments which are now essential to any professional translator.

BabelNet e il problema della Word Sense Disambiguation

La Word Sense Disambiguation è un problema informatico appartenente al campo di studi del Natural Language Processing, che consiste nel determinare il senso di una parola a seconda del contesto in cui essa viene utilizzata. Se un processo del genere può apparire banale per un essere umano, può risultare d'altra parte straordinariamente complicato se si cerca di codificarlo in una serie di istruzioni esguibili da una macchina. Il primo e principale problema necessario da affrontare per farlo è quello della conoscenza: per operare una disambiguazione sui termini di un testo, un computer deve poter attingere da un lessico che sia il più possibile coerente con quello di un essere umano. Sebbene esistano altri modi di agire in questo caso, quello di creare una fonte di conoscenza machine-readable è certamente il metodo che permette di affrontare il problema in maniera più diretta. Nel corso di questa tesi si cercherà, come prima cosa, di spiegare in cosa consiste la Word Sense Disambiguation, tramite una descrizione breve ma il più possibile dettagliata del problema. Nel capitolo 1 esso viene presentato partendo da alcuni cenni storici, per poi passare alla descrizione dei componenti fondamentali da tenere in considerazione durante il lavoro. Verranno illustrati concetti ripresi in seguito, che spaziano dalla normalizzazione del testo in input fino al riassunto dei metodi di classificazione comunemente usati in questo campo. Il capitolo 2 è invece dedicato alla descrizione di BabelNet, una risorsa lessico-semantica multilingua di recente costruzione nata all'Università La Sapienza di Roma. Verranno innanzitutto descritte le due fonti da cui BabelNet attinge la propria conoscenza, WordNet e Wikipedia. In seguito saranno illustrati i passi della sua creazione, dal mapping tra le due risorse base fino alla definizione di tutte le relazioni che legano gli insiemi di termini all'interno del lessico. Infine viene proposta una serie di esperimenti che mira a mettere BabelNet su un banco di prova, prima per verificare la consistenza del suo metodo di costruzione, poi per confrontarla, in termini di prestazioni, con altri sistemi allo stato dell'arte sottoponendola a diversi task estrapolati dai SemEval, eventi internazionali dedicati alla valutazione dei problemi WSD, che definiscono di fatto gli standard di questo campo. Nel capitolo finale vengono sviluppate alcune considerazioni sulla disambiguazione, introdotte da un elenco dei principali campi applicativi del problema. Vengono in questa sede delineati i possibili sviluppi futuri della ricerca, ma anche i problemi noti e le strade recentemente intraprese per cercare di portare le prestazioni della Word Sense Disambiguation oltre i limiti finora definiti.

Named Entity Extraction: la piattaforma Gate/Annie

In questo lavoro si introducono i concetti di base di Natural Language Processing, soffermandosi su Information Extraction e analizzandone gli ambiti applicativi, le attività principali e la differenza rispetto a Information Retrieval. Successivamente si analizza il processo di Named Entity Recognition, focalizzando l’attenzione sulle principali problematiche di annotazione di testi e sui metodi per la valutazione della qualità dell’estrazione di entità. Infine si fornisce una panoramica della piattaforma software open-source di language processing GATE/ANNIE, descrivendone l’architettura e i suoi componenti principali, con approfondimenti sugli strumenti che GATE offre per l'approccio rule-based a Named Entity Recognition.