Title: Neural Network Based Named Entity Recognition Author: Jana Straková Institute: Institute of Formal and Applied Linguistics Supervisor of the doctoral thesis: prof. RNDr. Jan Hajič, Dr., Institute of Formal and Applied Linguistics Abstract: Czech named entity recognition (the task of automatic identification and classification of proper names in text, such as names of people, locations and organizations) has become a well-established field since the publication of the Czech Named Entity Corpus (CNEC). This doctoral thesis presents the author's research of named entity recognition, mainly in the Czech language. It presents work and research carried out during CNEC publication and its evaluation. It fur- ther envelops the author's research results, which improved Czech state-of-the-art results in named entity recognition in recent years, with special focus on artificial neural network based solutions. Starting with a simple feed-forward neural net- work with softmax output layer, with a standard set of classification features for the task, the thesis presents methodology and results, which were later used in open-source software solution for named entity recognition, NameTag. The thesis finalizes with a recurrent neural network based recognizer with word embeddings and character-level word embeddings,...

Since the proposal of hypertext by Tim Berners-Lee to his employer CERN on March 12, 1989 the World Wide Web has grown to more than one billion Web pages and still grows. With the later proposed Semantic Web vision,Berners-Lee et al. suggested an extension of the existing (Document) Web to allow better reuse, sharing and understanding of data. Both the Document Web and the Web of Data (which is the current implementation of the Semantic Web) grow continuously. This is a mixed blessing, as the two forms of the Web grow concurrently and most commonly contain different pieces of information. Modern information systems must thus bridge a Semantic Gap to allow a holistic and unified access to information about a particular information independent of the representation of the data. One way to bridge the gap between the two forms of the Web is the extraction of structured data, i.e., RDF, from the growing amount of unstructured and semi-structured information (e.g., tables and XML) on the Document Web. Note, that unstructured data stands for any type of textual information like news, blogs or tweets. While extracting structured data from unstructured data allows the development of powerful information system, it requires high-quality and scalable knowledge extraction frameworks to lead to useful results. The dire need for such approaches has led to the development of a multitude of annotation frameworks and tools. However, most of these approaches are not evaluated on the same datasets or using the same measures. The resulting Evaluation Gap needs to be tackled by a concise evaluation framework to foster fine-grained and uniform evaluations of annotation tools and frameworks over any knowledge bases. Moreover, with the constant growth of data and the ongoing decentralization of knowledge, intuitive ways for non-experts to access the generated data are required. Humans adapted their search behavior to current Web data by access paradigms such as keyword search so as to retrieve high-quality results. Hence, most Web users only expect Web documents in return. However, humans think and most commonly express their information needs in their natural language rather than using keyword phrases. Answering complex information needs often requires the combination of knowledge from various, differently structured data sources. Thus, we observe an Information Gap between natural-language questions and current keyword-based search paradigms, which in addition do not make use of the available structured and unstructured data sources. Question Answering (QA) systems provide an easy and efficient way to bridge this gap by allowing to query data via natural language, thus reducing (1) a possible loss of precision and (2) potential loss of time while reformulating the search intention to transform it into a machine-readable way. Furthermore, QA systems enable answering natural language queries with concise results instead of links to verbose Web documents. Additionally, they allow as well as encourage the access to and the combination of knowledge from heterogeneous knowledge bases (KBs) within one answer. Consequently, three main research gaps are considered and addressed in this work: First, addressing the Semantic Gap between the unstructured Document Web and the Semantic Gap requires the development of scalable and accurate approaches for the extraction of structured data in RDF. This research challenge is addressed by several approaches within this thesis. This thesis presents CETUS, an approach for recognizing entity types to populate RDF KBs. Furthermore, our knowledge base-agnostic disambiguation framework AGDISTIS can efficiently detect the correct URIs for a given set of named entities. Additionally, we introduce REX, a Web-scale framework for RDF extraction from semi-structured (i.e., templated) websites which makes use of the semantics of the reference knowledge based to check the extracted data. The ongoing research on closing the Semantic Gap has already yielded a large number of annotation tools and frameworks. However, these approaches are currently still hard to compare since the published evaluation results are calculated on diverse datasets and evaluated based on different measures. On the other hand, the issue of comparability of results is not to be regarded as being intrinsic to the annotation task. Indeed, it is now well established that scientists spend between 60% and 80% of their time preparing data for experiments. Data preparation being such a tedious problem in the annotation domain is mostly due to the different formats of the gold standards as well as the different data representations across reference datasets. We tackle the resulting Evaluation Gap in two ways: First, we introduce a collection of three novel datasets, dubbed N3, to leverage the possibility of optimizing NER and NED algorithms via Linked Data and to ensure a maximal interoperability to overcome the need for corpus-specific parsers. Second, we present GERBIL, an evaluation framework for semantic entity annotation. The rationale behind our framework is to provide developers, end users and researchers with easy-to-use interfaces that allow for the agile, fine-grained and uniform evaluation of annotation tools and frameworks on multiple datasets. The decentral architecture behind the Web has led to pieces of information being distributed across data sources with varying structure. Moreover, the increasing the demand for natural-language interfaces as depicted by current mobile applications requires systems to deeply understand the underlying user information need. In conclusion, the natural language interface for asking questions requires a hybrid approach to data usage, i.e., simultaneously performing a search on full-texts and semantic knowledge bases. To close the Information Gap, this thesis presents HAWK, a novel entity search approach developed for hybrid QA based on combining structured RDF and unstructured full-text data sources.

Natural language processing (NLP) is a vibrant area of research with many practical applications today like sentiment analyses, text labeling, questioning an- swering, machine translation and automatic text summarizing. At the moment, research is mainly focused on the English language, although many other lan- guages are trying to catch up. This work focuses on an area within NLP called information extraction, and more specifically on relation extraction, that is, to ex- tract relations between entities in a text. What this work aims at is to use machine learning techniques to build a Swedish language processing pipeline with part-of- speech tagging, dependency parsing, named entity recognition and coreference resolution to use as a base for later relation extraction from archival texts. The obvious difficulty lies in the scarcity of Swedish annotated datasets. For exam- ple, no large enough Swedish dataset for coreference resolution exists today. An important part of this work, therefore, is to create a Swedish coreference solver using distantly supervised machine learning, which means creating a Swedish dataset by applying an English coreference solver on an unannotated bilingual corpus, and then using a word-aligner to translate this machine-annotated En- glish dataset to a Swedish dataset, and then training a Swedish model on this dataset. Using Allen NLP:s end-to-end coreference resolution model, both for creating the Swedish dataset and training the Swedish model, this work achieves an F1-score of 0.5. For named entity recognition this work uses the Swedish BERT models released by the Royal Library of Sweden in February 2020 and achieves an overall F1-score of 0.95. To put all of these NLP-models within a single Lan- guage Processing Pipeline, Spacy is used as a unifying framework. / Natural Language Processing (NLP) är ett stort och aktuellt forskningsområde idag med många praktiska tillämpningar som sentimentanalys, textkategoriser- ing, maskinöversättning och automatisk textsummering. Forskningen är för när- varande mest inriktad på det engelska språket, men många andra språkområ- den försöker komma ikapp. Det här arbetet fokuserar på ett område inom NLP som kallas informationsextraktion, och mer specifikt relationsextrahering, det vill säga att extrahera relationer mellan namngivna entiteter i en text. Vad det här ar- betet försöker göra är att använda olika maskininlärningstekniker för att skapa en svensk Language Processing Pipeline bestående av part-of-speech tagging, de- pendency parsing, named entity recognition och coreference resolution. Denna pipeline är sedan tänkt att användas som en bas for senare relationsextrahering från svenskt arkivmaterial. Den uppenbara svårigheten med detta ligger i att det är ont om stora, annoterade svenska dataset. Till exempel så finns det inget till- räckligt stort svenskt dataset för coreference resolution. En stor del av detta arbete går därför ut på att skapa en svensk coreference solver genom att implementera distantly supervised machine learning, med vilket menas att använda en engelsk coreference solver på ett oannoterat engelskt-svenskt corpus, och sen använda en word-aligner för att översätta detta maskinannoterade engelska dataset till ett svenskt, och sen träna en svensk coreference solver på detta dataset. Det här arbetet använder Allen NLP:s end-to-end coreference solver, både för att skapa det svenska datasetet, och för att träna den svenska modellen, och uppnår en F1-score på 0.5. Vad gäller named entity recognition så använder det här arbetet Kungliga Bibliotekets BERT-modeller som bas, och uppnår genom detta en F1- score på 0.95. Spacy används som ett enande ramverk för att samla alla dessa NLP-komponenter inom en enda pipeline.

Open-domain question answering (QA) is an established NLP task which enables users to search for speciVc pieces of information in large collections of texts. Instead of using keyword-based queries and a standard information retrieval engine, QA systems allow the use of natural language questions and return the exact answer (or a list of plausible answers) with supporting snippets of text. In the past decade, open-domain QA research has been dominated by evaluation fora such as TREC and CLEF, where shallow techniques relying on information redundancy have achieved very good performance. However, this performance is generally limited to simple factoid and deVnition questions because the answer is usually explicitly present in the document collection. Current approaches are much less successful in Vnding implicit answers and are diXcult to adapt to more complex question types which are likely to be posed by users. In order to advance the Veld of QA, this thesis proposes a shift in focus from simple factoid questions to encyclopaedic questions: list questions composed of several constraints. These questions have more than one correct answer which usually cannot be extracted from one small snippet of text. To correctly interpret the question, systems need to combine classic knowledge-based approaches with advanced NLP techniques. To Vnd and extract answers, systems need to aggregate atomic facts from heterogeneous sources as opposed to simply relying on keyword-based similarity. Encyclopaedic questions promote QA systems which use basic reasoning, making them more robust and easier to extend with new types of constraints and new types of questions. A novel semantic architecture is proposed which represents a paradigm shift in open-domain QA system design, using semantic concepts and knowledge representation instead of words and information retrieval. The architecture consists of two phases, analysis – responsible for interpreting questions and Vnding answers, and feedback – responsible for interacting with the user. This architecture provides the basis for EQUAL, a semantic QA system developed as part of the thesis, which uses Wikipedia as a source of world knowledge and iii employs simple forms of open-domain inference to answer encyclopaedic questions. EQUAL combines the output of a syntactic parser with semantic information from Wikipedia to analyse questions. To address natural language ambiguity, the system builds several formal interpretations containing the constraints speciVed by the user and addresses each interpretation in parallel. To Vnd answers, the system then tests these constraints individually for each candidate answer, considering information from diUerent documents and/or sources. The correctness of an answer is not proved using a logical formalism, instead a conVdence-based measure is employed. This measure reWects the validation of constraints from raw natural language, automatically extracted entities, relations and available structured and semi-structured knowledge from Wikipedia and the Semantic Web. When searching for and validating answers, EQUAL uses the Wikipedia link graph to Vnd relevant information. This method achieves good precision and allows only pages of a certain type to be considered, but is aUected by the incompleteness of the existing markup targeted towards human readers. In order to address this, a semantic analysis module which disambiguates entities is developed to enrich Wikipedia articles with additional links to other pages. The module increases recall, enabling the system to rely more on the link structure of Wikipedia than on word-based similarity between pages. It also allows authoritative information from diUerent sources to be linked to the encyclopaedia, further enhancing the coverage of the system. The viability of the proposed approach was evaluated in an independent setting by participating in two competitions at CLEF 2008 and 2009. In both competitions, EQUAL outperformed standard textual QA systems as well as semi-automatic approaches. Having established a feasible way forward for the design of open-domain QA systems, future work will attempt to further improve performance to take advantage of recent advances in information extraction and knowledge representation, as well as by experimenting with formal reasoning and inferencing capabilities.

Le traitement automatique des données complexes et variées est un processus fondamental dans les applications d'extraction d'information. L'explosion combinatoire dans la composition des textes journalistiques et l'évolution du vocabulaire rend la tâche d'extraction d'indicateurs sémantiques, tel que les entités nommées, plus complexe par les approches symboliques. Les modèles stochastiques structurels tel que les champs conditionnels aléatoires (CRF) permettent d'optimiser des systèmes d'extraction d'information avec une importante capacité de généralisation. La première contribution de cette thèse est consacrée à la définition du contexte optimal pour l'extraction des régularités entre les mots et les annotations dans la tâche de reconnaissance d'entités nommées. Nous allons intégrer diverses informations dans le but d'enrichir les observations et améliorer la qualité de prédiction du système. Dans la deuxième partie nous allons proposer une nouvelle approche d'adaptation d'annotations entre deux protocoles différents. Le principe de cette dernière est basé sur l'enrichissement d'observations par des données générées par d'autres systèmes. Ces travaux seront expérimentés et validés sur les données de la campagne ESTER. D'autre part, nous allons proposer une approche de couplage entre le niveau signal représenté par un indice de la qualité de voisement et le niveau sémantique. L'objectif de cette étude est de trouver le lien entre le degré d'articulation du locuteur et l'importance de son discours / Recent researches in Information Extraction are designed to extract fixed types of information from data. Sequence annotation systems are developed to associate structured annotations to input data presented in sequential form. The named entity recognition (NER) task consists of identifying and classifying every word in a document into some predefined categories such as person name, locations, organizations, and dates. The complexity of the NER is largely related to the definition of the task and to the complexity of the relationships between words and the semantic associated. Our first contribution is devoted to solving the NER problem using discriminative graphical models. The proposed approach investigates the use of various contexts of the words to improve recognition. NER systems are fixed in accordance with a specific annotation protocol. Thus, new applications are developed for new protocols. The challenge is how we can adapt an annotation system which is performed for a specific application to other target application? We will propose in this work an adaptation approach of sequence labelling task based on annotation enrichment using conditional random fields (CRF). Experimental results show that the proposed approach outperform rules-based approach in NER task. Finally, we propose a multimodal approach of NER by integrating low level features as contextual information in radio broadcast news data. The objective of this study is to measure the correlation between the speaker voicing quality and the importance of his speech

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal.

[pt] Sistemas de recomendação têm sido amplamente utilizados pelos grandes portais na Web, em decorrência do aumento do volume de dados disponíveis na Web. Tais sistemas são basicamente utilizados para sugerir informações relevantes para os seus usuários. Esta dissertação apresenta um sistema de recomendação de segundo nível para auxiliar equipes de jornalistas de portais de notícias no processo de recomendação de notícias relacionadas para os usuários do portal. O sistema é chamado de segundo nível pois apresenta recomendações aos jornalistas para que, por sua vez, geram recomendações aos usuários do portal. O modelo seguido pelo sistema consiste na recomendação de notícias relacionadas com base em características extraídas do próprio texto da notícia original. As características extraídas permitem a criação de consultas contra um banco de dados de notícias anteriormente publicadas. O resultado de uma consulta é uma lista de notícias candidatas à recomendação, ordenada pela similaridade com a notícia original e pela data de publicação, que o editor da notícia original manualmente processa para gerar a lista final de notícias relacionadas. / [en] Recommendation systems are widely used by major Web portals due to the increase in the volume of data available on the Web. Such systems are basically used to suggest information relevant to their users. This dissertation presents a second-level recommendation system, which aims at assisting the team of journalists of a news Web portal in the process of recommending related news for the users of the Web portal. The system is called second level since it creates recommendations to the journalists Who, in turn, generate recommendations to the users. The system follows a model based on features extracted from the text itself. The extracted features permit creating queries against a news database. The query result is a list of candidate news, sorted by score and date of publication, which the news editor manually processes to generate the final list of related news.

The thesis presents named-entity recognition in Czech historical newspapers from Modern Access to Historical Sources Project. Our goal was to create a specific corpus and annotation manual for the project and evaluate neural networks methods for named-entity recognition within the task. We created the corpus using scanned Czech historical newspapers. The scanned pages were converted to digitize text by optical character recognition (OCR) method. The data were preprocessed by deleting some OCR errors. We also defined specific named entities types for our task and created an annotation manual with examples for the project. Based on that, we annotated the final corpus. To find the most suitable neural networks model for our task, we experimented with different neural networks architectures, namely long short-term memory (LSTM), bidirectional LSTM and CNN-BiLSTM models. Moreover, we experimented with randomly initialized word embeddings that were trained during the training process and pretrained word embeddings for contemporary Czech published as open source by fastText. We achieved the best result F1 score 0.444 using CNN-BiLSTM model and the pretrained word embeddings by fastText. We found out that we do not need to normalize spelling of our historical texts to get closer to contemporary language if we use the neural network model. We provided a qualitative analysis of observed linguistics phenomena as well. We found out that some word forms and pair of words which were not frequent in our training data set were miss-tagged or not tagged at all. Based on that, we can say that larger data sets could improve the results.

Information Extraction (IE) is the process of extracting structured and important pieces of information from a set of unstructured text documents in natural language. The final goal of structured information extraction is to populate a database and reach data effectively. Our study focuses on named entity recognition (NER) which is an important subtask of IE. NER is the task that deals with extraction of named entities like person, location, organization names, temporal expressions (date and time) and numerical expressions (money and percent). NER research on Turkish is known to be rare. There are rule-based, learning based and hybrid systems for NER on Turkish texts. Some of the learning approaches used for NER in Turkish are conditional random fields (CRF), rote learning, rule extraction and generalization. In this thesis, we propose a learning based named entity recognizer for Turkish texts which employs a modified version of Bayesian learning as the learning scheme. To the best of our knowledge, this is the first learning based system that uses Bayesian approach for NER in Turkish. Several features (like token length, capitalization, lexical meaning, etc.) are used in the system to see the effects of different features on NER process. We also propose hybrid system where the Bayesian learning-based system is utilized along with a rule-based recognition system. There are two different versions of the hybrid system. Output of rule-based recognizer is utilized in different phases in these versions. We observed increase in F-Measure values for both hybrid versions. When partial scoring is active, hybrid system reached 91.44% F-Measure value / where rule-based system result is 87.43% and learning-based system result is 88.41%. The hybrid system can be improved by utilizing rule-based and learning-based components differently in the future. Hybrid system can also be improved by using different learning approaches and combining them with existing hybrid system or forming the hybrid system with a completely new approach.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal