Companies do not exist in isolation. They are embedded in structural relationships with each other. Mapping a given company’s relationships with other companies in terms of competitors, subsidiaries, suppliers, and customers are key to understanding a company’s major risk factors and opportunities. Conventionally, obtaining and staying up to date with this key knowledge was achieved by reading financial news and reports by highly skilled manual labor like a financial analyst. However, with the development of Natural Language Processing (NLP) and graph databases, it is now possible to systematically extract and store structured information from unstructured data sources. The current go-to method to effectively extract information uses supervised machine learning models, which require a large amount of labeled training data. The data labeling process is usually time-consuming and hard to get in a domain-specific area. This project explores an approach to construct a company domain-specific Knowledge Graph (KG) that contains company-related entities and relationships from the U.S. Securities and Exchange Commission (SEC) 10-K filings by combining a pre-trained general NLP with rule-based patterns in Named Entity Recognition (NER) and Relation Extraction (RE). This approach eliminates the time-consuming data-labeling task in the statistical approach, and by evaluating ten 10-k filings, the model has the overall Recall of 53.6%, Precision of 75.7%, and the F1-score of 62.8%. The result shows it is possible to extract company information using the hybrid methods, which does not require a large amount of labeled training data. However, the project requires the time-consuming process of finding lexical patterns from sentences to extract company-related entities and relationships. / Företag existerar inte som isolerade organisationer. De är inbäddade i strukturella relationer med varandra. Att kartlägga ett visst företags relationer med andra företag när det gäller konkurrenter, dotterbolag, leverantörer och kunder är nyckeln till att förstå företagets huvudsakliga riskfaktorer och möjligheter. Det konventionella sättet att hålla sig uppdaterad med denna viktiga kunskap var genom att läsa ekonomiska nyheter och rapporter från högkvalificerad manuell arbetskraft som till exempel en finansanalytiker. Men med utvecklingen av ”Natural Language Processing” (NLP) och grafdatabaser är det nu möjligt att systematiskt extrahera och lagra strukturerad information från ostrukturerade datakällor. Den nuvarande metoden för att effektivt extrahera information använder övervakade maskininlärningsmodeller som kräver en stor mängd märkta träningsdata. Datamärkningsprocessen är vanligtvis tidskrävande och svår att få i ett domänspecifikt område. Detta projekt utforskar ett tillvägagångssätt för att konstruera en företagsdomänspecifikt ”Knowledge Graph” (KG) som innehåller företagsrelaterade enheter och relationer från SEC 10-K-arkivering genom att kombinera en i förväg tränad allmän NLP med regelbaserade mönster i ”Named Entity Recognition” (NER) och ”Relation Extraction” (RE). Detta tillvägagångssätt eliminerar den tidskrävande datamärkningsuppgiften i det statistiska tillvägagångssättet och genom att utvärdera tio SEC 10-K arkiv har modellen den totala återkallelsen på 53,6 %, precision på 75,7 % och F1-poängen på 62,8 %. Resultatet visar att det är möjligt att extrahera företagsinformation med hybridmetoderna, vilket inte kräver en stor mängd märkta träningsdata. Projektet kräver dock en tidskrävande process för att hitta lexikala mönster från meningar för att extrahera företagsrelaterade enheter och relationer.

Analyzing the data produced by next-generation sequencing technologies relies on access to information synthesized based on previous research findings. The volume of data available in the literature is growing rapidly, and it is becoming increasingly necessary for researchers to use AI or other statistics-based approaches in the analysis of their datasets. In this project, knowledge graphs are explored as a tool for providing access to contextual gene-information available in scientific literature. The explorative method described in this thesis is based on the implementation and comparison of two approaches for knowledge graph construction, a rule-based statistical as well as a neural-network and co-occurrence based approach, -based on specific literature contexts. The results are presented both in the form of a quantitative comparison between approaches as well as in the form of a qualitative expert evaluation of the quantitative result. The quantitative comparison suggested that contrasting knowledge graphs constructed based on different approaches can provide valuable information for the interpretation and contextualization of key genes. It also demonstrated the limitations of some approaches e.g. in terms of scalability as well as the volume and type of information that can be extracted. The result further suggested that metrics based on the overlap of nodes and edges, as well as metrics that leverage the global topology of graphs are valuable for representing and comparing contextual information between knowledge graphs. The result based on the qualitative expert evaluation demonstrated that literature-derived knowledge graphs of gene-information can be valuable tools for identifying research biases related to genes and also shed light on the challenges related to biological entity normalization in the context of knowledge graph development. In light of these findings, automatic knowledge-graph construction presents as a promising approach for improving access to contextual information about genes in scientific literature. / För att analysera de stora mängder data som produceras med hjälp av next-generation sequencing krävs det att forskare har tillgång till och kan sammanställa information från tidigare forskning. I takt med att mängden data som finns tillgänglig i den vetenskapliga litteraturen ökar, så ökar även behovet av att använda AI och andra statistiska metoder för att få tillgång till denna data i analysen. I detta projekt utforskas kunskapsgrafer som verktyg för att tillgängliggöra kontextuell geninformation i vetenskapliga artiklar. Den explorativa metod som beskrivs i detta projekt är baserad på implementationen och jämförelsen av två olika tekniker för kunskapsgrafgenerering, en regelbaserad-statistisk metod samt en metod baserad på neurala-nätverk och co-occurrence, baserade på specifika kontexter inom litteraturen. Resultatet presenteras både i form av en kvantitativ jämförelse mellan metoder samt genom en kvalitativ expertutvärdering baserad på det kvantitativa resultatet. Den kvantitativa jämförelsen antydde att jämförelsen mellan kunskapsgrafer genererade med hjälp av olika metoder kan bidra med värdefull information för tolkningen och kontextualiseringen av viktiga gener. Resultatet visade även på begränsningar hos vissa metoder, till exempel gällande skalbarhet samt den mängd och typ av information som kan extraheras. Men även att metrics baserade på överlappning av hörn och kanter, samt metrics som tar hänsyn till den globala topologin i grafer kan vara användbara i jämförelsen av, samt för att representera skillnader mellan biologiska kunskapsgrafer. Resultatet från den kvalitativa expertutvärderingen visade att kunskapsgrafer baserade på geninformation extraherad från vetenskapliga artiklar kan vara värdefulla verktyg för att identifiera forskningsbias gällande gener, samt framhävde viktiga utmaningar gällande normalisering av biologiska entiteter inom området kunskapsgrafsutveckling. Baserat på dessa fynd framstår automatisk kunskapsgrafsgenerering som ett lovande tillvägagångssätt för att förbättra tillgängligheten av kontextuell geninformation i vetenskaplig litteratur.

Seit Jahrhunderten wird menschliches Wissen in Form von natürlicher Sprache ausgetauscht und in Dokumenten schriftlich aufgezeichnet. In den letzten Jahren konnte man auf dem Gebiet der Lebenswissenschaften eine exponentielle Zunahme wissenschaftlicher Publikationen beobachten. Diese Dissertation untersucht die automatische Extraktion von Beziehungen zwischen Eigennamen. Innerhalb dieses Gebietes beschäftigt sich die Arbeit mit der Steigerung der Robustheit für die Relationsextraktion. Zunächst wird der Einsatz von Ensemble-Methoden anhand von Daten aus der "Drug-drug-interaction challenge 2013" evaluiert. Ensemble-Methoden erhöhen die Robustheit durch Aggregation unterschiedlicher Klassifikationssysteme zu einem Modell. Weiterhin wird in dieser Arbeit das Problem der Relationsextraktion auf Dokumenten mit unbekannten Texteigenschaften beschrieben. Es wird gezeigt, dass die Verwendung des halb-überwachten Lernverfahrens self training in solchen Fällen eine höhere Robustheit erzielt als die Nutzung eines Klassifikators, der lediglich auf einem manuell annotierten Korpus trainiert wurde. Zur Ermittlung der Robustheit wird das Verfahren des cross-learnings verwendet. Zuletzt wird die Verwendung von distant-supervision untersucht. Korpora, welche mit der distant-supervision-Methode erzeugt wurden, weisen ein inhärentes Rauschen auf und profitieren daher von robusten Relationsextraktionsverfahren. Es werden zwei verschiedene Methoden untersucht, die auf solchen Korpora trainiert werden. Beide Ansätze zeigen eine vergleichbare Leistung wie vollständig überwachte Klassifikatoren, welche mit dem cross-learning-Verfahren evaluiert wurden. Um die Nutzung von Ergebnissen der Informationsextraktion zu erleichtern, wurde die semantische Suchmaschine GeneView entwickelt. Anforderungen an die Rechenkapazität beim Erstellen von GeneView werden diskutiert und Anwendungen auf den von verschiedenen Text-Mining-Komponenten extrahierten Daten präsentiert. / For several centuries, a great wealth of human knowledge has been communicated by natural language, often recorded in written documents. In the life sciences, an exponential increase of scientific articles has been observed, hindering the effective and fast reconciliation of previous finding into current research projects. This thesis studies the automatic extraction of relationships between named entities. Within this topic, it focuses on increasing robustness for relationship extraction. First, we evaluate the use of ensemble methods to improve performance using data provided by the drug-drug-interaction challenge 2013. Ensemble methods aggregate several classifiers into one model, increasing robustness by reducing the risk of choosing an inappropriate single classifier. Second, this work discusses the problem of applying relationship extraction to documents with unknown text characteristics. Robustness of a text mining component is assessed by cross-learning, where a model is evaluated on a corpus different from the training corpus. We apply self-training, a semi-supervised learning technique, in order to increase cross-learning performance and show that it is more robust in comparison to a classifier trained on manually annotated text only. Third, we investigate the use of distant supervision to overcome the need of manually annotated training instances. Corpora derived by distant supervision are inherently noisy, thus benefiting from robust relationship extraction methods. We compare two different methods and show that both approaches achieve similar performance as fully supervised classifiers, evaluated in the cross-learning scenario. To facilitate the usage of information extraction results, including those developed within this thesis, we develop the semantic search engine GeneView. We discuss computational requirements to build this resource and present some applications utilizing the data extracted by different text-mining components.

The aim of this thesis is to determine to what extent aircraft-borne radar remote sensing can be used as the sole method for making recordings of the surface of the earth as a basis for compiling topographical and relief maps of mountainous areas. This is done using three test areas: the Edelsberg area in the Allgäu Alps and the Silvretta and Verwall Groups in the Central Alps. The basis for discussion is provided by examination of the interaction between the objects to be imaged and the radar signal, the sensor-specific characteristics thereby being taken into account. Following this some data processing and conditioning methods used for extracting information on the relief and surface coverage for preparation of cartographical products are presented. Analysis of the quality of the results shows that, measured against the requirements of mountain cartography, radar remote sensing is a practical and useful tool for making maps in Alpine regions. As the sole source of information, however, aircraft-borne radar remote sensing p roves to date to be inadequate for cartographical applications in high-mountain regions. / Ziel der vorliegenden Arbeit ist, festzustellen, inwieweit die flugzeuggetragene Radarfernerkundung als alleinige Erfassungsmethode der Erdoberfläche zur Erstellung von topographischen und reliefbeschreibenden Karten in Gebirgslandschaften dienen kann. Dies wird anhand von drei Testgebieten, dem Edelsberggebiet in den Allgäuer Alpen sowie der Silvretta- und Verwallgrupe in den Zentralalpen, untersucht. Die Betrachtung der Interaktion zwischen den abzubildenden Objekten und dem Radar-Signal unter Berücksichtigung der sensorspezifischen Charakteristika bildet dabei die Diskussionsgrundlage. Im weiteren werden Methoden zur Datenprozessierung und -aufbereitung vorgestellt, die eine Informa-tionsextraktion bezüglich des Reliefs und der Oberflächenbedeckung für die Erstellung kartographischer Produkte ermöglichen. Die Qualitätsanalyse der Ergebnisse zeigt, dass die Radarfernerkundung, gemessen an den Anforderungen der Gebirgskartographie, ein sinnvolles und nutzbringendes Werkzeug für die Kartenerstellung in alpinen Regionen ist. Für kartographische Anwendungen in Hochgebirgsregionen erweist sich die flugzeuggetragene Radarfernerkundung als einzige Informationsquelle bislang allerdings als nicht ausreichend.

Searching is a fundamental task to support research. Current search engines are keyword-based. Semantic technologies promise a next generation of semantic search engines, which will be able to answer questions. Current approaches either apply natural language processing to unstructured text or they assume the existence of structured statements over which they can reason. This work provides a system for combining the classical keyword-based search engines with semantic annotation. Conventional search results are annotated using a customized annotation algorithm, which takes the textual properties and requirements such as speed and scalability into account. The biomedical background knowledge consists of the GeneOntology and Medical Subject Headings and other related entities, e.g. proteins/gene names and person names. Together they provide the relevant semantic context for a search engine for the life sciences. We develop the system GoWeb for semantic web search and evaluate it using three benchmarks. It is shown that GoWeb is able to aid question answering with success rates up to 79%. Furthermore, the system also includes semantic hyperlinks that enable semantic browsing of the knowledge space. The semantic hyperlinks facilitate the use of the eScience infrastructure, even complex workflows of composed web services. To complement the web search of GoWeb, other data source and more specialized information needs are tested in different prototypes. This includes patents and intranet search. Semantic search is applicable for these usage scenarios, but the developed systems also show limits of the semantic approach. That is the size, applicability and completeness of the integrated ontologies, as well as technical issues of text-extraction and meta-data information gathering. Additionally, semantic indexing as an alternative approach to implement semantic search is implemented and evaluated with a question answering benchmark. A semantic index can help to answer questions and address some limitations of GoWeb. Still the maintenance and optimization of such an index is a challenge, whereas GoWeb provides a straightforward system.

Each day, millions of people ask questions and search for answers on the World Wide Web. Due to this, the Internet has grown to a world wide database of questions and answers, accessible to almost everyone. Since this database is so huge, it is hard to find out whether a question has been answered or even asked before. As a consequence, users are asking the same questions again and again, producing a vicious circle of new content which hides the important information. One platform for questions and answers are Web forums, also known as discussion boards. They present discussions as item streams where each item contains the contribution of one author. These contributions contain questions and answers in human readable form. People use search engines to search for information on such platforms. However, current search engines are neither optimized to highlight individual questions and answers nor to show which questions are asked often and which ones are already answered. In order to close this gap, this thesis introduces the \\emph{Effingo} system. The Effingo system is intended to extract forums from around the Web and find question and answer items. It also needs to link equal questions and aggregate associated answers. That way it is possible to find out whether a question has been asked before and whether it has already been answered. Based on these information it is possible to derive the most urgent questions from the system, to determine which ones are new and which ones are discussed and answered frequently. As a result, users are prevented from creating useless discussions, thus reducing the server load and information overload for further searches. The first research area explored by this thesis is forum data extraction. The results from this area are intended be used to create a database of forum posts as large as possible. Furthermore, it uses question-answer detection in order to find out which forum items are questions and which ones are answers and, finally, topic detection to aggregate questions on the same topic as well as discover duplicate answers. These areas are either extended by Effingo, using forum specific features such as the user graph, forum item relations and forum link structure, or adapted as a means to cope with the specific problems created by user generated content. Such problems arise from poorly written and very short texts as well as from hidden or distributed information.

Konversationerna i ett kundforum rör sig över olika ämnen och språket är inkonsekvent. Texterna uppfyller inte de krav som brukar ställas på material inför automatisk nyckelordsextraktion. Uppsatsens undersöker hur nyckelord automatiskt kan extraheras ur ett kundforum trots dessa svårigheter. Fokus i undersökningen ligger på tre aspekter av nyckelordsextraktion. Den första faktorn rör hur den etablerade nyckelordsextraktionsmetoden TF*IDF presterar jämfört med fyra metoder som skapas med hänsyn till materialets ovanliga struktur. Nästa faktor som testas är om olika sätt att räkna ordfrekvens påverkar resultatet. Den tredje faktorn är hur metoderna presterar om de endast använder inläggen, rubrikerna eller båda texttyperna i sina extraktioner. Icke-parametriska test användes för utvärdering av extraktionerna. Ett antal Friedmans test visar att metoderna i några fall skiljer sig åt gällande förmåga att identifiera relevanta nyckelord. I post-hoc-test mellan de högst presterande metoderna ses en av de nya metoderna i ett fall prestera signifikant bättre än de andra nya metoderna men inte bättre än TF*IDF. Ingen skillnad hittades mellan användning av olika texttyper eller sätt att räkna ordfrekvens. För framtida forskning rekommenderas reliabilitetstest av manuellt annoterade nyckelord. Ett större stickprov bör användas än det i aktuell studie och olika förslag ges för att förbättra rättning av extraherade nyckelord. / Conversations in a customer forum span across different topics and the language is inconsistent. The text type do not meet the demands for automatic keyword extraction. This essay examines how keywords can be automatically extracted despite these difficulties. Focus in the study are three areas of keyword extraction. The first factor regards how the established keyword extraction method TF*IDF performs compared to four methods created with the unusual material in mind. The next factor deals with different ways to calculate word frequency. The third factor regards if the methods use only posts, only titles, or both in their extractions. Non-parametric tests were conducted to evaluate the extractions. A number of Friedman's tests shows the methods in some cases differ in their ability to identify relevant keywords. In post-hoc tests performed between the highest performing methods, one of the new methods perform significantly better than the other new methods but not better than TF*IDF. No difference was found between the use of different text types or ways to calculate word frequency. For future research reliability test of manually annotated keywords is recommended. A larger sample size should be used than in the current study and further suggestions are given to improve the results of keyword extractions.

Each day, millions of people ask questions and search for answers on the World Wide Web. Due to this, the Internet has grown to a world wide database of questions and answers, accessible to almost everyone. Since this database is so huge, it is hard to find out whether a question has been answered or even asked before. As a consequence, users are asking the same questions again and again, producing a vicious circle of new content which hides the important information. One platform for questions and answers are Web forums, also known as discussion boards. They present discussions as item streams where each item contains the contribution of one author. These contributions contain questions and answers in human readable form. People use search engines to search for information on such platforms. However, current search engines are neither optimized to highlight individual questions and answers nor to show which questions are asked often and which ones are already answered. In order to close this gap, this thesis introduces the \\emph{Effingo} system. The Effingo system is intended to extract forums from around the Web and find question and answer items. It also needs to link equal questions and aggregate associated answers. That way it is possible to find out whether a question has been asked before and whether it has already been answered. Based on these information it is possible to derive the most urgent questions from the system, to determine which ones are new and which ones are discussed and answered frequently. As a result, users are prevented from creating useless discussions, thus reducing the server load and information overload for further searches. The first research area explored by this thesis is forum data extraction. The results from this area are intended be used to create a database of forum posts as large as possible. Furthermore, it uses question-answer detection in order to find out which forum items are questions and which ones are answers and, finally, topic detection to aggregate questions on the same topic as well as discover duplicate answers. These areas are either extended by Effingo, using forum specific features such as the user graph, forum item relations and forum link structure, or adapted as a means to cope with the specific problems created by user generated content. Such problems arise from poorly written and very short texts as well as from hidden or distributed information.

Searching is a fundamental task to support research. Current search engines are keyword-based. Semantic technologies promise a next generation of semantic search engines, which will be able to answer questions. Current approaches either apply natural language processing to unstructured text or they assume the existence of structured statements over which they can reason. This work provides a system for combining the classical keyword-based search engines with semantic annotation. Conventional search results are annotated using a customized annotation algorithm, which takes the textual properties and requirements such as speed and scalability into account. The biomedical background knowledge consists of the GeneOntology and Medical Subject Headings and other related entities, e.g. proteins/gene names and person names. Together they provide the relevant semantic context for a search engine for the life sciences. We develop the system GoWeb for semantic web search and evaluate it using three benchmarks. It is shown that GoWeb is able to aid question answering with success rates up to 79%. Furthermore, the system also includes semantic hyperlinks that enable semantic browsing of the knowledge space. The semantic hyperlinks facilitate the use of the eScience infrastructure, even complex workflows of composed web services. To complement the web search of GoWeb, other data source and more specialized information needs are tested in different prototypes. This includes patents and intranet search. Semantic search is applicable for these usage scenarios, but the developed systems also show limits of the semantic approach. That is the size, applicability and completeness of the integrated ontologies, as well as technical issues of text-extraction and meta-data information gathering. Additionally, semantic indexing as an alternative approach to implement semantic search is implemented and evaluated with a question answering benchmark. A semantic index can help to answer questions and address some limitations of GoWeb. Still the maintenance and optimization of such an index is a challenge, whereas GoWeb provides a straightforward system.

Das Finden von zuverlässigen Informationen über gegebene Ereignisse aus großen und dynamischen Textsammlungen, wie dem Web, ist ein wichtiges Thema. Zum Beispiel sind Rettungsteams und Versicherungsunternehmen an prägnanten Fakten über Schäden nach Katastrophen interessiert, die heutzutage online in Web-Blogs, Zeitungsartikeln, Social Media etc. zu finden sind. Solche Fakten helfen, die erforderlichen Hilfsmaßnahmen zu bestimmen und unterstützen deren Koordination. Allerdings ist das Finden, Extrahieren und Aggregieren nützlicher Informationen ein hochkomplexes Unterfangen: Es erfordert die Ermittlung geeigneter Textquellen und deren zeitliche Einordung, die Extraktion relevanter Fakten in diesen Texten und deren Aggregation zu einer verdichteten Sicht auf die Ereignisse, trotz Inkonsistenzen, vagen Angaben und Veränderungen über die Zeit. In dieser Arbeit präsentieren und evaluieren wir Techniken und Lösungen für jedes dieser Probleme, eingebettet in ein vierstufiges Framework. Die angewandten Methoden beruhen auf Verfahren des Musterabgleichs, der Verarbeitung natürlicher Sprache und des maschinellen Lernens. Zusätzlich berichten wir über die Ergebnisse zweier Fallstudien, basierend auf dem Einsatz des gesamten Frameworks: Die Ermittlung von Daten über Erdbeben und Überschwemmungen aus Webdokumenten. Unsere Ergebnisse zeigen, dass es unter bestimmten Umständen möglich ist, automatisch zuverlässige und zeitgerechte Daten aus dem Internet zu erhalten. / Finding reliable information about given events from large and dynamic text collections, such as the web, is a topic of great interest. For instance, rescue teams and insurance companies are interested in concise facts about damages after disasters, which can be found today in web blogs, online newspaper articles, social media, etc. Knowing these facts helps to determine the required scale of relief operations and supports their coordination. However, finding, extracting, and condensing specific facts is a highly complex undertaking: It requires identifying appropriate textual sources and their temporal alignment, recognizing relevant facts within these texts, and aggregating extracted facts into a condensed answer despite inconsistencies, uncertainty, and changes over time. In this thesis, we present and evaluate techniques and solutions for each of these problems, embedded in a four-step framework. Applied methods are pattern matching, natural language processing, and machine learning. We also report the results for two case studies applying our entire framework: gathering data on earthquakes and floods from web documents. Our results show that it is, under certain circumstances, possible to automatically obtain reliable and timely data from the web.