Natural Language Interfaces Natural Langauge Processing NLP The

Natural Language Interfaces

Natural Langauge Processing (NLP) The European Union (EU) is a political and economic union of 28 member states that are located primarily in Europe. It has an area of 4, 475, 757 km 2 (1, 728, 099 sq mi), and an estimated population of over 510 million. The EU has developed an internal single market through a standardised system of laws that apply in all member states. EU policies aim to ensure the free movement of people, goods, services, and capital within the internal market, [13] enact legislation in justice and home affairs, and maintain common policies on trade, [14] agriculture, [15] fisheries, and regional development. [16] Within the Schengen Area, passport controls have been abolished. [17] A monetary union was established in 1999 and came into full force in 2002, and is composed of 19 EU member states which use the euro currency. The EU traces its origins from the European Coal and Steel Community (ECSC) and the European Economic Community (EEC), established, respectively, by the 1951 Treaty of Paris and 1957 Treaty of Rome. The original members of what came to be known as the European Communities, were the Inner Six; Belgium, France, Italy, Luxembourg, the Netherlands and West Germany. The Communities and its successors have grown in size by the accession of new member states and in power by the addition of policy areas to its remit. While no member state has left the EU or its antecedent organisations, the United Kingdom enacted the result of a membership referendum in June 2016 and is currently negotiating its withdrawal. The Maastricht Treaty established the European Union in 1993 and introduced European citizenship. [18] The latest major amendment to the constitutional basis of the EU, the Treaty of Lisbon, came into force in 2009. The European Union provides more foreign aid than any other economic union. [19] Covering 7. 3% of the world population, [20] the EU in 2016 generated a nominal gross domestic product (GDP) of 16. 477 trillion US dollars, constituting approximately 22. 2% of global nominal GDP and 16. 9% when measured in terms of purchasing power parity. [citation needed] Additionally, 27 out of 28 EU countries have a very high Human Development Index, according to the United Nations Development Programme. In 2012, the EU was awarded the Nobel Peace Prize. [21] Through the Common Foreign and Security Policy, the EU has developed a role in external relations and defence. The union maintains permanent diplomatic missions throughout the world and represents itself at the United Nations, the World Trade Organization, the G 7, and the G 20. Because of its global influence, the European Union has been described as an emerging superpower. [22]

IBM Watson

Speech recognition

Speech recognition

Chatbots

Dialogue systems (chatbots) Database lookup • Webshops • Call centers • Decision support systems Navigation systems Tutoring

Theorical architecture Text processing/understanding Semantic representation Dialogue strategy Semantic representation of the answer generation Text

Turing test

Chatbots today • Question templates (regular expressions) and slot-based answers (bag-of-words) • only works in very narrow domains • question-answer bank • try to control the discussion, ask very simple questions • they collect dialouges and use data for hand-crafting rules (or machine learning)

Question answering

Question answering (QA) • Input: a natural language question • Output: – the document with an answer (similiar to information retrieval) – the relevant paragraph (or an abstractive summary) – the answer itself

Type of questions • • • yes/no factual (person name, date, etc. ) definition list How? Why?

QA Architecture • extracting the key words from the question • creating queries built from the keywords Where did Petőfi born? „Petőfi * born in” • querying big datasets • relevancy scoring of paragraphs/sentences (similarity, in-document position etc)

(other) NLP applications

Machine Translation • translating whole texts from a source language to a target language • Computer Aided Translation (CAT) • Why and how? – EU spends 1 billion € per year on official translations – Quick access of internet text in foreign language (Google Translate)

Differences among languages • lexical differences – red vs. vörös, piros

Document classification of documents into pre-defined categories (document can be multimodal, like text + image)

Application areas for document classification since 1961! • Filtering (spam, news) • Organisation (e. g. advertisment) • CRM routing • automatic assesment of exams Topic detection

Document clustering and automatic labeling Linguistics Machine Learning Probability therory

Information extraction

Information retrieval vs Information extraction

Named Entity Recognition person, organisations, locations, etc United States Department of Homeland Security semantic class: Ford normalization: FC Barcelona and Barca

Event extraction • Relations among entities • Events

Sentiment analysis opinion about products, parties, ideas based on various aspects

Summarisation • Summary: short, but reliable representation of the documents content • short? • content from which aspect? „I took a speed-reading course and read War and Peace in twenty minutes. It involves Russia. ” Woody Alen

Keyword extraction Set of words/phrases(=multiword expressions) to describe the content of document(s)