Bei der Informationsextraktion (IE) geht es um die automatische Extraktion von Information aus Volltexten. Die Anwendungen erstrecken sich von der Unterstützung von Internet-Suchmaschinen bis hin zum automatischen Aufbau von Fachdatenbanken. Die Methoden reichen von der Analyse natürlicher Sprache über automatische Termerkennung bis zu automatischen Lernverfahren, wobei symbolische, statistische und hybride Methoden zum Einsatz kommen. Komplexe Informationsstrukturen können mit sogenannten Templates (Informationsmustern) repräsentiert werden. In der Veranstaltung werden verschiedene Anwendungen und Methoden für diverse Anwendungsdomänen betrachtet.
Inhalte:
Das Seminar behandelt Ansätze, Verfahren und Werkzeuge der Informationsextraktion und legt einen besonderen Fokus auf die Erkennung von Eigennamen und von domänen- bzw. fachspezifischer Information. Auch sollen Fragestellungen der Evaluation derartiger Verfahren diskutiert werden.
Lernziele:
Die Teilnehmer sollen lernen, wie sie Ressourcen für IE Systeme bewerten können. Außerdem sollen sie befähigt werden, dass sie bei der Entwicklung, beim Einsatz und bei der Bewertung von IE-Systemen mitwirken können.
Here is a link to the Lecture
Email Address: SubstituteMyLastName@cis.uni-muenchen.de
Tutor: Tobias Eder
Email Address: tobias.eder@in.tum.de
There are *two separate seminars*. You EITHER go on Wednesdays, OR you go on Thursdays, NOT BOTH!
Wed: 10:00 c.t., Room U139
Thurs: 10:00 c.t., Room U139
For a LaTeX template for the Hausarbeit, click here.
If this web page does not seem to be up to date, use the refresh button in your browser.
| Date | Topic | Materials |
| October 17th and October 18th | Introduction, Information on Participants | |
| October 24th and October 25th | Referat Topics | fraser.pdf hangya.pdf huck.pdf stojanovski.pdf |
| October 31st and November 1st | Cancelled (EMNLP and WMT conferences) and holiday | |
| November 7th and November 8th | Only office hours - (Nov 7th Lecture will take place!) | |
| November 14th and November 15th | First Exercise - in Gobi | IE_exercise1.tgz IE_exercise1_notes.txt |
| December 19th and December 20th | Second Exercise - in Gobi | IE_exercise2.tar.xz IE_exercise2_notes.txt |
| February 6th (starting early at 9:30am s.t. ! ) and February 7th | Third Exercise - in Gobi | IE_exercise3.tar.gz IE_exercise3_notes.txt |
WEDNESDAY Referatsthemen (name: topic)
| Date | Topic | Language | Materials | Hausarbeit Received |
| 21.11 | (AF) History of IE | EN | see Thursday (Möller) | |
| 21.11 | (AF) Florian Babl: Rule-based vs. Statistical | EN | slides | yes |
| 28.11 | (AF) Stefanos Potamianakis: Annotating for NER with Amazon Mechanical Turk | DE | slides | yes |
| 5.12 | (MH) Jing Hu: NER E-Commerce | DE | slides | yes |
| 5.12 | (MH) Cornelia Dekinger: Biomed NER Deep Learning | DE | slides | yes |
| 12.12 | (MH) Wilma Hampl: Rule-based Relation Extraction | DE | slides | yes |
| 12.12 | (MH) Phuong-Anh Tran: Relation Extraction Using Dependency Trees | DE | slides | yes |
| 12.12 | (MH) Andreas Maier: Relation Extraction and Scoring for Question Answering | DE | slides | yes |
| 19.12 | EXERCISE IN GOBI (SEE ABOVE)! | |||
| 9.1 | (DS) Mingfei Cui: Multilingual NER | EN | slides | yes |
| 9.1 | (DS) Viktoriia Mokriak: Aspect-based Sentiment | EN | slides | yes |
| 16.1 | (DS) Wei Hong: Twitter Sentiment | EN | slides | yes |
| 16.1 | (VH) Yi-Chen Liu: Cross-lingual Sentiment | EN | slides | yes |
| 16.1 | (VH) Adrian Reitz: Joint aspect and sentiment classification | EN | slides | yes |
| 23.1 | (AF) Dorian David: Detecting Social Spambots | DE | slides | yes |
| 23.1 | (MH) Olga Melnikova: Automatic Biomed Knowledge Extraction | DE | slides | yes |
| 23.1 | (AF) Miriam Rupprecht: Event Extraction - Disasters | DE | slides | yes |
| 30.1 | (DS) Elisabeth Busse: Document Event Extraction | EN | slides | yes |
| 30.1 | (AF) Simone Glück: Fake News | EN | slides | yes |
| 6.2 at 9:30am | (AF) Hanna Razumava: Never-ending Learning (Open IE) | DE | slides | yes |
| 20.02 | (AF) Robert Gruber: Fine-Grained Entity Recognition | DE | slides | yes |
| 20.02 | (AF) Irina Trefilova: NER Twitter | DE | slides | yes |
THURSDAY Referatsthemen (name: topic)
| Date | Topic | Language | Materials | Hausarbeit Received |
| 22.11 | (AF) Yannic Möller: History of IE | EN | slides | yes |
| 22.11 | (AF) Son Tien Nguyen: Rule-based vs. Statistical | EN | slides | yes |
| 29.11 | (AF) Olesia Melnychuk: NER Twitter | DE | slides | yes |
| 29.11 | (MH) Angelina Koltsova: NER E-Commerce | DE | slides | yes |
| 29.11 | (AF) Behzat Cinar: Entity Linking in the Digital Humanities | DE | slides | yes |
| 6.12 | (MH) Valiantsina Vauraniuk: Relation Extraction and Scoring for Question Answering | DE | slides | yes |
| 6.12 | (AF) Marika Rionidze: Event Extraction - Disasters | DE | slides | yes |
| 13.12 | (DS) Korbinian Schmidhuber: Multilingual NER | EN | slides | yes |
| 13.12 | (MH) Erik Weber: Biomed NER Deep Learning | EN | slides | yes |
| 13.12 | (VH) Daniel Teplan: Parallel Sentence Extraction - BUCC Shared Task | EN | slides | yes |
| 20.12 | EXERCISE IN GOBI (SEE ABOVE)! | |||
| 10.1 | (VH) Johanna Strebl: Parallel Sentence Extraction - Sentence Representation | EN | slides | yes |
| 10.1 | (DS) Leopold Martin: Twitter Sentiment | EN | slides | yes |
| 17.1 | (VH) Andreas Wassermayr: Cross-lingual Sentiment | EN | slides | yes |
| 17.1 | (VH) Dominik Braun: Joint aspect and sentiment classification | EN | slides | yes |
| 17.1 | (VH) Manuel Pleier: Abstractive Neural Text Summarization | EN | slides | yes |
| 24.1 | (DS) Manja Faulhaber: Document Event Extraction | EN | slides | yes |
| 24.1 | (DS) Tianqi Bao: Coreference | EN | slides | yes |
| 24.1 | (VH) Ario Cecchettini: Document Classification with Hierarchical Attention | EN | slides | yes |
| 31.1 | (AF) Mischan Malek: IE from Audio | EN | slides | yes |
| 31.1 | (AF) Jakob Jungmaier: Incomplete Knowledge Bases | EN | slides | yes |
| 31.1 | (AF) Leah Michel: Fake News | EN | slides | yes |