Seminar – Ixa Group. Language Technology.

Seminar: Irish Machine Translation and Resources (M. Dowling, 2019-03-11)

Kepa Sarasola — Mon, 11 Mar 2019 11:46:24 +0000

Seminar title: Irish Machine Translation and Resources
Speaker: Meghan Dowling (PhD student in the ADAPT Centre, Dublin City University)
When: Mon, 11 Mars, 15:00pm
Where: Faculty of informatics, room 3.2 gelan map

Summary: As an official language in both Ireland and the EU, there is a high demand for Irish language translation in public administration. The difficulty that translators face in meeting this demand leads to an even greater need for English-Irish (EN-GA) machine translation (MT). This presentation will discuss the advances in EN-GA MT that have been made so far, the language resources gathered, as well as some possible avenues for future research.

Bio: Meghan Dowling is a PhD student in the ADAPT Centre, Dublin City University. Her PhD topic focuses on improving English-Irish machine translation. She is visiting IXA group for 3 months to learn about how Basque resources and MT have improved, with a view to using this knowledge to improve resources and MT for Irish.

Presentation: Research groups in the Faculty of Informatics (2017-10-10, 10:00-11:10)

Kepa Sarasola — Mon, 09 Oct 2017 12:49:57 +0000

Tomorrow morning the research groups in the Faculty of Informatics will present their work to the students.

Date: Tuesday, October 10
Time: 10:05-11:10
Where: Ada-Lovelace room
Audience: Students of 3rd & 4th levels
Subject: Presentation of research subjects and groups in the Faculty.
IXA Group’s collaboration with students: job opportunities for undergraduate students, scholarships…

Research groups in the Faculty of Informatics

Seminar: Big Data and NLP at Trivago (Min Fang, 2016-06-08)

Kepa Sarasola — Tue, 07 Jun 2016 11:50:29 +0000

Talk: Big Data and NLP at Trivago

Speaker: Min Fang

………..2013 – 2015: Master Erasmus Mundus Language and Communication Technologies, summa cum laude

………..2015-… : (Trivago, hotel metasearch)

When: Wed, 8 June, 10pm – 11pm

Where: Faculty of informatics, room 3.2 gelan map

Abstract:

I’m interested in getting insights from data by applying natural language processing, machine learning and statistical analyses. Ideally, those insights can then be turned into useful applications or facilitate higher level decisions.

Together with our software engineers I take care of our NLP capabilities: We work on improving and maintaining a highly flexible and scalable pipeline that is geared towards aspect-based sentiment analysis (and more in the future). Extracting knowledge from a large number of natural language texts allows us to understand our domain better and enhance the experience for our users.

Our technology stack includes:
– Python and Java
– R for analysis
– AWS for infrastructure

Talk: Irish language technology (T. Lynn, 2015/10/13)

Kepa Sarasola — Fri, 09 Oct 2015 09:33:30 +0000

Speaker: Teresa Lynn ADAPT Centre, School of Computing, Dublin City University
Data: October 13th 2015, Tuesday
Time: 15:00-16:00
Room: 3.2 room. Faculty of Informatics (UPV/EHU)
Title: Overview of the current status of Irish language technology

The Irish government have asked a group (two members from Dublin City University, and three from Trinity College Dublin) to compile a digital strategy for Irish language technology. Teresa Lynn will be next week visiting Basque Country to gain insight from our experience in the area of developing resources for Basque, another low-resourced language, and to see what advice they could harness from those experiences. We asked her to present us an overview of the current status of Irish language technology.

Seminar: ‘The Lexikoaren Behatokia project’ & ‘Enriching EDBL with Hiztegi Batua’ (12/11/2013)

Kepa Sarasola — Fri, 13 Dec 2013 09:01:11 +0000

Topics : The Lexikoaren Behatokia project (X. Artola) + Enriching EDBL with Hiztegi Batua (Gorka Labaka – Xabier Artola)
Speakers: Xabier Artola and Gorka Labaka
Day: December 11th 2013, Wednesday

The Basque language academy, Euskaltzaindia, launched in 2008 the Lexikoaren Behatokia project (“The Lexicon Observatory”), led by Andoni Sagarna. The objective of the project was to create a labelled and linguistically annotated corpus for research. In order to carry out the corpus, it was expected to use a variety of sources, mostly media, especially general interest ones. In late 2012, the corpus consisted of 26,565,924 words, and has been expanded year after year. Euskaltzaindia, the IXA research group, the Elhuyar foundation and UZEI collaborate on the project.

The Lexikoaren Behatokia corpus is available here.

On the other hand, the latest version of the prescriptive dictionary Hiztegi Batua has provided new entries for the Basque lexical database EDBL. This enriching process was explained in the seminar.

Seminar: Unsupervised Relation Extraction with General Domain Knowledge (2013/07/10)

Kepa Sarasola — Wed, 17 Jul 2013 16:45:06 +0000

Last week Oier Lopez de Lacalle participated in our IXA seminar with a presentation entitled “Unsupervised Relation Extraction with General Domain Knowledge“.

Currently, Oier Lopez de Lacalle is a Research Fellow at University of Edinburgh.

Abstract:

In this work we present an unsupervised approach to relational information extraction. Our model partitions tuples representing an observed syntactic relationship between two named entities (e.g., “X was born in Y” and “X is from Y”) into clusters corresponding to underlying semantic relation types (e.g., BornIn, Located). Our approach incorporates general domain knowledge which we encode as First Order Logic rules and automatically combine with a topic model developed specifically for the relation extraction task. Evaluation results on the ACE 2007 English Relation Detection and Categorization (RDC) task show that our model outperforms competitive unsupervised approaches by a wide margin and is able to produce clusters shaped by both the data and the rules.

Seminar: HPSG grammars. (2013/07/15).

Kepa Sarasola — Wed, 10 Jul 2013 15:16:04 +0000

Speakers: Montse Marimón and Lluis Padró
Title: HPSG grammars
Date: July 15, 2013, Monday
Time: 10:00-13:00
Where: Computer Science Faculty, Room 3.

Outline:

What is a HPSG Grammar? Head-driven phrase structure grammar
Spanish HPSG grammar. Examples.
HPSG for other languages and Parallel deepbanks.

References:

Montse Marimón. (2013) The Spanish DELPH-IN grammar. Language Resources and Evaluation, 1-27.
Marimon, M., & Padró, L. (2012). A hybrid approach to treebank construction. Procesamiento del Lenguaje Natural, 49, 139-146.
Marimon Felipe, M. (2010). The Tibidabo Treebank. Procesamiento del lenguaje natural, 2010, vol. 45, num. 1, p. 113-119.
Gojenola K., (1998) Guneak zuzendutako egitura sintagmatikoen gramatika (HPSG) eta Euskararako aplikazioa. UPV/EHU/LSI/TR 5-98.

Seminar. First steps towards Quechua’s processing. (2012/11/15)

Kepa Sarasola — Mon, 12 Nov 2012 20:00:50 +0000

Hugo and Richard visiting Aholab in Bilbao.

Speakers: Hugo Quispe and Richard Castro (Universidad UNSAAC of Cusco, Peru),
………………Olatz Arregi, Xabier Artola eta Kepa Sarasola (Ixa Group)
Title: Primera aproximación al procesamiento automático del Quechua
………(First steps towards Quechua’s processing.)
Date: November 15, 2012, Thursday
Time: 16:00–17:00
Where: Computer Science Faculty, Room 3.2

Abstract

El Quechua (Runa Simi) como lengua oriunda de la cultura Inca en el Perú, es una familia de lenguas en Latinoamérica. La situación actual de la lengua, por factores como la occidentalización entre otros, ha hecho que el quechua sea una lengua vulnerable, en vías de extinción.

Un grupo de profesores e investigadores del grupo IXA de la UPV/EHU, en conjunto con la UNSAAC en Cusco, Perú, estamos realizando un trabajo para sentar las bases de lo que pretende ser el centro de ingeniería lingüística
de Cusco. Se trata de desarrollar los primeros recursos básicos y herramientas para al procesamiento automático del quechua. Los temas en los que estamos trabajando son: recopilación de un corpus textual, una base de datos léxica para la lengua quechua (BDLQ) y futuras herramientas derivadas de la misma, uso de la herramienta FOMA en el análisis morfológico y creación de un TTS como herramientas básicas para el tratamiento de la lengua.

De esta manera, se ha consolidado las bases de apoyo y trabajo en equipo entre las dos universidades, en bien de una lengua en situación crítica.

Hugo and Richard visiting Ixa Group in Donostia.

Quechua (Runa Simi) is a native South American language family and dialect cluster spoken primarily in the Andes of South America. It is the most widely spoken language family of the indigenous peoples of the Americas, with a total of probably some 8 to 10 million speaker. Like Basque Quechua remains alive but in last centuries suffered continuous regression. The region in which Quechua is spoken is becaming smaller and smaller. Similar with what happened with Basque, Quechua was not an official language, it has been out of educational systems, out of media, and out of industrial environments. Today Quechua holds co-official language status in Peru and Bolivia, even it is not regulated. But, although there have been several changes in the last years, Quechua is still associated with lack of education, stigmatized as uneducated, rural, or holding low economic and power resources, as it was Basque some years ago. Language technology may help to the Quechua speakers’ community and to scholars to built a standard. So opening a door to face Quechua’s future in the digital world. Corpus tools, lexical data-bases and spelling checkers have proven to be useful tools in that way for other languages such as Basque.
The group created by Prof. Juan Cruz in UNSAAC University in Cusco (Peru) has been collaborating with Ixa Group and Aholab since the beginning of 2012. Hugo Quispe and Richard Castro will present in this seminar the work they are doing on the definition of a lexical data-base and a TTS system (Text to Speech) for Quechua.

The group of Cusco (January 2012)