2015

TraMOOC Progress Timeline

13/02

TraMOOC kicks-off

TraMOOC Kicks-off

Humboldt Universität zu Berlin hosted a one-day event on February 13th, 2015 during which the official launch of the TraMOOC project took place. Besides the formal project kick off, the meeting provided an opportunity for all 10 partners to meet, obtain an overview of the entire project and initiate the first interactions.

Starting the meeting, Valia Kordoni, the Project Coordinator, welcomed all participants and provided an overview presentation of the project and its expected results. The management structure and the review processes were explained to all participants to allow for smooth and timely communication between the consortium members. Also, Saila Rinne, the EC project officer presented the Horizon 2020 procedures including the novel points compared to FP7. She also encouraged the consortium to seek possible synergies with other relevant projects. Following that, the individual partners gave an overview presentation about their scheduled activities during the course of the project.

Read More
17/02

TraMOOC related article @LinkedIn

TraMOOC related article @LinkedIn

A TraMOOC related web based article titled "Automatic translation of MOOCs into 11+ languages " was published at LinkedIn in February 2015.

Please follow this link in order to view the article.

Read More
18/02

Press Release – 3 Million euros to boost the quality of automatic translation

Press Release – 3 Million euros to boost the quality of automatic translation

Press Release – 3 Million euros to boost the quality of automatic translation (Link)

01/03

TraMOOC at the SEQUENT Consultation Sessions

TraMOOC at the SEQUENT Consultation Sessions

TraMOOC was present at the SEQUENT Consultation Sessions, which took place in March 2015 at Jozef Stefan Institute, Ljubljana, Slovenia, making an oral presentation titled "TraMOOC, language tools for Open Education".

10/03

Translation for MOOCs to be developed by the TraMOOC project

Translation for Massive Open Online Courses to be developed by the TraMOOC project

Massive Open Online Courses have been growing rapidly in size and impact. Yet the language barrier constitutes a major growth impediment in reaching out to all peoples and educating all citizens.

The traMOOC project launched in February 2015 aims at tackling this impediment by developing high-quality automatic translation of various types of texts included in MOOCs from English into eleven European and BRIC languages. Phrase-based and syntax-based statistical machine translation models will be developed for addressing language diversity and supporting the language-independent nature of the methodology. For a high quality, automatic translation approach and for adding value to existing infrastructure, extensive advanced bootstrapping of new resources will be performed.An innovative multi-modal automatic and human evaluation schema will further ensure translation quality. For human evaluation, an innovative, strict-access control, time- and cost-efficient crowdsourcing setup will be used. Translation experts, domain experts and end users will also be involved.The project results will be showcased and tested on the Iversity MOOC platform and on the VideoLectures.NET digital video lecture library (Link).

Read More
01/04

TraMOOC at the Open Education Global Conference 2015

TraMOOC at the Open Education Global Conference 2015

TraMOOC was present at the Open Education Global Conference 2015, which took place in April 2015 at Banff, Alberta, Canada, making an oral presentation titled "TraMOOC, a project fitting the “Opening up Education” communication".

07/04

Quality risk and management mechanisms defined

Quality risk and management mechanisms defined

The quality risk and management mechanisms to be implemented by the consortium have been defined. The mechanisms are based on the Grant and Consortium Agreements and follow good project management practices.The relevant consortium entities and their responsibility with respect to the quality and risk management have also been defined.

08/04

Language Resources & Tools Management Plan

Language Resources & Tools Management Plan

The project-level management mechanisms of the language resources and tools to be employed/adapted/developed in TraMOOC have been defined. Project activities involving language resources and tools are spread across multiple work packages. It is therefore important to have good interaction between different tasks, work packages, and teams.

The responsibilities of each partner and the timeline for the project activities involving language resources and tools have been clearly defined. The timeline is organised around the machine translation prototypes which will be produced within the project, thus making sure that the language resources and tools needed for each prototype will be ready on time.

Read More
15/04

TraMOOC in the EASN Newsletter

TraMOOC in the EASN Newsletter

The Translation for Massive Open Online Courses (TraMOOC) project is a Horizon 2020 collaborative project aiming at providing reliable machine translation for Massive Open Online Courses (MOOCs).

Coordinated by the Humboldt-Universität zu Berlin, it brings together 10 partners from 6 European countries with wide experience in their field, allowing for an unhindered course towards achieving the project’s goals.  Following the successful Kick-off meeting held on the Humboldt-Universität zu Berlin premises on February 13th, 2015, the TraMOOC project launched its activities, focusing on identifying the project’s use cases and on creating the first language pairings for the machine translation process. The main outcome of the project is an online translation platform which will provide high-quality translation of all types of text genre included in MOOCs from English into eleven European and BRIC languages, which currently have weak Machine Translation support. After the online TraMOOC service launch, EASN network members, natives of all eleven targeted languages, will be identified as domain experts and they will be asked via questionnaires and interviews to evaluate the translation output of course material pertaining to their domain of expertise. Information will soon be available on the TraMOOC project website (Link).

Read More
22/04

TraMOOC, a project fitting the "Opening up Education"

TraMOOC, a project fitting the “Opening up Education” in the Open Education Global Conference 2015

TRAMOOC was present at the Open Education Global Conference 2015 held between the 22nd and the 24th of April, 2015 at Banff, Alberta, Canada.

27/04

TraMOOC at the META-FORUM 2015

TraMOOC at the META-FORUM 2015

TRAMOOC was present at the META-FORUM 2015 - Technologies for the Multilingual Digital Single Market, which was held on the 27th of April, 2015 at Riga, Latvia, presenting an overview poster presentation of the project.

29/04

TraMOOC Intranet & public website

TraMOOC Intranet & public website

The Intranet is the main communication, monitoring, and information exchange mechanism employed by the consortium. The Intranet is a wiki-based online tool to which all partners and the EU project officer have equal access and can upload, download or edit information.

The collaborative nature of the Intranet facilitates monitoring of the work progress and enables all partners and the project officer to monitor the results of the project at any particular point of time as well as the achievement of the overall goals of the project. The Intranet is available at the following web address: http://www.tramooc.eu/intranet. In addition, the TraMOOC public website has been designed, created and implemented on-line including general information about the project and the Consortium. The public area of the website aims at providing information about the project, its progress, main results and achievements. The website is available at the following web address: http://tramooc.eu/.

Read More
01/05

TraMOOC at the OPENING UP EDUCATION

TraMOOC at the OPENING UP EDUCATION - Irish National MOOC Symposium

TRAMOOC was present at the OPENING UP EDUCATION - Irish National MOOC Symposium held on the 1st of May, 2015 at Dublin, Ireland. Dr. John Judge, represented the TraMOOC project in this event and presented the project overview and its expected impact.

10/05

TraMOOC at the German National Commission for UNESCO meeting

TraMOOC at the meeting held by the German National Commission for UNESCO

TRAMOOC was present at the Slovenian National Commission for UNESCO, Ljubljana, Slovenia, May 2015 making a presentation titled "TraMOOC, an asset for global education and UNESCO ". 

11/05

TraMOOC at the EAMT2015

TraMOOC Overview Presentation at the 18th Annual Conference of the European Association for Machine Translation (EAMT2015)

TRAMOOC was present at the 18th Annual Conference of the European Association for Machine Translation (EAMT2015) held between the 11th and the 13th of May, 2015 at Antalya, Turkey.

This event was a great opportunity for increasing awareness about the scope and objectives of the project and for creating momentum about the project from its beginning. The project coordinator, Prof. Valia Kordoni, represented the TraMOOC project in this event and presented the project overview and its expected impact. For more information please follow this link.

Read More
12/05

TraMOOC in the Slovenian National Commission for UNESCO meeting

TraMOOC, an asset for global education and UNESCO” in the Slovenian National Commission for UNESCO meeting

TraMOOC was present at the Slovenian National Commission for UNESCO meeting held at Ljubljana, Slovenia in May 2015.

18/06

TraMOOC EAB and 1st Plenary Meeting

TraMOOC External Advisory Board (EAB) and 1st Plenary Meeting

The progress achieved so far in the scope of TraMOOC, along with the overall project goals were presented to the project External Advisory Board meeting which was held between June 18th and 19th, 2015 in Corfu, Greece.

The External Advisory Board (EAB) provided extremely insightful and positive feedback which will be taken into consideration while proceeding with the work to be performed throughout the course of the project. During the second half of the second day of the meeting, the Project Coordination Group was convened in order to decide upon the course of action for the next period and on issues of its responsibility, thus ensuring the seamless advancement of the project’s work.

Read More
01/07

TraMOOC at the COMPSAC 2015

TraMOOC at the COMPSAC 2015: The 39th Annual International Computers, Software & Applications Conference

TRAMOOC was present at the COMPSAC 2015: The 39th Annual International Computers, Software & Applications Conference held on the 1st of July, 2015 at Taichung, Taiwan.

The presentation titled “Open Education Resources: Some Research Issue” was made at this event, which was a great opportunity for increasing awareness about the scope and objectives of the project.

Read More
20/07

TraMOOC in the EASN Newsletter

TraMOOC in the EASN Newsletter

The TraMOOC project is in its sixth month of activities and the partners’ efforts are going full steam ahead on the preparation of the first Machine Translation (MT) prototype.

More specifically, the first months of the project were dedicated to the definition of the end user requirements, which will lead to the identification of the system’s functionalities, a necessary step towards the realization of the TraMOOC system architecture. Additionally, during this period other activities involved the collection of relevant parallel data for training the MT system and the conversion of all data acquired to a suitable common format so as to enable their use for training the MT system.  The progress achieved so far, along with the overall project goals and objectives were presented to the project External Advisory Board meeting which was held between June 18th and 19th, 2015 in Corfu, Greece. The External Advisory Board (EAB) provided extremely insightful and positive feedback to the Consortium which will definitely be taken into consideration throughout the course of the project (Link).

Read More
27/07

Data Management Plan

Data Management Plan

The Data Management Plan of the project provides a short, general outline of the consortium’s policy for data management, including the types of data which will be generated/ collected, the standards which will be used, how this data will be shared and/or made accessible for verification and use and how this data will be curated and preserved.

28/07

Initial Communication Plan

Initial Communication Plan

Communicating and disseminating the TraMOOC project concept, main objectives, achievements and outcomes to the wider European scientific community and the general public are of primary importance to the TraMOOC project consortium.

In order to raise awareness about the project and its results, a set of communication tools establishing the project’s “visual identity” has been developed (including the project logo, poster, leaflet, service mock-up demo video, presentation and deliverable templates, social media pages) and shall be updated throughout the project’s lifetime.

Read More
30/07

TraMOOC Mock-up video

TraMOOC Mock-up Video

This video aims at visualizing the project's main outcome, i.e., the translation platform and its functionalities. The platform is designed within the frame of the European ρesearch project TraMOOC (http://www.tramooc.eu) and explores a promising new reality, as the goal of TraMOOC is to provide accurate, reliable and coherent machine translation for all types of educational textual data available on a MOOC platform (Link).

31/07

Translation Prototype Version 1

Translation Prototype Version 1

The translation prototype version 1 has been developed. The prototype provides four of the eleven language pairs and will be used to perform initial implicit and explicit evaluations to estimate the accuracy of the translated text. It will be succeeded by the TraMOOC translation prototype version 2 in June 2016.

01/08

TraMOOC related article

TraMOOC related article

A TraMOOC related web based article titled "TraMOOC – Translation for Massive Open Online Courses was published at Opening Up Slovenia 2015 website in August 2015. 

Please view the article by following this link.

Read More
05/08

TraMOOC Facebook Group

TraMOOC Facebook Group

The TraMOOC Facebook Group is online!!!

20/08

TraMOOC press release

TraMOOC press release

Ireland’s ADAPT Centre at heart of €3 million EU project to make online learning accessible across languages (Link).

21/08

TraMOOC press release

Online language service lauded

An Irish online service aimed at developing linguistic skills in the EU and for the Government could save millions of euro. (Link)

22/08

TraMOOC press release

DCU researchers help secure €3m for machine-based translation project

Major new EU funding for research aimed at removing language barriers to distance learning was announced today by a group comprising the Adapt Centre for Digital Content Technology at Dublin City University (DCU) and nine other European partners. (Link)

02/09

TraMOOC at the Conference on Empirical Methods in Natural Language Processing (EMNLP 2015)

TraMOOC at the Conference on Empirical Methods in Natural Language Processing (EMNLP 2015)

TRAMOOC was present at the Conference on Empirical Methods in Natural Language Processing (EMNLP 2015), which was held on the 2nd of September, 2015 at Lisbon, Portugal, and the presentation titled "A Joint Dependency Model of Morphological and Syntactic Structure for Statistical Machine Translation " was made.  

In this paper the authors propose to model syntactic and morphological structure jointly in a dependency translation model, allowing the system to generalize to the level of morphemes. They also present a dependency representation of German compounds and particle verbs that results in improvements in translation quality of 1.4–1.8 BLEU in the WMT English–German translation task. For more information please follow this link.

Read More
07/09

TraMOOC LinkedIn Group

TraMOOC LinkedIn Group

The TraMOOC LinkedIn Group is online!!!

17/09

TraMOOC at the Tenth Workshop on Statistical Machine Translation

TraMOOC at the Tenth Workshop on Statistical Machine Translation

TRAMOOC was present at the Tenth Workshop on Statistical Machine Translation, which was held between the 17th and the 18th of September, 2015 at Lisbon, Portugal, and the presentations titled "CHRF: character n-gram F-score for automatic MT evaluation" and "The Edinburgh/JHU Phrase-based Machine Translation Systems for WMT 2015 (Poster Presentation)" were made.

In the paper titled "CHRF: character n-gram F-score for automatic MT evaluation" the authors propose the use of character n-gram F-score for automatic evaluation of machine translation output. Also, the poster presentation titled "The Edinburgh/JHU Phrase-based Machine Translation Systems for WMT 2015 (Poster Presentation)" presented the Phrase-based translation systems using the open source Moses toolkit.    

Read More
18/09

TraMOOC at the 2nd Internet of Education Conference 2015

TraMOOC at the 2nd Internet of Education Conference 2015

A project overview presentation was made at the 2nd Internet of Education Conference 2015 held on the 18th of September, 2015 at Sarajevo, Bosna and Herzegovina.

19/09

TraMOOC at the EMNLP 2015

TraMOOC at the Conference on Empirical Methods in Natural Language Processing (EMNLP 2015)

TRAMOOC was present at the Conference on Empirical Methods in Natural Language Processing (EMNLP 2015), which was held between the 17th and the 21st of September, 2015 at Lisbon, Portugal, presenting the paper titled "Mixed-Domain vs. Multi-Domain Statistical Machine Translation".

In this paper the authors present an approach to domain adaptation which results in large-scale, general purpose machine translation systems. First, their translation models are tuned to multiple individual domains. Then, by means of source-side domain classification, the authors are able to predict the domain of individual input sentences and thereby select the appropriate domain-specific model parameters. This approach is called multi-domain translation. For more information please follow this link.

Read More
20/09

TraMOOC at the EMNLP 2015

TraMOOC at the Conference on Empirical Methods in Natural Language Processing (EMNLP 2015)

TRAMOOC was present at the Conference on Empirical Methods in Natural Language Processing (EMNLP 2015), which was held between the 17th and the 21st of September, 2015 at Lisbon, Portugal, presenting the paper titled "A Joint Dependency Model of Morphological and Syntactic Structure for Statistical Machine Translation". 

In this paper the authors tackle the issue that arise when translating between two languages that differ in their degree of morphological synthesis and syntactic structures in one language may be realized as morphological structures in the other. In this case the Statistical Machine Translation (SMT) models need a mechanism to learn such translations. In this paper the authors propose to model syntactic and morphological structure jointly in a dependency translation model For more information please follow this link.

Read More
21/09

TraMOOC at the EMNLP 2015

TraMOOC at the Conference on Empirical Methods in Natural Language Processing (EMNLP 2015)

TRAMOOC was present at the Conference on Empirical Methods in Natural Language Processing (EMNLP 2015), which was held between the 17th and the 21st of September, 2015 at Lisbon, Portugal, presenting the paper titled "The Edinburgh/JHU Phrase-based Machine Translation Systems for WMT 2015".

In this paper the authors describe the submission of the University of Edinburgh and the Johns Hopkins University for the shared translation task of the EMNLP 2015 Tenth Workshop on Statistical Machine Translation (WMT 2015). The authors have set up phrase-based statistical machine translation systems for all ten language pairs of this year’s evaluation campaign, which are English paired with Czech, Finnish, French, German, and Russian in both translation directions. For more information please follow this link.

Read More
22/09

TraMOOC YouTube Channel

TraMOOC YouTube Channel

The TraMOOC YouTube Channel is online!!! 

23/09

TraMOOC at the fOSSa 2015 Conference

TraMOOC at the fOSSa 2015 Conference

TRAMOOC was present at the fOSSa 2015 Conference - Free Open Source Software for Academia, which was held between the 23rd and the 25th of September, 2015 at Nantes, France, and presented the paper titled "Opening up Slovenia teaching? What about the technologies impact on teaching?".

24/09

TraMOOC press release

3M Euro TraMOOC Project May be Commercialized

Slator recently covered the European Union’s latest attempt at machine translation: Translation for Massive Open Online Courses (TraMOOC). As explained in the article: TraMOOC “aims to leverage an online translation platform utilizing a wide array of linguistic infrastructure tools to reliably translate multimedia data of MOOCs from English into eleven other languages: German, Italian, Portuguese, Greek, Dutch, Bulgarian, Czech, Croatian, Polish, Russian and Chinese.” (Link)

25/09

TraMOOC at the fOSSa 2015 Conference

TraMOOC at the fOSSa 2015 Conference

TRAMOOC was present at the fOSSa 2015 Conference - Free Open Source Software for Academia, which was held between the 23rd and the 25th of September, 2015 at Nantes, France, and presented the paper titled "Debate about crowd teaching & sciences". 

15/10

TraMOOC at the meeting held by German National Commission for UNESCO and UNESCO Chair in Hydrological Change and Water Resources Management

TraMOOC at the meeting held by German National Commission for UNESCO and UNESCO Chair in Hydrological Change and Water Resources Management

TraMOOC was present at the meeting held by German National Commission for UNESCO and UNESCO Chair in Hydrological Change and Water Resources Management, which took place in October 2015 at the RWTH Aachen University, Aachen, Germany, making a presentation titled "TraMOOC, an asset for global education and UNESCO".

30/10

TraMOOC at the Machine Translation Summit XV

TraMOOC at the Machine Translation Summit XV

Domain adaptation boosts translation quality on in-domain data, but translation quality for domain adapted systems on out-of-domain data tends to suffer. Users of web-based translation services expect high quality translation across a wide range of diverse domains, and what makes the task even more difficult is that no domain label is provided with the translation request.

In this paper the authors present an approach to domain adaptation which results in large-scale, general purpose machine translation systems. First, we tune our translation models to multiple individual domains. Then, by means of source-side domain classification, the authors are able to predict the domain of individual input sentences and thereby select the appropriate domain-specific model parameters. The authors call this approach multi-domain translation. The authors develop state-of-the-art, domain-adapted translation engines for three broadly-defined domains: TED talks, Europarl, and News. Their results suggest that multi-domain translation performs better than a mixed-domain approach, which deploys a system that has been tuned on a development set composed of samples from many domains. (Link)

Read More
01/11

TraMOOC at a Master's course of Tilburg University

TraMOOC at a Master's course of Tilburg University

TraMOOC made a presentation titled "Social Intelligence" within the framework of a Master's course at Tilburg University, Tilburg, The Netherlands in November 2015.

03/12

TraMOOC at the IWSLT

TraMOOC at the 12th International Workshop on Spoken Language Translation (IWSLT)

TRAMOOC was present at the 12th International Workshop on Spoken Language Translation (IWSLT), which was held between the 3rd and the 4th of December, 2015 at Da Nang, Vietnam, presenting the paper titled "The Edinburgh Machine Translation Systems for IWSLT 2015". 

In this paper the authors describe the University of Edinburgh’s machine translation (MT) systems for the IWSLT 2015 evaluation campaign. The authors’ submissions are based on preliminary systems which are under development for the purpose of lecture translation in the TraMOOC project, funded by the European Union. The authors participated in the English→Chinese and the English→German translation tasks in the MT track, utilizing only data supplied by the organizers or listed as permissible. They built phrase-based translation systems for both tasks. For English→German, the authors made use of syntax-based translation and system combination. For more information please follow this link.

Read More
16/12

TraMOOC in the EASN Newsletter

TraMOOC in the EASN Newsletter

The TraMOOC project is in its 10th month of activities and the partners, after having developed the first Machine Translation (MT) prototype, are going full steam ahead its integration into the Iversity platform.

More specifically, during this period the responsibilities of each partner and the language resources and tools to be employed/adapted/developed in TraMOOC have been defined and agreed providing a timeline for the project activities. The timeline is organised around the machine translation prototypes which will be produced within the project, thus making sure that the language resources and tools needed for each prototype will be ready on time. Additionally, a focused survey on existing and readily-available MT infrastructure, which could be employed within TraMOOC for the 11 languages targeted, has been performed. This includes statistical machine translation systems, parallel and monolingual corpora (both in- and out-of-domain), and language tools, the employment of which could be beneficial for the machine translation quality. Finally, during this period the first MT prototype has been developed which provides four of the final eleven language pairs and will be used to perform initial implicit and explicit evaluations for estimating the accuracy of the translated text. It is worth highlighting that the first MT prototype will be succeeded by the second version of the TraMOOC translation prototype in June 2016 (Link).

Read More
18/12

TraMOOC at the Computational Linguistics in the Netherlands (CLIN)

TraMOOC at the Computational Linguistics in the Netherlands (CLIN)

TRAMOOC was present at the Computational Linguistics in the Netherlands (CLIN), which was held on the 18th of December, 2015 at Amsterdam, The Netherlands, presenting the papers titled "Implicit evaluation for machine translation” and “Wikification of MOOCs". This event was a great opportunity for increasing awareness about the scope and objectives of the project.

2016

TraMOOC Progress Timeline

25/01

Initial Dissemination and Exploitation Plan

The Initial Dissemination and Exploitation Plan for the project results has been constructed

The consortium’s initial strategy and concrete actions to disseminate, exploit and protect the foreground generated within the project has been constructed. In particular, dissemination activities which demonstrate the added-value and positive impact of the project on the European Union have been planned to be performed within (February 2015 – January 2018) and after the end of the project. It is worth mentioning that as the resources dedicated to dissemination are restricted, cost-effective ways have been chosen to achieve the maximum of publicity for the project and its results. Also, the strategic exploitation plan to be carried out jointly from Consortium partners or by individual partners has been developed. The initial exploitation scenarios, the plans of industrial partners in the Consortium, and the IPR regulating issues have been defined.

26/01

Initial TraMOOC Business Plan

The Initial TraMOOC Business Plan has been developed

An initial planning on how to bring the TraMOOC solution closer to market, allowing the technology and know-how generated within the project to be used and further developed beyond the project lifetime, has been made. The initial business streams, business models and business solutions for the TraMOOC platform have been defined. The plan takes into account the different market segments which can benefit from the use of the TraMOOC platform. Finally, a series of SWOT analyses for the use of the platform and the related business models have also been prepared. 

27/01

System Prototype Version 1

System Prototype Version 1 is available

An initial prototype of the automated translation functionality on iversity, connecting the iversity platform and its content with the TraMOOC translation engine has become available. The objectives of this prototype are the following: i) to draw conclusions towards further refining the integration (both for API and client), ii) to provide the consortium with information on the usability of such functionality, as experienced directly on a live website, and iii) to gather feedback regarding the quality of translations. This prototype will also serve as a basis for future extended implementation and integration efforts.

The prototype deals with the 4 languages which are processed by Machine Translation prototype 1, i.e., German, Italian, Greek, and Portuguese. From the user perspective, the integration consists of a translation feature on a section of text, including a language switcher and a button, which sends a translation request to the TraMOOC API. On the successful return of the translated text, the original source text on the site is replaced by the translation. The initial implementation applies the translation feature to announcements in courses. 

Read More
28/01

Machine Translation Version 1

Machine Translation Version 1 is available

Baseline phrase-based machine translation systems have been developed which translate from English into German, Italian, European Portuguese, and Greek and have been put online as Translation Prototype Version 1. Also, two alternative translation solutions have been developed in order to compare these with the phrase-based baseline models. Experiments have been performed with syntax-based (for German, Chinese and Czech) and neural machine translation models (for German and Czech). Extensive domain adaptation experiments have been also performed for the aforementioned four phrase-based models.

29/01

User Requirements Analysis

User Requirements Analysis completed

User data have been gathered in order to define the TraMOOC use cases and user requirements and, through them, a first version of the TraMOOC System Architecture has become available. The definition of the use cases has led to a set of requirements, which have led to the specification of the overall architecture. Besides the System Architecture, the basic TraMOOC functionality and the related system workflows have been also defined.

30/01

TraMOOC at the UNESCO 2nd OER World Congress

TraMOOC at the UNESCO 2nd OER World Congress

TraMOOC was present at the UNESCO 2nd OER World Congress, which was held between the 28th and the 30th of January 2016 at Paris, France, making a presentation titled "TraMOOC, an asset for global education and UNESCO".

01/05

TraMOOC at the 10th edition of the Language Resources and Evaluation Conference

TraMOOC at the 10th edition of the Language Resources and Evaluation Conference

TraMOOC was present at the 10th edition of the Language Resources and Evaluation Conference, which took place in May 2016 at Portorož, Slovenia, presenting the paper titled "PE2rr Corpus: Manual Error Annotation of Automatically pre-annotated MT Post-edits".

This paper presents a freely available corpus containing source language texts from different domains along with their automatically generated translations into several distinct morphologically rich languages, their post-edited versions, and error annotations of the performed post-edit operations. The authors believe that the corpus will be useful for many different applications. The main advantage of the approach used for creation of the corpus is the fusion of post-editing and error classification tasks, which have usually been seen as two independent tasks, although naturally they are not.  For more information, please follow this link.

Read More
10/05

TraMOOC Meeting

TraMOOC Meeting

On the 10th and 11th of May 2016 the TraMOOC Consortium conducted successfully the M15 meeting in Dublin, Ireland where the progress performed during the first 15 months of the project was presented to the consortium. The aim of the M15 project meeting was to review the progress of each WP and the activities planned up to the Review Meeting in September 2016.

In a nutshell, important technical work has been performed during this period and all WPs have achieved most of this period’s objectives. More information about the project's progress and latest results is available on the project's progress timeline. For more information please visit the project's website here.

Read More
23/05

TraMOOC at LREC 2016

TraMOOC at the 10th edition of the Language Resources and Evaluation Conference

TRAMOOC was present at the 10th edition of the Language Resources and Evaluation Conference, which was held between the 23rd and the 28th of May, 2016 at Portoloz, Slovenia, presenting the paper titled "Enhancing Access to Online Education: Quality Machine Translation of MOOC Content".

This paper presents an overview of the TraMOOC (Translation for Massive Open Online Courses) research and innovation project, which is a machine translation approach for online educational content. For more information please follow this link.

Read More
30/05

TraMOOC at EAMT 2016

TraMOOC at the 19th Annual Conference of the European Association for Machine Translation (EAMT 2016)

TraMOOC was present at the 19th Annual Conference of the European Association for Machine Translation (EAMT 2016) which was held at Riga, Latvia from the 30th of May till the 1st of June. The TraMOOC poster presentation gave an overview of the project's aims and objectives and progress and achievements. In brief, the TraMOOC consortium has completed an analysis of in-domain and out-of-domain data available in each of the project’s language pairs. More information can be found here.

02/06

TraMOOC at the 19th Annual Conference of the European Association of Machine Translation

TraMOOC at the 19th Annual Conference of the European Association of Machine Translation

TraMOOC was present at the 19th Annual Conference of the European Association of Machine Translation, which took place in June 2016 at Riga, Latvia, presenting the paper titled "Potential and Limits of Using Post-edits as Reference Translations for MT Evaluation".

This work investigates the potential use of post-edited machine translation (MT) outputs as reference translations for automatic machine translation evaluation, focusing mainly on the following important question: Is it necessary to take into account the machine translation system and the source language from which the given post-edits are generated? For more information, please follow this link.

Read More
08/06

The parallel data on which the MT will be trained have been collected and prepared

The parallel data on which the MT will be trained have been collected and prepared

The parallel data on which the MT will be trained have been collected and prepared. The data will be used in the development of MT systems in work package 4. The language tools are also available, the employment of which could be beneficial for the machine translation quality. Finally, potential issues and challenges with the infrastructure created within work package 3 have been identified and analysed, many of which arise from the specific application domain of the project, namely MOOCs.

21/06

TraMOOC in the EASN Newsletter

TraMOOC in the EASN Newsletter

TraMOOC, being in its 2nd year of development, conducted on the 10th and 11th of May 2016 the M15 meeting in Dublin, Ireland where the progress performed during the first 15 months of the project was presented to the consortium. The aim of the M15 project meeting was to review the progress and the activities planned up to the Review Meeting in September 2016.

The highlights of this period include: i) the development of the translation prototype version 1, ii) the creation of the initial prototype of the automated translation functionality on iversity, connecting the iversity platform and its content with the TraMOOC translation engine, and iii) the creation of the Machine Translation Version 1 (baseline phrase-based machine translation systems) which translates from English into German, Italian, European Portuguese and Greek. In a nutshell, important technical work has been performed during this period and most of this period’s objectives have been achieved. More information about the project's progress, latest results and performed dissemination activities is available on the project's progress timeline. (Link)

Read More
27/07

Final System Architecture defined

Final System Architecture defined

This  second  WP2 deliverable  reports  on  updates  in  the  overall TraMOOC use  cases, functional  and  non-functional  requirements, System  Architecture and  basic  workflows  as these  were  reported  in  D2.1.  Moreover,  system-wide requirements  are mapped  onto  the architectural blocks of TraMOOC, showcasing how the system architecture plans to achieve the users' and project's ambitious goals. 

07/08

TraMOOC at ACL 2016

TraMOOC at the 54th Annual Meeting of the Association for Computational Linguistics (ACL2016)

TraMOOC was present at the 54th Annual Meeting of the Association for Computational Linguistics (ACL), which took place at the Humboldt University, Berlin, in August 7-12, 2016, presenting three scientific papers during the ACL2016 First Conference on Machine Translation (WMT16). TraMOOC was also hosting a dedicated booth where three project related posters were displayed and leaflets were available for distribution.

11/08

TraMOOC at the ACL 2016 First Conference on Machine Translation (WMT16)

TraMOOC at the ACL 2016 First Conference on Machine Translation (WMT16)

TraMOOC was present at the 54th Annual Meeting of the Association for Computational Linguistics (ACL), which took place at the Humboldt University, Berlin, in August 7-12, 2016, presenting the paper titled "Edinburgh Neural Machine Translation Systems for WMT 16".

The authors participated in the WMT 2016 shared news translation task by building neural translation systems for four language pairs, each trained in both directions: English↔Czech, English↔German, English↔Romanian and English↔Russian. Their systems are based on an attentional encoder-decoder, using BPE subword segmentation for open-vocabulary translation with a fixed vocabulary. The authors experimented with using automatic back-translations of the monolingual News corpus as additional training data, pervasive dropout, and target-bidirectional models. All reported methods gave substantial improvements, and improvements of 4.3–11.2 BLEU were achieved over the baseline systems. In the human evaluation, their systems were the (tied) best constrained system for 7 out of 8 translation directions in which they participated. For more information please follow this link.

Read More
12/08

TraMOOC at the ACL 2016 First Conference on Machine Translation (WMT16)

TraMOOC at the ACL 2016 First Conference on Machine Translation (WMT16)

TraMOOC was present at the 54th Annual Meeting of the Association for Computational Linguistics (ACL), which took place at the Humboldt University, Berlin, in August 7-12, 2016, presenting the paper titled "The AMU-UEDIN Submission to the WMT16 News Translation Task: Attention-based NMT Models as Feature Functions in Phrase-based SMT".

This paper describes the AMU-UEDIN submissions to the WMT 2016 shared task on news translation. For more information, please follow this link.

Read More
13/08

TraMOOC at the ACL 2016 First Conference on Machine Translation (WMT16)

TraMOOC at the ACL 2016 First Conference on Machine Translation (WMT16)

TraMOOC was present at the 54th Annual Meeting of the Association for Computational Linguistics (ACL), which took place at the Humboldt University, Berlin, in August 7-12, 2016, presenting the paper titled "Edinburgh’s Statistical Machine Translation Systems for WMT16".

This paper describes the University of Edinburgh’s phrase-based and syntax-based submissions to the shared translation tasks of the ACL 2016 First Conference on Machine Translation (WMT16). The authors submitted five phrase-based and five syntaxbased systems for the news task, plus one phrase-based system for the biomedical task. For more information, please follow this link.

Read More
14/08

TraMOOC at the Association for Computational Linguistics

TraMOOC at the Association for Computational Linguistics

TraMOOC was present at the Association for Computational Linguistics, which took place in Augist 2016 at Berlin, Germany, presenting the paper titled "chrF deconstructed: beta parameters and n-gram weights".

In this work, the authors investigated CHRF in more details, namely β parameters in range from 1/6 to 6, and they found out that CHRF2 is the most promising version.  For more information, please follow this link.

Read More
16/08

TraMOOC at the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016) First Conference on Machine Translation (WMT16)

TraMOOC at the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016) First Conference on Machine Translation (WMT16)

TraMOOC was present at the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016) First Conference on  Machine Translation (WMT16), which took place in August 2016 at Berlin, Germany, making a presentation titled "The AMU-UEDIN Submission to the WMT16 News Translation Task: Attention-based NMT Models as Feature Functions in Phrase-based SMT".

17/08

TraMOOC at the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016) First Conference on Machine Translation (WMT16)

TraMOOC at the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016) First Conference on Machine Translation (WMT16)

TraMOOC was present at the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016) First Conference on  Machine Translation (WMT16), which took place in August 2016 at Berlin, Germany, making a presentation titled "Edinburgh’s Statistical Machine Translation Systems for WMT16".

18/08

TraMOOC at the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016) First Conference on Machine Translation (WMT16)

TraMOOC at the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016) First Conference on Machine Translation (WMT16)

TraMOOC was present at the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016) First Conference on  Machine Translation (WMT16), which took place in August 2016 at Berlin, Germany, making a presentation titled "Edinburgh Neural Machine Translation Systems for WMT 16".

31/08

Translation Prototype Version 2 is available

Translation Prototype Version 2 is available

TraMOOC (Translation for Massive Open Online Courses) is a Horizon 2020 collaborative project aiming at providing reliable machine Translation for Massive Open Online Course (MOOCs). The goal of this project is to overcome language barriers and enable people around the world to access leading educational resources. In particular,the project aims at developing an innovative,high-quality machine translation (MT) service for producing accurate translations of multi-genre and heterogeneous textual course materialvarying from videolectures,assignments,tutorial text to user-generated content posted on MOOC blogs and fora.

This deliverable documents the TraMOOC demo: the translation prototype version 2. This demo is the third deliverable of the machine translation workpackage,WP4. The prototype is provided for eleven language pairs and it will be used to perform initial implicit and explicit evaluations in WP5 and WP6.This report accompanies the deliverable. Its purpose is to detail the API to the demo's translation models. It also describes the data used to train the models,and performance results.

Read More
01/09

The impact and continuation plans for TraMOOC services and software are available

The impact and continuation plans for TraMOOC services and software are available

The impact and continuation plans for TraMOOC services and software have been developed. These plans describe the current status, plans and mechanisms for further impact in order to ensure that the services developed within TraMOOC are available within the project and after project’s end. These plans also explore the possibility for further development of a community around the created open-source software and is being completed in synergy with the dissemination, exploitation and business strategy.

15/09

TraMOOC at the 8th EST Congress Translation Studies: Moving Boundaries

TraMOOC at the 8th EST Congress Translation Studies: Moving Boundaries

TraMOOC was present at the 8th EST Congress Translation Studies: Moving Boundaries, which took place in September 2016 at Aarhus, Denmark, making a presentation titled "Translation as a Microtask: Investigating a Paradox".

27/09

TraMOOC Mid-Term Review Meeting, Luxembourg, September 27th, 2016

TraMOOC Mid-Term Review Meeting, Luxembourg, September 27th, 2016

On the 27th of September 2016 the TraMOOC Consortium conducted successfully the M18 mid-term review meeting in Luxembourg where the progress performed during the first half of the project was presented to the European Commission Project Officer and the EC reviewers who provided their valuable feedback to the TraMOOC partners. 

The aim of the M18 mid-term review meeting was to highlight the scientific and technical achievements of the project and provide sufficient information to allow the EC to perform the project assessment against the applicable Description of Action. The highlights of this period include: The definition of the quality risk and management mechanisms The definition of the project-level management mechanisms of the language resources and tools The creation of the TraMOOC Intranet and public project website The development of the Data Management Plan of the project The creation of the project Initial Communication Plan The development of the translation prototype version 1 The creation of the Initial Dissemination and Exploitation Plan for the project results The development of the Initial TraMOOC Business Plan The creation of the initial prototype of the automated translation functionality on iversity, connecting the iversity platform and its content with the TraMOOC translation engine The creation of the Machine Translation Version 1 (baseline phrase-based machine translation systems) which translates from English into German, Italian, European Portuguese and Greek The analysis and definition of the user requirements  In a nutshell, important technical work has been performed during this period and all WPs have achieved most of this period’s objectives. More information about the project's progress and latest results is available on the project's progress timeline.

Read More
29/09

TraMOOC at the 10th conference on Language Technologies and Digital Humanities

TraMOOC at the 10th conference on Language Technologies and Digital Humanities

TraMOOC was present at the 10th conference on Language Technologies and Digital Humanities, which took place in September 2016 at Ljubljana, Slovenia, presenting the paper titled "Asistent – A Machine Translation System for Slovene, Serbian and Croatian".

The META-NET research on language technologies in 2012 showed a weak support on tools for crossing the language barrier for many European languages, including the south Slavic languages. Therefore, the authors describe a statistical machine translation system, called Asistent, which enables automatic translations between English, Slovene, Croatian and Serbian. In addition to make this system publicly accessible, the authors focused on parallel data preparation as well as on using multiple pivot languages for translation quality improvement of the targeted Slavic languages. A comparison of translations generated by the Asistent translation system showed a significant improvement of translation quality over Google Translate. For more information, please follow this link.

Read More
03/10

TraMOOC is live on Twitter!!!

TraMOOC is live on Twitter!!!

Follow us on TraMOOC's twitter account here!!!

13/10

TraMOOC at the ELIA's 20th Networking Days

TraMOOC at the ELIA's 20th Networking Days

TraMOOC was present at the ELIA's 20th Networking Days, which took place in October 2016 at Brussels, Belgium, making a presentation titled "Microtask Translation Workflows".

20/10

TraMOOC at the 11th International Workshop on Semantic and Social Media Adaptation and Personalization

TraMOOC at the 11th International Workshop on Semantic and Social Media Adaptation and Personalization

TraMOOC was present at the 11th International Workshop on Semantic and Social Media Adaptation and Personalization, which took place in October 2015 at Thessaloniki, Greece, making a presentation titled "Social Media and NLP tasks: Challenges in Crowdsourcing Linguistic Information".

21/10

TraMOOC at ATILA 2016

TraMOOC at ATILA 2016

TraMOOC was present at the ATILA 2016 Conference, which took place in October 2016 at Ravenstein,The Netherlands, making a presentation titled "A crowdsourcing annotation interface for Wikification in 11 language pairs".

24/10

TraMOOC at Languages and the Media 2016

TraMOOC at Languages and the Media 2016

Mrs. Georgakopoulou (DME) gave a very interesting interview in the  "languagesmedia Newsletter"  highlighting the significant results achieved so far within the frame of the project as well as the expected impacts. The interview is available online here.  

28/10

TraMOOC at Translating Europe Forum 2016

TraMOOC at Translating Europe Forum 2016

TraMOOC was present at the Translating Europe Forum 2016, which took place in October 2016 at Brussels, Belgium, making a presentation titled "Quality Machine Translation of Massive Open Online Courses".

02/11

TraMOOC at the 11th International Conference on Language Transfer in Audiovisual Media

TraMOOC at the 11th International Conference on Language Transfer in Audiovisual Media

TraMOOC was present at the 11th International Conference on Language Transfer in Audiovisual Media, which took place in November 2016 at Berlin, Germany, making a presentation titled "Crowdsourcing for the Creation of Parallel Translation Corpora: The Case of TraMOOC".

17/11

TraMOOC at the 38th edition of the annual Translating and the Computer Conference (TC38)

TraMOOC at the 38th edition of the annual Translating and the Computer Conference (TC38)

TraMOOC was present at the 38th edition of the annual Translating and the Computer Conference (TC38), which took place in November 2016 at London, UK, making a presentation titled " A Comparative Evaluation of Phrase-Based SMT and Neural Machine Translation".

11/12

TraMOOC at 26th International Conference on Computational Linguistics

TraMOOC at 26th International Conference on Computational Linguistics

TraMOOC was present at 26th International Conference on Computational Linguistics, which took place in December 2016 at Osaka, Japan, presenting the paper titled "Enlarging scarce in-domain English-Croatian corpus for SMT of MOOCs using Serbian".

In this work, the authors investigated statistical machine translation of lecture subtitles from English into Croatian, which is morphologically rich and generally weakly supported, especially for the educational domain.The authors showed that results comparable with publicly available systems trained on much larger data can be achieved if a small in-domain training set is used in combination with additional in-domain corpus originating from the closely related Serbian language. For more information, please follow this link.

Read More
16/12

TraMOOC at the Third Workshop on NLP for similar Languages, Varieties and Dialects (VarDial3)

TraMOOC at the Third Workshop on NLP for similar Languages, Varieties and Dialects (VarDial3)

TraMOOC was present at the Third Workshop on NLP for similar Languages, Varieties and Dialects (VarDial3), which took place in December 2016 at Osaka, Japan, presenting the paper titled "Language Related Issues for Machine Translation between Closely Related South Slavic Languages".

This work explores the obstacles for machine translation systems between closely related South Slavic languages, namely Croatian, Serbian and Slovenian. Statistical systems for all language pairs and translation directions are trained using parallel texts from different domains, however mainly on spoken language i.e. subtitles. For translation between Serbian and Croatian, a rule-based system is also explored. It is shown that for all language pairs and for both translation systems, the main obstacles are the differences between syntactic properties. For more information, please follow this link.

Read More
2017

TraMOOC Progress Timeline

09/01

TraMOOC at the January 2017 issue of the EASN Newsletter

TraMOOC at the January 2017 issue of the EASN Newsletter

TraMOOC, being in its 2nd year of development, conducted successfully on the 27th of September 2016 the M18 mid-term review meeting in Luxembourg where the progress performed during the first half of the project was presented to the European Commission Project Officer and the EC reviewers who provided their valuable feedback. The highlights of this period include the initial integration of the Machine Translation systems in a real-live MOOC platform which proved the viability of the work performed within the project and the thorough comparison of phrase-based, syntax-based, and neural machine translation algorithms on MOOC-related test data. 

In terms of disseminating the important research results of the project, during the last months TraMOOC was present at the Translating Europe Forum 2016, the 11th International Workshop on Semantic and Social Media Adaptation and Personalization (SMAP 2016), the Languages and the Media 2016 and the 12th IFIP International Conference on Artificial Intelligence Applications and Innovations (AIAI 2016). In a nutshell, important technical work has been performed during this period and most of this period’s objectives have been achieved. More information about the project's progress and latest results is available on the project's progress timeline. You may also follow us on our social media groups and pages (Facebook, Twitter, LinkedIn, YouTube).Link

Read More
16/01

TraMOOC M24 Progress Meeting in Athens, Greece, January 16-17, 2017

TraMOOC M24 Progress Meeting in Athens, Greece, January 16-17, 2017

On the 16th and 17th of January 2017 the TraMOOC Consortium conducted successfully the M24 meeting in Athens, Greece where the progress performed after the M18 review meeting was presented to the consortium. The aim of the M24 project meeting was to review the progress of each WP and the activities planned up to the next project meeting in June 2017.

16/01

TraMOOC M24 Progress Meeting in Athens, Greece, January 16-17, 2017

TraMOOC M24 Progress Meeting in Athens, Greece, January 16-17, 2017

On the 16th and 17th of January 2017 the TraMOOC Consortium conducted successfully the M24 meeting in Athens, Greece where the progress performed after the M18 review meeting was presented to the consortium. The aim of the M24 project meeting was to review the progress of each WP and the activities planned up to the next project meeting in June 2017.

The highlights of this period include the initial integration of the Machine Translation systems in a real-live MOOC platform which proved the viability of the work performed within the project and the thorough comparison of phrase-based, syntax-based, and neural machine translation algorithms on MOOC-related test data. In terms of disseminating the important research results of the project, during the last months TraMOOC was present at the Translating Europe Forum 2016, the 11th International Workshop on Semantic and Social Media Adaptation and Personalization (SMAP 2016), the Languages and the Media 2016 and the 12th IFIP International Conference on Artificial Intelligence Applications and Innovations (AIAI 2016). In a nutshell, important technical work has been performed during this period and most of this period’s objectives have been achieved.

Read More
26/01

Intermediate Dissemination and Exploitation plan of the project results is available

Intermediate Dissemination and Exploitation plan of the project results is available

In Year 2 (M13-M24) of the project, dissemination activities were focused on amplifying the project’s scientific impact, through participations at conferences and scientific publications; these activities were mainly addressed by the academic partners. At the same time, activities also sook to promote the TraMOOC platform developed within the project, thus pursuing exploitation opportunities. These activities were mainly driven by SME partners, who work towards pushing the commercial applications of this technology. It is important to observe how dissemination actions were often realized as joint undertakings, achieving a complete and unified project presence in various activities (joint publications, joint exhibits, etc.). This justifies the active involvement of the entire consortium in communicating the project’s results.

27/01

Extensive evaluation of different MT technologies has been finalized

Extensive evaluation of different MT technologies has been finalized

Extensive evaluation of different MT technologies has been finalized, including phrase-based,syntax-based,neural,and hybrid phrase-based/neural approaches.We participated in the WMT2016 shared evaluation campaign,where our neural systems achieved top-ranked performance,outperforming other research systems and commercial systems.These results are the basis of our decision to focus on neural MT for the TraMOOC MT systems,including translation prototype 2. 

Apart from translation quality,where our results support the use of neural MT,one important consideration is ease of deployment.We evaluated the efficiency of phrase-based and neural MT,and developed a highly efficient inference engine for neural MT.A major activity in year 2 was the creation of the second translation prototype,which supports all eleven TraMOOC language pairs,and is based on neural MT.The second prototype is deployable via Docker containers,and provides an interface for integration into the TraMOOC platform.

Read More
10/02

TraMOOC @ CLIN 27

TraMOOC @ CLIN 27

TraMOOC was present at the 27th Meeting of Computational Linguistics in The Netherlands (CLIN 27), which was organised by the Centre for Computational Linguistics group of KU Leuven and took place at the Faculty of Arts in Leuven on February 10, 2017. 

The poster entitled "Wikification for Implicit MT evaluation, First results for Dutch" was displayed by Iris Hendrickx  (Radboud University). Based on the authors of this work machine translation evaluation requires a human in the loop: either for providing gold-standard written translations or for human judgements. As such human effort is expensive and time-consuming, the authors aim to look at methods to evaluate translation quality without human supervision. They focus on topics and entities in the text and make smart use of the inter-lingual connections in Wikipedia to determine whether the MT system managed to translate the entity correctly. They use a Wikifier (Ratinov et al, 2011) to automatically find entities with Wikipedia links in the English text and compare them against the linked entities in the target language's equivalent Wikipedia page. We present the first outcomes of this method on Dutch MOOC data and discuss the limits of this approach such as Wikipedia coverage and Wikifier errors.

Read More
02/03

TraMOOC at the Content Supply Chain & Workflow Management Forum

TraMOOC at the Content Supply Chain & Workflow Management Forum

TraMOOC was present at the Content Supply Chain & Workflow Management Forum, which took place in March 2017 at London, UK making a presentation titled "TraMOOC: SMT vs. NMT evaluation".

03/04

TraMOOC at the 15th Conference of the European Association for Computational Linguistics (EACL 2017)

TraMOOC at the 15th Conference of the European Association for Computational Linguistics (EACL 2017)

TraMOOC was present at the 15th Conference of the European Association for Computational Linguistics (EACL 2017), which took place in April 2017 at Valencia, Spain presenting the paper titled "Nematus: a Toolkit for Neural Machine Translation".

This work presents  Nematus,  a  toolkit  for  Neural Machine Translation.  The toolkit prioritizes high translation accuracy,  usability,  and  extensibility.   Nematus  has  been used to build top-performing submissions to  shared  translation  tasks  at  WMT  and IWSLT, and has been used to train systems for production environments. For more information, please follow this link.

Read More
04/04

TraMOOC at the QT21 Workshop on MQM-based evaluation at EACL 2017 - 15th Conference of the European Chapter of the Association for Computational Linguistics

TraMOOC at the QT21 Workshop on MQM-based evaluation at EACL 2017 - 15th Conference of the European Chapter of the Association for Computational Linguistics

TraMOOC was present at the QT21 Workshop on MQM-based evaluation at EACL 2017 - 15th Conference of the European Chapter of the Association for Computational Linguistics, which took place in April 2017 at Valencia, Spain making a presentation titled "TraMOOC: Translation for Massive Open Online Courses".

25/04

TraMOOC at the April 2017 issue of the EASN Newsletter

TraMOOC at the April 2017 issue of the EASN Newsletter

TraMOOC, by having achieved most of the objectives planned for its second year of activities, is moving ahead to its final 10 months of activities. 

The highlights of this period include the development of machine translation prototypes for 11 target languages, from English into German, Italian, Portuguese, Dutch, Bulgarian, Greek, Polish, Czech, Croatian, Russian and Chinese. The translation systems are based on phrase-based SMT and Neural Machine Translation. The latter has achieved state-of-the-art performance in recent evaluation campaigns. The translation systems have been adapted to MOOC texts to maximize translation quality on this domain. During this period, a comparative human evaluation of phrase-based SMT and NMT has been completed for four language pairs to compare educational domain output from both systems using a variety of metrics. These include automatic evaluation, human rankings of adequacy and fluency, error-type markup, and post-editing effort. The results show a preference for NMT in side-by-side ranking for all language pairs, texts, and segment lengths. In a nutshell, important technical work has been performed during this period and most of this period’s objectives have been achieved. More information about the project's progress and latest results is available on the project's progress timeline. You may also follow us on our social media groups and pages (Facebook, Twitter, LinkedIn, YouTube). Link

Read More
26/04

Human and automatic translation evaluation completed

Human and automatic translation evaluation completed

TraMOOC (Translation for Massive Open Online Courses) is a Horizon 2020 collaborative project aiming at providing reliable machine translation (MT) for Massive Open Online Courses (MOOCs). The goal of this project is to overcome language barriers and enable people around the world to access leading educational resources. In particular, the project aims at developing an innovative, high-quality MT service for producing accurate translations of multi-genre and heterogeneous course materials, ranging from video lectures, assignments, tutorial texts to social web text posted on MOOC blogs and fora.

This first deliverable of the explicit evaluation work package (WP5) reported a comparative evaluation between the two machine translation prototypes (v1 and v2). The evaluation was carried out in close collaboration with Deluxe Media. Professional translators were asked to post-edit and evaluate the MT output in a range of target languages from the different prototypes in order to evaluate whether a shift to Neural MT (NMT) models is justified. 

Read More
22/05

TraMOOC at the Fifth European MOOCs Stakeholders Summit (EMOOCs 2017 Conference)

TraMOOC at the Fifth European MOOCs Stakeholders Summit (EMOOCs 2017 Conference)

TraMOOC was present at the Fifth European MOOCs Stakeholders Summit (EMOOCs 2017 Conference), which took place in May 2017 at Madrid, Spain making a presentation titled "MOOC tailored content translation service boosting revenues and business intelligence".

30/05

TraMOOC at the 20th Annual Conference of the European Association for Machine Translation (EAMT2017)

TraMOOC at the 20th Annual Conference of the European Association for Machine Translation (EAMT2017)

TraMOOC was present at the 20th Annual Conference of the European Association for Machine Translation (EAMT 2017) which is held at Prague, Czech Republic from the 29th till the 31st of May presenting a poster entitled "TraMOOC - Translation for Massive Open Online Courses: Recent Developments in Machine Translation", a presentation entitled "Comparing language related issues for NMT and PBMT between German and English" and  a presentation entitled "Is Neural Machine Translation the New State-of-the-Art?".     

The TraMOOC poster presentation gives an overview of the project's aims and objectives and progress and achievements, while the TraMOOC related presentation entitled "Comparing language related issues for NMT and PBMT between German and English" presents an extensive comparison of language related problems for neural machine translation and phrase-based machine translation between German and English. Finally, the TraMOOC related presentation entitled "Is Neural Machine Translation the New State-of-the-Art?" discusses neural machine translation (NMT), a new paradigm in the MT field,comparing the quality of NMT systems with statistical MT by describing three studies using automatic and human evaluation methods. 

Read More
31/05

Crowdsourcing Platform Configuration

Crowdsourcing Platform Configuration

TraMOOC  (Translation  for  Massive  Open  Online  Courses)  aims  at  developing  high-quality machine  translation  technology  for  online educational  content.  A  significant  part  of  the development  process  relies  on  crowdsourcing  solutions  for  building  and  evaluating  textual resources. In more detail, the crowdsourcing activities include human translation from English into 11 languages, human evaluation of machine-translated output for the translation engines of  the  11  language  pairs,  human  annotation  of  terms/entities  in  11  languages,  and  human sentiment annotation of English text.

This  deliverable  documents the  selection  process  of  the  crowdsourcing  platform used in TraMOOC, namely CrowdFlower, and describes the design and development process of the jobs built for all the aforementioned crowdsourcing activities, as well as the configuration and customization  process  for  setting  the  platform parameters  for  all  activities  and  for  all  11 language pairs. Job-specific and language-specific challenges and limitations are presented, along with the means adopted to address them.

Read More
01/06

D2.3: Crowdsourcing Platform Configuration has been delivered

D2.3: Crowdsourcing Platform Configuration has been delivered

The  deliverable (D2.3) regarding the selection  process  of  the  crowdsourcing  platform used in TraMOOC, namely CrowdFlower, and the design and development process of the jobs built for all the aforementioned crowdsourcing activities, as well as the configuration and customization  process  for  setting  the  platform parameters  for  all  activities  and  for  all  11 language pairs has been finalized. 

Job-specific and language-specific challenges and limitations are presented, along with the means adopted to address them.

Read More
09/06

TraMOOC M29 Progress Meeting in Corfu, Greece, June 9th, 2017

TraMOOC M29 Progress Meeting in Corfu, Greece, June 9th, 2017

On the 9th of June 2017 the TraMOOC Consortium conducted the TraMOOC M29 Progress Meeting in Corfu, Greece hosted by the Ionian University (IURC). Within the frame of this meeting the progress achieved so far and the next steps to be taken were presented while proceeding towards the end of the project.

The project plans to create three MT prototypes, the second of which has been developed for 11 target languages: from English into German, Italian, Portuguese, Dutch, Bulgarian, Greek, Polish, Czech, Croatian, Russian, and Chinese. The current translation systems are based on phrase-based SMT and neural machine translation (NMT), with a view to move to NMT for the final prototype. Our NMT systems have achieved state-of-the-art performance in recent evaluation campaigns. These systems use the Nematus toolkit for training, and the translation server is based on the amuNMT toolkit. The translation systems have been adapted to MOOC texts via fine-tuning of the model parameters on in-domain training data to maximize translation quality on this domain. In a nutshell, important technical work has been performed during this period and most of this period’s objectives have been achieved. More information about the project's progress and latest results is available on the related announcement in the news section (https://tramooc.eu/content/tramooc-m29-progress-meeting-corfu-greece-june-9th-2017-0) of the project's public website. You may also follow us on our social media groups and pages (FacebookTwitterLinkedInYouTube).

Read More
15/06

TraMOOC @ the CLS Language & Social Media day

TraMOOC at the CLS Language & Social Media day

TraMOOC was present at the CLS Language & Social Media day which was held at Radboud University, Nijmegen, The Netherlands in June 2017 presenting a poster entitled "Automatically monitoring of MOOC fora for evaluating translation quality".     

The poster describes the results from the evaluation of the translation Quality of a machine-translated Massive Open Online Course (MOOC) by automatically monitoring the sentiment of MOOC fora.

Read More
30/06

D6.1:Report on Crowdsourcing Activity 3 delivered

D6.1:Report on Crowdsourcing Activity 3 delivered

This  document follows  deliverable  D2.3  "Crowdsourcing  Platform  Configuration",  which describes   and   explains the   configuration   and   customization   process   for   setting   the crowdsourcing  platform  parameters  for  all  activities  (including  the  two  annotation  tasks described  here in)  and  for  all  11  language  pairsexamined  in  the  project. This document presents  the  specific  crowdsourcing  experiments  themselves,  the  data  used,  the  task interfaces,as well as the results obtained.

03/07

TraMOOC @ EDULEARN17

TraMOOC @ the 9th annual International Conference on Education and New Learning Technologies (EDULEARN17)

TraMOOC was present at the 9th annual International Conference on Education and New Learning Technologies (EDULEARN17) which was held at Barcelona, Spain from the 3rd till the 5th of July 2017 presenting a paper entitled "Integrating Machine Translation into MOOCs".      

Massive Open Online Courses (MOOCs) offer valuable learning opportunities in several disciplines to many students, regardless of their background, location, and personal circumstances. MOOCs are typically available in the local language that is shared between tutors and students, which has the added bonus of encouraging interactions on social platforms and fora accompanying formal instruction. However, language barriers impede broad use of high-quality MOOC materials across national and language boundaries: English is often chosen as the common language of MOOCs with international users, even though this prevents large groups of users from fully engaging in a fulfilling MOOC experience, thus wasting precious learning opportunities for innumerable potential students around the world. Key lessons learned in the ongoing TraMOOC project that are relevant to the wider community of international professionals with an interest in the multilingual aspects of innovative education and new learning technologies, assessing the pros and cons of integrating state-of-the-art MT into MOOC platforms. In particular, the authors share reflections on how best to integrate MT into MOOCs, to build multilingual communities in which students can truly learn and interact across language barriers.

Read More
16/07

TraMOOC @ The Journal of Specialised Translation

TraMOOC @ The Journal of Specialised Translation

Casting some light on experts’ experience with translation crowdsourcing by V. Sosoni (IURC) This paper reports on an empirical study concerning professional translators’ attitudes towards  and  experience  with  translation  crowdsourcing.  In particular,it  seeks  to explore  how  professional  translators  perceive  translation  crowdsourcing  and  what concerns they raise, if any. It also aims at identifying any problems they may face as crowdworkers  during  the  translation  process. </p> 

The  investigation  takes  place  in  the framework of the TraMOOC (Translation for Massive Open Online Courses) research and innovation project where a crowd of professional translatorsis used for the translation into Greek (EL) of English (EN) MOOC (Massive Open Online Courses) educational data on the CrowdFlower platform with the end goal of using such translations to train and tune a machine  translation  (MT)  system. It  concludes highlighting the  unexpected benefits that crowdsourcing may bring to professional translators. (Link)

Read More
31/07

TraMOOC @ ACL 2017

TraMOOC @ 55th annual meeting of the Association for Computational Linguistics (ACL 2017)

The use of machine translation (MT) has become  widespread  since  statistical  machine translation (SMT) became the dominant paradigm. However, there is growing interest in the research community in the  possibilities  of  neural  machine  translation (NMT) based largely on impressive results in automatic evaluation. 

This paper reports on a comparative human evaluation of phrase-based SMT and NMT in four language pairs, using the PET tool to compare educational domain output from both  systems  using  a  variety  of  metrics. Our results show a preference for NMT in side by  side  ranking  for  all  language  pairs, texts,  and  segment  lengths.   In  addition, perceived  fluency  is  improved  and  annotated errors fewer in the NMT output. Results are mixed for perceived accuracy and for errors of omission, addition, and mis-translation. 

Read More
30/08

CA1 has been completed

CA1 has been completed

The  pilot  trials  as  well  as  the  actual  tasks  of  the  crowdsourcing  activity pertaining to human translation (CΑ1) have been completed (D3.3). 

The experimental setup of the activity and the obtained  results have been obtained for  all  eleven  language pairs  included  in  the TraMOOC project. Challenges that emerged due to language-specific idiosyncrasies, as well as  platform-dependent  deficiencies,  are  pointed  out  and  solutions  for  overcoming  them have been  identified.  Significant  conclusions  as  well  as  good practices  and  guidelines  have  been derived from this activity.  

Read More
07/09

TraMOOC @ EMNLP 2017 (WMT17)

TraMOOC at EMNLP 2017 Second Conference on Machine Translation (WMT17)

TraMOOC was present at the EMNLP 2017 Second Conference on Machine Translation (WMT17) which was held at Copenhagen, Denmark in September 2017 presenting a paper entitled "The University of Edinburgh’s Neural MT Systems for WMT17" as well as a paper entitled "Regularization Techniques for Fine-tuning in Neural Machine Translation". 

19/09

TraMOOC at 16th Machine Translation Summit

TraMOOC at 16th Machine Translation Summit

On the 22nd of September, 2017 the TraMOOC project was present at 16th Machine Translation Summit that was held at Nagoya University in Japan, 

making  two presentations under the titles “A Comparative Quality Evaluation of PBSMT and NMT using Professional Translators” and “TraMOOC - Translation for Massive Open Online Courses: Recent Developments”. Machine Translation (MT) Summit is an international conference, which brings together people from the academic and commercial world developing MT and users of MT systems. The conference overviews state-of-the-art technologies in MT, its major contemporary trends and practical applications. This year’s conference was organized by the Asia-Pacific Association for Machine Translation (AAMT) and overall, the event was a great success.

Read More
30/09

TraMOOC @ the First Conference for professional translators and interpreters in Greece

TraMOOC @ the First Conference for professional translators and interpreters in Greece

TraMOOC was present at the First Conference for professional translators and interpreters in Greece (http://ti-events.org/) which was held at Athens, Greece in September 2017 making a presentation entitled "Machine Translation: Latest developments and contribution to the translation workflow".       

In recent years, the explosion of global content has prompted the translation industry to  focus more  on machin etranslation.  Machine translation  is  the  sub-field  of computational  linguistics  that  studies  the  use  of  software  to  automatically  translate texts  from a natural  language  into  another. Today,  there  are  three available neural  machine translation systems online: Google Translate,   Microsoft   Translator   and   Systran   Pure   Neural   Machine   Translation. Another system is  Moses  SMT (Koehn et al., 2007), a promising, and ever-growing, free software, which is mainly used for research purposes. The EU supports under its funding scheme for research and innovation, Horizon 2020,research projects focusing on machine translation such as QT21(Quality Translation 21), HimL(Health in my Language),  MMT(Modern  Machine  Translation)  and  TraMOOC(Translation  of MOOCs).   At   the   same   time,   the   translation   industry closely follows these developments.  Machine translation  has been  incorporated  into  various CAT  tools, such  as  SDL  Trados  Studio,  MateCat  and  Memsource. As  expected,  these  changes created  a  wave  of  discontent  and  suspicion  among translators.  However, the controversy “human translator versus machine” is  just  a myth. The  MT  output  is  not  of  an  equal  standard  to  anything  the human  brain  can produce, but MT can be an essential tool in the translation lifecycle, giving also rise to new opportunities for translators in the areas of evaluation, post-editing, data analysis and "training" of machin etranslation systems.

Read More
02/10

D5.3 Report on crowdsourcing activity 2 has been delivered

D5.3 Report on crowdsourcing activity 2 has been delivered

The deliverable D5.3 describing the crowdsourcing experiments pertaining to the two linguistic human translation evaluation tasks involved in TraMOOC (Translation for Massive Open Online Courses), involving expert and non-expert translators has been finalised.

Both tasks aim at collecting linguistic metadata that can be used to indirectly provide a novel means to automatically and manually evaluate the output of the machine translation engines developed in TraMOOC. Both translation evaluation tasks are performed for all eleven language pairs involved in the project. This document follows deliverable D2.3 "Crowdsourcing Platform Configuration", which describes and explains the configuration and customization process for setting the crowdsourcing platform parameters for all activities (including the two translation evaluation tasks described herein) and for all 11 language pairs examined in the project. This document presents the specific crowdsourcing experiments themselves, the data used, the task interfaces, as well as the results obtained.

Read More
05/10

TraMOOC at Language Technology Industry Summit

TraMOOC at Language Technology Industry Summit

TraMOOC gladly announces its participation in LT Industry Summit, taking place from 9 to 11 of October, 2017 in Brussels, Belgium.  

 LT Industry Summit is one of the highlights of the European events calendar. It is the yearly point of convergence between the Language Technology Industry, its clients, research partners and policy makers. An overview TraMOOC presentation will be performed in this major event by the project Coordinator Prof. Valia Kordoni under the title “TraMOOC: Translation for Massive Open Online Courses” in the frame of "What is hot in LT? Technology And Research Spotlights" Session, which will be held on the 11th of October 2017 from 12:00 to 13:00. We are looking forward to meeting you there!  

Read More
20/11

TraMOOC at 39th edition of the annual Translating and the Computer Conference (TC39)

TraMOOC at 39th edition of the annual Translating and the Computer Conference (TC39)

On 17th of November, 2017, TraMOOC was present at 39th edition of the annual Translating and the Computer Conference (TC39) in London.

The TC conference series has emerged as a leading forum for users, developers and vendors of Translation Technology tools where latest developments and practices are presented and discussed. It is a distinctive event where translators, interpreters, researchers and business people from translation companies, international organisations, universities and research labs as well as freelance professionals, interact, exchange ideas and discuss current trends. The conference provided a great opportunity to present the high-quality TraMOOC machine translation service to the stakeholders of the translation technology field.

Read More