News and Announcements

25
Nov 2022
NLP Tea & Talk

On Wednesday, November 30, at 15:30, in class EEB 5302, we are waiting for everyone to join our Tea & Talk event, which will be held with the participation of our group coordinator Assoc. Dr. Gülşen Eryiğit!
teatalk2

Tea&Talk2022
Tea&Talk2022



07
Oct 2022
Search Conference

To discuss future plans and visions that are practical and can be implemented for our research group and to identify specific actions which ought to be taken accordingly, we organized a "search conference" with the participation of our undergraduate and graduate students.

SearchConference



06
Oct 2022
Prometeia Banking & Insurance Day

Our Natural Language Processing Group Coordinator Assoc. Prof. Dr. Gülşen Eryiğit made a presentation on "Natural Language Processing in Banking & Finance" as part of the "Prometeia Banking & Insurance Day" event, which included nearly 400 participants and 39 speakers from national and international institutions.

PrometeiaSunum



25
Jun 2022
Best Graduation Project Award

Furkan Pala, one of the undergraduate students in our study group, received the Best Graduation Project Award with his graduation project "Information Extraction from Structured, Semi-Structured, and Unstructured Banking Documents" under the supervision of Assoc. Prof. Dr. Gülşen Eryiğit, one of the faculty members of our group.
FurkanPala2
bitirme



31
May 2022
Celebration and Experience Transfer Meeting

Our graduate student Erif Oral presented her article titled "AMR Alignment for Morphologically-rich and Pro-drop Languages" and Berke Oral presented his article titled ""Fusion of Visual Representations for Multimodal Information Extraction from Unstructured Transactional Documents" while Ali Şentaş presented his article titled "Gamified Crowdsourcing for Idiom Corpora Construction", Fatih Bektaş presented his article titled "Gamification of complex morphology learning: the case of Turkish" and shared their experiences during their research at the meeting attended by our teammates from undergraduate, graduate and doctoral levels. 



31
May 2022
Multi - Modal Information Extraction

document-types
Oral, B., Eryiğit, G. Fusion of visual representations for multimodal information extraction from unstructured transactional documentsIJDAR 25, 187–205 (2022). https://doi.org/10.1007/s10032-022-00399-3



27
May 2022
İTÜRK

This text is an automatic translation from the its original Turkish version:

AI-Powered Language Learning Technology for Agglutinative Languages

As we know, words in Turkish are formed by putting various suffixes one after the other on some noun or verb roots. For example, in a sentence such as ""Hala yakınlaşamadığımız kişilerden misin?"" ("Are you one of those with whom we still cannot get close?"), it is seen that many suffixes are added to the word "yakın" ("close"). One of the most challenging points for foreign language learners trying to learn our language is  the order in which such suffixes should be used and the different phonetic harmony rules.

In recent years, many language learning applications have been released (e.g.: Duolingo, Busuu, etc.). However, these applications are generally developed by considering languages ​​such as English-German in the first place. And basically, they do some interlingual translation exercises using some sentences prepared by hand. However, such exercises do not work well for languages ​​with very complex and long suffixes, such as Turkish. Language learners need to practice a lot, especially on suffix and sound harmony.

At this point, the idea that learning Turkish and similar languages ​​could be facilitated by using artificial intelligence technologies was put forward, and this idea was presented to foreign language learners for the first time in the world by researchers from ITU in a gamified mobile application environment. This technology, in which the exercises are created dynamically, so that the student can practice with different words each time, has been published in the past months by "Computer Assisted Language Learning", one of the leading international journals in this field. A patent application has been made for the project .The project has been led by ITU Artificial Intelligence and Data Engineering faculty member Assoc. Gülşen Eryiğit, the coordinator of the ITU Natural Language Processing Laboratory.

Currently, studies are continuing to support different languages ​​of the application and to increase the supported structures in Turkish. The first prototype tests of the study, which brought an innovative approach to language learning, were made on foreign students and the study was highly appreciated.

New foreign students who will start learning Turkish at the ITU Turkish Learning Application and Research Center (İTÜ TÖMER), which was established with the vision of transferring ITU's unique experience in natural language processing to language learning and starting artificial intelligence-supported language learning research in Turkey, will benefit from this technology during their education. Working as the center director, Assoc. Gülşen Eryiğit gives the good news that the number of such applications will increase in the near future.



12
May 2022
ITU NLP 2022 New Members

We welcomed our new members to our ITU NLP team – October, 2021.

IMG_8441



27
Jan 2022
STEM_Valorise Interview

stemvalorise



https://stemvalorise.eu/gulsen-eryigit/

Interview with Dr Gülşen Eryigit

Entrepreneur & Professor of Artificial Intelligence and Data Engineering at Istanbul Technical University 

Gülşen’s Background

Dr. Gülşen Eryiğit is currently working as a professor at the Artificial Intelligence and Data Engineering Department, İstanbul Technical University (ITU), from where she received her MSc and PhD. degrees. She is the founder of the ITU Turkish Natural Language Processing (NLP) Pipeline servicing many academic studies in the field. She has been serving as a technical consultant to several companies from different sectors. She also acted as the Turkey’s representative in CLARIN (EU 7th Framework Program, CLARIN – Common language resources and technology infrastructure), EU Cost Actions PARSEME and ENETCOLLECT. Gülşen’s experience of more than 20 years in the field of Turkish language processing technologies make her an expert in the field; she explains her motivation as emotional and predictive. As a young graduate, she did not understand why it was preferable to purchase an item from abroad if it were possible to produce it domestically with Turkish engineers. Consequently, she chose to work with her own language in order to produce intelligent solutions in the field domestically. 

 

Gülşen’s valorisation story

Although ITU’s Technology Transfer Office encouraged Gülşen to build her own company, she was reluctant to take this path because she was afraid of moving away from research. Additionally, she did not want to deal with all the inputs and processes which go into making a business successful. Gülşen prefers to collaborate with other researchers and industry to generate social or economic value from research. For instance, for a collaborative project that focused on children with hearing disability, Gülşen and her NLP team worked with two other research groups specializing in human-computer-interaction and linguistics. The product itself is a new technology that solves reading problems for disabled children. Her team was mainly responsible of producing the technology for the translation, enabling the translation of Turkish texts in children’s books to Turkish sign language. 

  

In order to endow the machine with the capacity to translate, the research team developed machine readable representations for the Turkish sign language. One of the other teams worked on creating an Avatar system to perform the signs so that the children could watch and understand. This project, therefore, represents an example of successful multidisciplinary teamwork. After successfully receiving a patent, this project is now at the stage of commercialization. 

  

Gülşen and her team have made communication between machines and humans natural and efficient. One of their projects is also known as the first licensing success from Turkey to the US through İTU TTO. Developing machines capable of natural language processing is a critical technology that is still evolving in the world today. Turkish, which belongs to a different language family than many other languages, is among the most difficult for machines to learn. When the limited capabilities of the previous versions in the field are considered, this project’s impact has a critical value. 

 

Challenges and advice for future entrepreneurs

In terms of challenges, Gülşen points to a lack of valuable data sets or problems to be solved on the purely academic side. For this reason, she believes that UBC based projects will enable win-win situations in terms of technology transfer and use case gathering. Another challenge Gülşen faced is the short-term orientation of small enterprises and their expectation of getting quick results from the research; this is the main reason that she primarily works with large companies. 

  

Gülşen lists key success factors, especially when building relationships with outside companies, as follows: a good motivation and the energy in the team, university industry collaboration in terms of supporting the graduate theses or engaging in joint research, and a compatible long-term vision of the managers in partner companies. Gülşen thinks that it is important for young people to have an interest in research and hold the belief that they can be successful and develop new technologies from any project. 

  

Gülşen recommends that young researchers in academia or in industry should work in an area that makes them motivated and curious. She sees curiosity and motivation as key and believes that this motivation should come from within. Gülşen closes with a passage from a book: ‘’be careful regarding what you are curious about. If you are prepared to keep on thinking about it for the coming five years, then keep being curious about it, otherwise just let it go.”



07
Sep 2021


13
Apr 2021
Best Article Award

Best Paper Award to Our Faculty Member and Her Student

Department of Artificial Intelligence and Data Engineering faculty member Assoc. Prof. Dr. Gülşen Eryiğit’s work titled “Information Extraction from Text Intensive and Visually Rich Banking Documents” conducted together with her master’s student Berke Oral received the 2020 Master Degree Paper Award given by Elsevier’s prestigious journal “Information Processing & Management”.

News: İTÜ Media and Communication Office

The award-winning publication is the outcome of the TÜBİTAK 1505 project titled “Information Extraction from Banking Documents using Visual Features” jointly funded by TÜBİTAK and Yapı Kredi Technology and carried out at Faculty of Computer and Informatics as part of Assoc. Prof. Dr. Gülşen Eryiğit’s studies since 2016 within the framework of the university-industry collaboration with Yapı Kredi Technology at ITU Teknokent.

ITU Natural Language Processing group member and ITU Computer Engineering 2018 graduate Berke Oral signed a prestigious publication while he was at the master’s level as a result of his studies with Assoc. Prof. Dr. Eryiğit.

We congratulate our faculty member and our graduate and wish them continued success.

Assoc Prof. Dr. Gülşen Eryiğit

Gülşen Eryiğit is an Associate Professor at Istanbul Technical University (ITU), Faculty of Computer and Informatics. Her research focus is on Turkish Natural Language Processing (NLP). She is the founding member and the head of the ITU NLP Research Group and the coordinator of the ITU Language Technologies and Social Robotic Laboratory. She is the builder of the ITU Turkish NLP Pipeline (tools.nlp.itu.edu.tr) servicing many academic studies in the field.  Gülşen Eryiğit received her M.Sc. and Ph.D. degrees from ITU Computer Engineering Department in 2002 and 2007, respectively. In 2006, she worked as a visiting student researcher at Växjö University, Sweden where she worked in the first ranked team of CoNLL shared tasks in 2006 and 2007 on multi-lingual dependency parsing. In 2007, she won Siemens Excellence award for her Ph.D. thesis. She acted as a reviewer or co-author of various publications in prestigious journals and conferences. She was Turkey’s representative in CLARIN (EU 7th Framework Programme, CLARIN - Common language resources and technology infrastructure). Recently, she conducted or worked as researcher in many EU (Cost), TÜBİTAK (1001 and 1003) and Ministry of Science, Industry and Technology (Santez) supported projects and served as consultant in 9 TÜBİTAK TEYDEB supported projects. She has also been a reviewer for many research and industrial proposals for European Commission (H2020 and Cost Action Programs), TÜBİTAK, and Turkey’s Ministry of Science, Industry and Technology. She is the owner of one international licensing and one patent.

Berke Oral

Berke Oral completed his bachelor's and master's degree at Istanbul Technical University Computer Engineering Department, in 2018 and 2021 respectively. He is an active member of ITU NLP Research Group and has been working for Yapı Kredi Technology since 2018. He is mainly responsible for developing and maintaining machine learning pipelines for the automatic processing of scanned documents. He is the co-author of several scientific papers and articles in prestigious conferences and journals.

https://www.journals.elsevier.com/information-processing-and-management/news/ipm-2020-master-degree-paper-award-winner


oralen



28
Mar 2021
From Turkish to Turkish Sign Language Patent

We congratulate our team member Assoc.Prof.Dr. Gülşen Eryiğit who is among the inventors of work on translation from written Turkish to Turkish Sign Language (TİD) which has been patented by the Turkish Patent and Trademark Office.

We also congratulate Cihat Eryiğit,  Prof.Dr. Meltem Kelepir and Assoc.Prof.Dr.Hatice Köse for their valuable contributions. This work was part of a Tubitak 1003 research project (grant no: 114E263) carried out at Istanbul Technical University Faculty of Computer and Informatics Engineering in collaboration with the linguistic department of Boğaziçi University.
patent



02
Mar 2021
KUIS AI Talks


Assoc. Prof. Dr. Gülşen Eryiğit from İTÜ Department of Artificial Intelligence and Data Engineering, and NLP researcher Berke Oral were invited to KUIS AI Talks.  

You can watch the event video of Assoc. Prof. Dr. Gülşen Eryiğit detailing her latest project titled "Extracting from text-heavy and visually rich scanned documents" on YouTube.
  



01
Jun 2020
PITSTOP Webinars

You are invited to the webinar where ITU Faculty of Computer and IT Faculty, Assoc. Prof. Dr. Gülşen Eryiğit is our invited speaker.

Developing machines with natural language processing capability is a very critical technology that is still developing in the world today. Especially Turkish, which belongs to a different language family compared to popular languages, is among the most challenging languages ​​for machines to learn.

ITU Faculty of Computer and Informatics Assoc. Dr. Gülşen Eryiğit explains Turkish language processing studies from past to present.

Register for the Event and CLICK to watch the seminar ... https://youtu.be/Ei3SeseAq0g



07
May 2020
Turkish Touch to Artificial Intelligence

Interview by ITU corporate communication department could be reached from here.
(The interview is in Turkish language)


DSC_8850
corporate communication department



05
Feb 2020
EnetCollect CrowdFest MultiWordExpress

ITU NLP Team acted as team leader and task participants in EnetCollect Crowdfest organized at Combra. The task was about "A game for Implicit Crowdsourcing for Multiword Expressions Learning Material". We congratulate all the team members.

IMG_4249
IMG_4248



30
Jan 2020
İTÜ NLP TEAM is selected as the Lab of the Month by enetCollect

Natural Language Processing (NLP) Group  is selected as the lab of the month (January, 2020) by the European Cost Action "European Network for Combining Language Learning with Crowdsourcing Techniques"(enetCollect) CA16105

We congratulate all our team members, and wish continued success. 
enetcollectfacebook



27
Dec 2019
İTÜ NLP & İstanbul Provincial Directorate for National Education

On December 26 Thursday, Assos. Prof. Gülşen Cebiroğlu Eryiğit (İTÜ Faculty Member) gave a seminar to İstanbul Provincial Directorate for National Educations teachers, executive members and students about Natural language processing which is one of the important topics of Artificial Intelligence. 
https://istanbulakademi.meb.gov.tr/haberler.php?pID=93

27122843_1



27
Jul 2018
İsmail Arı Machine Learning Summer School at Bosporus

One of our faculty members Assos. Prof. Gülşen Cebiroğlu Eryiğit was attended as a lecturer to İsmail Arı Machine Learning Summer School at Bosporus. You can reach the speech by clicking the below link.