small changes

_news/burak-award.md

@@ -4,7 +4,7 @@ date: 2024-05-23
 inline: true
 ---
 
-Congratulations to Burak for winning the Second Place Award at the Regeneron ISEF in the category of Technology Enhances the Arts! 🎉
+Congratulations to our high-school mentee Burak Akbudak for winning the Second Place Award at the Regeneron ISEF in the category of Technology Enhances the Arts! 🎉
 
 ***
 We are proud to announce that Burak has received the Second Place Award at the Regeneron ISEF in the category of "Technology Enhances the Arts" for his project, “AI-Composed Musically Rich Classical Guitar Pieces.” You can check the details of this project in [here](https://projectboard.world/isef/project/teca001-ai-composed-musically-rich-classical-guitar-pieces). Congratulations, Burak!

_pages/about.md

@@ -23,14 +23,11 @@ social: false # includes social icons at the bottom of the page
 ---
 
 
-GGLab (pronounced as "cici" in Turkish) is a Natural Language Processing (NLP) research lab led by [Asst. Prof. Gözde Gül Şahin](https://gozdesahin.github.io/) with a focus on procedural language understanding, in particular, representation and evaluation of procedural text. We have a keen interest in conducting fundamental research in core methodologies, including but not limited to areas such as learning under low-resource settings, incorporating linguistic structures in language models and developing interpretable AI systems. Additionally, we explore how these methodologies can be applied to various tasks, such as text simplification, semantic analysis, morphological analysis, grammar error correction and answering questions. We are part of [Computer Science Department](https://cs.ku.edu.tr/) at [Koç University](https://www.ku.edu.tr/) and affiliated with [KUIS AI Lab](https://ai.ku.edu.tr/), located in the north of Istanbul, Türkiye. GGLab is partly funded by [Scientific and Technological Research Council of Türkiye](https://www.tubitak.gov.tr/) via Tübitak 2232B International Fellowship for Outstanding Researchers programme. 
+GGLab (pronounced as "cici" in Turkish) is a Natural Language Processing (NLP) research lab led by [Asst. Prof. Gözde Gül Şahin](https://gozdesahin.github.io/). We have a keen interest in conducting fundamental research in core methodologies, including but not limited to areas such as learning/performing under low-resource settings and incorporating expert knowledge in language technologies. Additionally, we explore how these methodologies can be applied to various tasks, such as text simplification, linguistic structure analysis, grammatical error correction, dialogue understanding and question answering. We are part of [Computer Science Department](https://cs.ku.edu.tr/) at [Koç University](https://www.ku.edu.tr/) and affiliated with [KUIS AI Lab](https://ai.ku.edu.tr/), located in the north of Istanbul, Türkiye. GGLab is partly funded by [Scientific and Technological Research Council of Türkiye](https://www.tubitak.gov.tr/) via Tübitak 2232B International Fellowship for Outstanding Researchers program, and [Wikimedia Foundation](https://wikimediafoundation.org/) via [Wikimedia Research Fund](https://meta.wikimedia.org/wiki/Grants:Programs/Wikimedia_Research_%26_Technology_Fund/Wikimedia_Research_Fund). 
 
 [Talk to us](mailto:gosahin@ku.edu.tr) or
 [join our group]({{ '/open-positions' | relative_url }})
 when you are interested in these topics or our work.
-Students at Koç University,
-please find [our courses]({{ '/teaching' | relative_url }})(coming soon).
-{: class="clearfix"}
 
 {% assign members = site.members | where: "team_frontpage", true | sort: "lastname" %}
 <div class="d-flex flex-wrap align-content-stretch justify-content-center m-n2 pt-5 no-gutters">

_pages/open-positions.md

@@ -7,26 +7,17 @@ nav_rank: 5
 nav: true
 ---
 
-We are looking for talented students who are interested in doing research on natural language processing.
+We are always looking for talented, enthusiastic students who are interested in doing research on natural language processing.
 At the moment,
-**Undergraduate internships**, **Masters**, and **PhD positions** are available.
-
-The positions will focus on understanding and processing procedural language provided as natural language instructions (e.g., step-by-step instructions to fix WiFi). Some of the topics are (but are not limited to):
-
-- Investigating and designing neural/hybrid models with long-range reasoning on procedural text,
-- Document-grounded task-oriented dialogue on procedural text, and 
-- Designing evaluation metrics and benchmarks for procedural text
-
-The position will also be affiliated with KUIS AI, a thriving international research environment with more than 50 graduate students working on NLP-related topics spread over computer vision, robotics, machine learning, and human-computer interaction. Successful candidates will have the opportunity to collaborate closely with other researchers at the KUIS AI and at Koç University. The primary communication language is English. The preferred starting date is as soon as possible. 
-
+**Undergraduate internships**, **Masters**, and **PhD positions** are available. Please check our on-going projects page, and latest publications before applying. 
 
 **Qualifications**
 
 The ideal candidates should be enthusiastic about natural language processing, have excellent mathematical and programming skills, should be team players, and have Bachelor's degrees in computer science, computational linguistics, or a related discipline.
 
 **Offer**
 
-Salary: Current KUIS AI fellowship salaries are 11.200TL/mo for MSc and 18.600TL/mo for Ph.D. students. Scholarship raises after a qualified publication as the first author: 1000 TL/mo for MS, 1500 TL/mo for Ph.D., to be re-evaluated every semester. 
+Salary: Please check the [KUIS AI page](https://ai.ku.edu.tr/education/programs/graduate-programs/ai-fellowships/) for the latest information on fellowships.  
 
 Travel: Full support for top-tier conference publications.
 
@@ -38,7 +29,7 @@ Side benefits: Private health insurance, meal card, and student housing based on
 
 If you are interested in any of the positions, please send an email to [<i class="fas fa-envelope"></i> gosahin [at] ku.edu.tr](mailto:gosahin@ku.edu.tr) with the subject "Ph.D. (or MS) application". Please send a CV (including publications, if applicable), transcript (or records), a copy of a work that you're most proud of (e.g., Master's thesis, conference paper), and a motivation letter stating your research interests. 
 
-The official application still needs to be done via the application portal as instructed here. The official deadline is the 4th of June, however, the applications will be reviewed on a regular basis. Please indicate "NLP" and "Procedural text" as your research interests and mention Dr. Gözde Gül Şahin in your Statement of Purpose. 
+The official application still needs to be done via the application portal as instructed here. Please indicate "NLP" as your research interest and mention Dr. Gözde Gül Şahin in your Statement of Purpose, so that it won't be missed. 
 
 
 **Why Koç University?**

_projects/project_alfalfas.md

@@ -9,11 +9,11 @@ publications: 'projects^=*ALfaLFas'
 
 ---
 
-## Automatic Learning oF ProcedurAl Language From NAtural Language InStructions for Intelligent Assistance 
+### Automatic Learning oF ProcedurAl Language From NAtural Language InStructions for Intelligent Assistance 
 
 <br>
 
-### Motivation
+#### Motivation
 Despite the number of studies that report exceptionally high scores (Devlin et. al., 2019) for downstream natural language processing (NLP) tasks, a growing number of studies discuss the gap between their performance on such tasks and on real-world tasks that require “understanding” ([Bender & Koller, 2020](https://aclanthology.org/2020.acl-main.463.pdf)). Some of the major causes for this are (i) neural models not being able to generalize to out-of-domain data, (ii) downstream tasks not containing the challenges of real-world scenarios and iii) not having suitable evaluation measures.
 
 In order to bring the performance on real-life and downstream tasks closer, this project proposes a novel task for understanding natural language utterances within a more realistic and challenging scope: **understanding human-written instructions**. Giving step-by-step instructions is one of the primary way of human communication to teach someone a new topic or a task. The research plan envisions a future, where people would also be able to instruct machines with such step-by-step instructions, and this research project aims to take the first step towards that goal by developing necessary tools to parse natural language utterances into a sequence of procedures. The advantage of having such a representation would be having the ability to reduce the statements automatically by using an off-the-shelf interpreter and enrich the final model with domain-specific knowledge/rules which cannot be easily learned from data.
@@ -24,7 +24,7 @@ Another major challenge arises when we want build models to *parse instructions
 
 Final obstacle that stands on the way is the *right evaluation measures* to track the progress in the direction of generalizable natural language processing models. The task we define already sets up a realistic measure, however, measuring the progress with one single score has been shown to be problematic. The reasons are as follows. First of all, neural models are not interpretable. Hence their strengths and weaknesses can not be analyzed simply by looking at a single score. Second, a single score only shows how good a model performs on this specific test set, rather than how good a model is by means of the skills required by the task (e.g., logical deduction, mathematical reasoning). As mentioned, this project hypothesizes that improving the ability of long-range reasoning would yield more generalizable models. That brings us to the final research question: “Which reasoning/logical skills are required for processing instructions?” and “How can we measure these skills adequately?”
 
-### Goals
+#### Goals
 
 The project will investigate three major research directions to answer the aforementioned RQs:
 
@@ -37,14 +37,15 @@ encloses various linguistic and reasoning challenges. Nonetheless, the researche
 evaluate only on the end result using a single score. 
 - Neural/Hybrid Models with Long-Range Reasoning Abilities for Procedural Text: We will contribute with i) investigating the generalizability of existing techniques on the procedural text, and ii) developing novel models inspired from existing cognitive architectures. 
 
-### Team
+#### Team
 
- - Asst. Prof. Gözde Gül Şahin
- - Abdalfatah Rashid Safa, MSc, Müge Kural, MSc
+ - PI: Gözde Gül Şahin
+ - Graduate Students: Abdulfatah Rashid Safa, Tamta Kapanadze
+ - Interns: Arda Uzunoğlu
 
-### Funding
+#### Funding
 
 This project (ID:1109B322100424) is funded within the scope of Tübitak 2232B International Fellowship for Outstanding Researchers funding scheme. (Funding Period: 09.2022-09.2025)
 
-### Credit
+#### Credit
 <a href="https://www.freepik.com/free-photo/3d-render-robot-with-books_1166338.htm#query=robot%20reading%20instruction&position=0&from_view=search&track=ais">Image by kjpargeter</a> on Freepik

_projects/project_wikimedia.md

@@ -8,6 +8,10 @@ img_contains_title: true
 publications: 'projects^=*terminology'
 ---
 
+### Bridging the Gap Between Wikipedians and Scientists with Terminology-Aware Translation: A Case Study in Turkish 
+
+<br> This project addresses the gap between the escalating volume of English-to-Turkish Wikipedia translations and the insufficient number of contributors, particularly in technical domains. Leveraging expertise from academics’ collaborative terminology dictionary effort, we propose a pipeline system to enhance translation quality. Our focus is on bridging academic and Wikipedia communities, creating datasets, and developing NLP models for terminology identification and retrieval, and terminology-aware translation. The aim is to foster sustained contributions and improve the overall quality of Turkish Wikipedia articles.
+
 <div class="row">
  <div class="col-sm mt-3 mt-md-0">
   {% include figure.liquid loading="eager" path="assets/img/projects/wikimedia.webp" title="pipeline" class="img-fluid rounded z-depth-1" %}
@@ -17,11 +21,7 @@ publications: 'projects^=*terminology'
  The pipeline of our proposed project.
 </div>
 
-## Bridging the Gap Between Wikipedians and Scientists with Terminology-Aware Translation: A Case Study in Turkish 
-
-<br> This project addresses the gap between the escalating volume of English-to-Turkish Wikipedia translations and the insufficient number of contributors, particularly in technical domains. Leveraging expertise from academics’ collaborative terminology dictionary effort, we propose a pipeline system to enhance translation quality. Our focus is on bridging academic and Wikipedia communities, creating datasets, and developing NLP models for terminology identification and retrieval, and terminology-aware translation. The aim is to foster sustained contributions and improve the overall quality of Turkish Wikipedia articles.
-
-### Goals
+#### Goals
 
 The project will focus on the following tasks:
 
@@ -33,10 +33,14 @@ The project will focus on the following tasks:
 
 - **Terminology-Aware Translation:** We will build post-editing and translation systems that will be constrained with the terminology database. 
 
-### Team
+- **Build an Effective Communication Channel:** We will survey both communities (Wikipedians and scientists) to identify the best practices to build the bridges, and the ways these two communities can help each other in a sustainable way. We will publish reports, best practices and guidelines. 
+
+#### Team
 
- - Asst. Prof. Gözde Gül Şahin
- - Ali Gebeşçe (Masters student)
+ - PI: Gözde Gül Şahin
+ - Graduate student(s): Ali Gebeşçe
+ - Interns: Ege Uğur Amasya, Mina Durhasan
+ - Duration: 01.06.2024 - 31.05.2025
 
-### Funding
+#### Funding
 This project is funded by Wikimedia Research Fund. Official URL for the funded project is [here](https://meta.wikimedia.org/wiki/Grants:Programs/Wikimedia_Research_Fund/Bridging_the_Gap_Between_Wikipedians_and_Scientists_with_Terminology-Aware_Translation:_A_Case_Study_in_Turkish).