Emine Yilmaz
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Emine Yilmaz has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Fanghua Ye
University College London London United Kingdom
2 shared publications - Jordan Massiah
Amazon & University College London, London, United Kingdom
2 shared publications Serhii Havrylov
2 shared publications- Aldo Lipani
University College London, London, United Kingdom
1 shared publication - Bağnu ÇOLAKOĞLU
Tekirdag Namik Kemal Universitesi
1 shared publication - Deniz Sari Gedi̇k
Tekirdag Namik Kemal University
1 shared publication
Latest Publications
- Open Access
- 27 May 2024
- Published by Tekirdag Ziraat Fakultesi Dergisi in Tekirdağ Ziraat Fakültesi Dergisi
- Vol. 21 (3) , 666-682
The Effect of Wild Mushroom Collection in The Mountains of Strandja on The Economic Development of The People Living in The Region
There is a need for innovative approaches that can accommodate the rural population and manage to keep the rural population in the settlement. At this point, it is important to carry out actions that will adequately meet the living conditions of rural areas and to increase the income levels of rural people. The protection of forests and the maintenance of biological diversity are important in terms of non-wood forest products collected from the forests by the villagers living in that region, and it is very important for the sustainability and economic future of life in the region. In this study, a questionnaire was collected from 130 households living in the region to determine the importance of mushrooms, which grew naturally in the mountains of Strandja and were used for consumption, in terms of family economy and development. The empirical analysis was based on a face-to-face survey of 130 respondents living in villages close to the Strandja Mountains and collecting mushrooms in 2019. The answers given to the questions, which determined the opinions of the participants about the mushrooms growing in nature, were taken with a 5-point Likert scale and factor analysis was performed. Villagers collecting mushrooms could not get the income they wanted from the products they obtained. It was determined that the income of mushroom collectors from mushrooms contributed to their daily needs, but did not make a significant contribution to increasing their welfare level. For this, villagers needed to establish cooperatives and create their own sales points. Since the mushroom collectors used the mushrooms in their diets, the villagers should be trained in mushroom drying and canning to ensure mushroom consumption throughout the year. In the research, it was observed that the local people were aware of the contribution of mushrooms to the economy and that a large part of their food consumption consisted of mushrooms grown in nature and therefore it was important in rural development.
- 27 May 2024
- Published by Tekirdag Ziraat Fakultesi Dergisi in Tekirdağ Ziraat Fakültesi Dergisi
- Vol. 21 (3) , 666-682
Open AccessThe Effect of Wild Mushroom Collection in The Mountains of Strandja on The Economic Development of The People Living in The Region
There is a need for innovative approaches that can accommodate the rural population and manage to keep the rural population in the settlement. At this point, it is important to carry out actions that will adequately meet the living conditions of rural areas and to increase the income levels of rural people. The protection of forests and the maintenance of biological diversity are important in terms of non-wood forest products collected from the forests by the villagers living in that region, and it is very important for the sustainability and economic future of life in the region. In this study, a questionnaire was collected from 130 households living in the region to determine the importance of mushrooms, which grew naturally in the mountains of Strandja and were used for consumption, in terms of family economy and development. The empirical analysis was based on a face-to-face survey of 130 respondents living in villages close to the Strandja Mountains and collecting mushrooms in 2019. The answers given to the questions, which determined the opinions of the participants about the mushrooms growing in nature, were taken with a 5-point Likert scale and factor analysis was performed. Villagers collecting mushrooms could not get the income they wanted from the products they obtained. It was determined that the income of mushroom collectors from mushrooms contributed to their daily needs, but did not make a significant contribution to increasing their welfare level. For this, villagers needed to establish cooperatives and create their own sales points. Since the mushroom collectors used the mushrooms in their diets, the villagers should be trained in mushroom drying and canning to ensure mushroom consumption throughout the year. In the research, it was observed that the local people were aware of the contribution of mushrooms to the economy and that a large part of their food consumption consisted of mushrooms grown in nature and therefore it was important in rural development.
- Open Access
- 13 March 2024
- Published by Tekirdag Ziraat Fakultesi Dergisi in Tekirdağ Ziraat Fakültesi Dergisi
- Vol. 21 (2) , 335-347
Food Demand and Nutritional Habits of Consumers During the Covid-19 Outbreak: A Case Study in Türkiye
The Covid-19 pandemic, which has affected the entire world, has caused changes in many areas of life worldwide. These changes have also influenced and continue to influence consumer behaviors. During the restrictions, people staying at home and only going out for essential needs such as grocery shopping has caused changes in their shopping and consumption habits. The changing consumer behaviors related to the process have brought about many studies. Since the outbreak of Covid-19, it has caused significant changes in the attitudes and behaviors of individuals in the world and in Türkiye. Numerous studies have been conducted to understand these changes. The purpose of this study is to examine the effects of the pandemic on the food demand and nutritional habits of individuals residing in Tekirdağ province. To achieve this, a survey was conducted with consumers living in Tekirdağ. According to the research's predicted 10% margin of error and 99% significance level, the sample size was calculated to be 166 people. Factor analysis was applied during the statistical analysis of the findings. Based on the results of the factor analysis, it was found that the desire of individuals to consume organic products and their awareness of healthy nutrition were high. The negative conditions that occurred during the pandemic increased the desire of individuals to eat. It has shown that consumers are more sensitive about healthy eating during the Covid-19 pandemic process. An increase in the consumption of food products that form the basis of a healthy diet, such as fruits, vegetables and eggs, and a corresponding decrease in the consumption of carbonated drinks and prepackaged food also support this finding. However, the high increase in the consumption of pastries and nuts can be given as examples of the negative effects of the need to stay at home and the restriction of physical activities due to the Covid-19 outbreak. In addition, their buying behavior changed, and the number of people who procure food products through online shopping increased. The aim of this study is to inform consumers and researchers working on this subject about how food consumption of consumers has changed after Covid-19.
- 13 March 2024
- Published by Tekirdag Ziraat Fakultesi Dergisi in Tekirdağ Ziraat Fakültesi Dergisi
- Vol. 21 (2) , 335-347
Open AccessFood Demand and Nutritional Habits of Consumers During the Covid-19 Outbreak: A Case Study in Türkiye
The Covid-19 pandemic, which has affected the entire world, has caused changes in many areas of life worldwide. These changes have also influenced and continue to influence consumer behaviors. During the restrictions, people staying at home and only going out for essential needs such as grocery shopping has caused changes in their shopping and consumption habits. The changing consumer behaviors related to the process have brought about many studies. Since the outbreak of Covid-19, it has caused significant changes in the attitudes and behaviors of individuals in the world and in Türkiye. Numerous studies have been conducted to understand these changes. The purpose of this study is to examine the effects of the pandemic on the food demand and nutritional habits of individuals residing in Tekirdağ province. To achieve this, a survey was conducted with consumers living in Tekirdağ. According to the research's predicted 10% margin of error and 99% significance level, the sample size was calculated to be 166 people. Factor analysis was applied during the statistical analysis of the findings. Based on the results of the factor analysis, it was found that the desire of individuals to consume organic products and their awareness of healthy nutrition were high. The negative conditions that occurred during the pandemic increased the desire of individuals to eat. It has shown that consumers are more sensitive about healthy eating during the Covid-19 pandemic process. An increase in the consumption of food products that form the basis of a healthy diet, such as fruits, vegetables and eggs, and a corresponding decrease in the consumption of carbonated drinks and prepackaged food also support this finding. However, the high increase in the consumption of pastries and nuts can be given as examples of the negative effects of the need to stay at home and the restriction of physical activities due to the Covid-19 outbreak. In addition, their buying behavior changed, and the number of people who procure food products through online shopping increased. The aim of this study is to inform consumers and researchers working on this subject about how food consumption of consumers has changed after Covid-19.
- Open Access
- 22 May 2023
- Published by Tekirdag Ziraat Fakultesi Dergisi in Tekirdağ Ziraat Fakültesi Dergisi
- Vol. 20 (2) , 418-429
Determination of the Factors Affecting the Level of Benefit from Young Farmer Project Support in Rural Development: Tekirdag Sample, Türkiye
Achieving development in the field of agriculture in a country passes through the development of rural areas. Various plans, programs and policies are being developed for the sustainability of production and solving the problems of those living in rural areas. In recent years, it has been given importance to support development and entrepreneurship in agriculture in Türkiye in order to solve the existing problems. For this purpose, the Young Farmer Project was put into effect with the decision of the Council of Ministers in 2016. With this project, it is aimed to ensure that young farmers stay in the agricultural sector by providing the support they need, and to prevent migration from rural to urban by encouraging their agricultural production and activities. In this research, the application process of the farmers who received support between 2016 and 2017 in Tekirdag province with the Young Farmer Project, how they carried out the project, what kind of difficulties they faced while carrying out, how they continued it, and where they could carry it was determined. A survey was conducted with 106 young farmers who received support within the scope of the research. 54% of the young farmers participating in the research are woman farmers. The fact that male farmers have existing businesses registered on them that they work in other paid jobs besides farming, positive discrimination against women within the scope of the project has contributed to the increase in the number of female young farmer applicants and winners. 82.1% of the young farmers involved in the project also reside in the rural area before the project, and 13.2% live both in the city center and in the rural areas. The rate of those who started living in rural areas after receiving project support is 4.7%. It has been determined that young farmers who own livestock enterprises do not have an increase in their income levels, and they have difficulty in meeting their borrowing and operating input costs. Young people who receive support within the scope of the Young Farmer Project express the opinion that the project will not encourage agriculture and animal husbandry. For women farmers, the most important contribution of the project is that they have owned businesses for the first time, and for male farmers, they have further expanded their existing businesses.
- 22 May 2023
- Published by Tekirdag Ziraat Fakultesi Dergisi in Tekirdağ Ziraat Fakültesi Dergisi
- Vol. 20 (2) , 418-429
Open AccessDetermination of the Factors Affecting the Level of Benefit from Young Farmer Project Support in Rural Development: Tekirdag Sample, Türkiye
Achieving development in the field of agriculture in a country passes through the development of rural areas. Various plans, programs and policies are being developed for the sustainability of production and solving the problems of those living in rural areas. In recent years, it has been given importance to support development and entrepreneurship in agriculture in Türkiye in order to solve the existing problems. For this purpose, the Young Farmer Project was put into effect with the decision of the Council of Ministers in 2016. With this project, it is aimed to ensure that young farmers stay in the agricultural sector by providing the support they need, and to prevent migration from rural to urban by encouraging their agricultural production and activities. In this research, the application process of the farmers who received support between 2016 and 2017 in Tekirdag province with the Young Farmer Project, how they carried out the project, what kind of difficulties they faced while carrying out, how they continued it, and where they could carry it was determined. A survey was conducted with 106 young farmers who received support within the scope of the research. 54% of the young farmers participating in the research are woman farmers. The fact that male farmers have existing businesses registered on them that they work in other paid jobs besides farming, positive discrimination against women within the scope of the project has contributed to the increase in the number of female young farmer applicants and winners. 82.1% of the young farmers involved in the project also reside in the rural area before the project, and 13.2% live both in the city center and in the rural areas. The rate of those who started living in rural areas after receiving project support is 4.7%. It has been determined that young farmers who own livestock enterprises do not have an increase in their income levels, and they have difficulty in meeting their borrowing and operating input costs. Young people who receive support within the scope of the Young Farmer Project express the opinion that the project will not encourage agriculture and animal husbandry. For women farmers, the most important contribution of the project is that they have owned businesses for the first time, and for male farmers, they have further expanded their existing businesses.
- Open Access
MetaASSIST: Robust Dialogue State Tracking with Meta Learning
Existing dialogue datasets contain lots of noise in their state annotations. Such noise can hurt model training and ultimately lead to poor generalization performance. A general framework named ASSIST has recently been proposed to train robust dialogue state tracking (DST) models. It introduces an auxiliary model to generate pseudo labels for the noisy training set. These pseudo labels are combined with vanilla labels by a common fixed weighting parameter to train the primary DST model. Notwithstanding the improvements of ASSIST on DST, tuning the weighting parameter is challenging. Moreover, a single parameter shared by all slots and all instances may be suboptimal. To overcome these limitations, we propose a meta learning-based framework MetaASSIST to adaptively learn the weighting parameter. Specifically, we propose three schemes with varying degrees of flexibility, ranging from slot-wise to both slot-wise and instance-wise, to convert the weighting parameter into learnable functions. These functions are trained in a meta-learning manner by taking the validation set as meta data. Experimental results demonstrate that all three schemes can achieve competitive performance. Most impressively, we achieve a state-of-the-art joint goal accuracy of 80.10% on MultiWOZ 2.4.
Open AccessMetaASSIST: Robust Dialogue State Tracking with Meta Learning
Existing dialogue datasets contain lots of noise in their state annotations. Such noise can hurt model training and ultimately lead to poor generalization performance. A general framework named ASSIST has recently been proposed to train robust dialogue state tracking (DST) models. It introduces an auxiliary model to generate pseudo labels for the noisy training set. These pseudo labels are combined with vanilla labels by a common fixed weighting parameter to train the primary DST model. Notwithstanding the improvements of ASSIST on DST, tuning the weighting parameter is challenging. Moreover, a single parameter shared by all slots and all instances may be suboptimal. To overcome these limitations, we propose a meta learning-based framework MetaASSIST to adaptively learn the weighting parameter. Specifically, we propose three schemes with varying degrees of flexibility, ranging from slot-wise to both slot-wise and instance-wise, to convert the weighting parameter into learnable functions. These functions are trained in a meta-learning manner by taking the validation set as meta data. Experimental results demonstrate that all three schemes can achieve competitive performance. Most impressively, we achieve a state-of-the-art joint goal accuracy of 80.10% on MultiWOZ 2.4.
- Open Access
Just Mix Once: Worst-group Generalization by Group Interpolation
Advances in deep learning theory have revealed how average generalization relies on superficial patterns in data. The consequences are brittle models with poor performance with shift in group distribution at test time. When group annotation is available, we can use robust optimization tools to tackle the problem. However, identification and annotation are time-consuming, especially on large datasets. A recent line of work leverages self-supervision and oversampling to improve generalization on minority groups without group annotation. We propose to unify and generalize these approaches using a class-conditional variant of mixup tailored for worst-group generalization. Our approach, Just Mix Once (JM1), interpolates samples during learning, augmenting the training distribution with a continuous mixture of groups. JM1 is domain agnostic and computationally efficient, can be used with any level of group annotation, and performs on par or better than the state-of-the-art on worst-group generalization. Additionally, we provide a simple explanation of why JM1 works.
Open AccessJust Mix Once: Worst-group Generalization by Group Interpolation
Advances in deep learning theory have revealed how average generalization relies on superficial patterns in data. The consequences are brittle models with poor performance with shift in group distribution at test time. When group annotation is available, we can use robust optimization tools to tackle the problem. However, identification and annotation are time-consuming, especially on large datasets. A recent line of work leverages self-supervision and oversampling to improve generalization on minority groups without group annotation. We propose to unify and generalize these approaches using a class-conditional variant of mixup tailored for worst-group generalization. Our approach, Just Mix Once (JM1), interpolates samples during learning, augmenting the training distribution with a continuous mixture of groups. JM1 is domain agnostic and computationally efficient, can be used with any level of group annotation, and performs on par or better than the state-of-the-art on worst-group generalization. Additionally, we provide a simple explanation of why JM1 works.
- 25 August 2022
- Published by Edward Elgar Publishing
- p. 324-326
Blue Flag
"Blue Flag" published on Jul 2022 by Edward Elgar Publishing.
- 25 August 2022
- Published by Edward Elgar Publishing
- p. 324-326
Blue Flag
"Blue Flag" published on Jul 2022 by Edward Elgar Publishing.
- 25 August 2022
- Published by Edward Elgar Publishing
- p. 260-262
Beach
"Beach" published on Jul 2022 by Edward Elgar Publishing.
- 25 August 2022
- Published by Edward Elgar Publishing
- p. 260-262
Beach
"Beach" published on Jul 2022 by Edward Elgar Publishing.
- Open Access
Trans-Encoder: Unsupervised sentence-pair modelling through self- and mutual-distillations
In NLP, a large volume of tasks involve pairwise comparison between two sequences (e.g. sentence similarity and paraphrase identification). Predominantly, two formulations are used for sentence-pair tasks: bi-encoders and cross-encoders. Bi-encoders produce fixed-dimensional sentence representations and are computationally efficient, however, they usually underperform cross-encoders. Cross-encoders can leverage their attention heads to exploit inter-sentence interactions for better performance but they require task fine-tuning and are computationally more expensive. In this paper, we present a completely unsupervised sentence representation model termed as Trans-Encoder that combines the two learning paradigms into an iterative joint framework to simultaneously learn enhanced bi- and cross-encoders. Specifically, on top of a pre-trained Language Model (PLM), we start with converting it to an unsupervised bi-encoder, and then alternate between the bi- and cross-encoder task formulations. In each alternation, one task formulation will produce pseudo-labels which are used as learning signals for the other task formulation. We then propose an extension to conduct such self-distillation approach on multiple PLMs in parallel and use the average of their pseudo-labels for mutual-distillation. Trans-Encoder creates, to the best of our knowledge, the first completely unsupervised cross-encoder and also a state-of-the-art unsupervised bi-encoder for sentence similarity. Both the bi-encoder and cross-encoder formulations of Trans-Encoder outperform recently proposed state-of-the-art unsupervised sentence encoders such as Mirror-BERT and SimCSE by up to 5% on the sentence similarity benchmarks.
Open AccessTrans-Encoder: Unsupervised sentence-pair modelling through self- and mutual-distillations
In NLP, a large volume of tasks involve pairwise comparison between two sequences (e.g. sentence similarity and paraphrase identification). Predominantly, two formulations are used for sentence-pair tasks: bi-encoders and cross-encoders. Bi-encoders produce fixed-dimensional sentence representations and are computationally efficient, however, they usually underperform cross-encoders. Cross-encoders can leverage their attention heads to exploit inter-sentence interactions for better performance but they require task fine-tuning and are computationally more expensive. In this paper, we present a completely unsupervised sentence representation model termed as Trans-Encoder that combines the two learning paradigms into an iterative joint framework to simultaneously learn enhanced bi- and cross-encoders. Specifically, on top of a pre-trained Language Model (PLM), we start with converting it to an unsupervised bi-encoder, and then alternate between the bi- and cross-encoder task formulations. In each alternation, one task formulation will produce pseudo-labels which are used as learning signals for the other task formulation. We then propose an extension to conduct such self-distillation approach on multiple PLMs in parallel and use the average of their pseudo-labels for mutual-distillation. Trans-Encoder creates, to the best of our knowledge, the first completely unsupervised cross-encoder and also a state-of-the-art unsupervised bi-encoder for sentence similarity. Both the bi-encoder and cross-encoder formulations of Trans-Encoder outperform recently proposed state-of-the-art unsupervised sentence encoders such as Mirror-BERT and SimCSE by up to 5% on the sentence similarity benchmarks.
- Open Access
ASSIST: Towards Label Noise-Robust Dialogue State Tracking
The MultiWOZ 2.0 dataset has greatly boosted the research on dialogue state tracking (DST). However, substantial noise has been discovered in its state annotations. Such noise brings about huge challenges for training DST models robustly. Although several refined versions, including MultiWOZ 2.1-2.4, have been published recently, there are still lots of noisy labels, especially in the training set. Besides, it is costly to rectify all the problematic annotations. In this paper, instead of improving the annotation quality further, we propose a general framework, named ASSIST (lAbel noiSe-robuSt dIalogue State Tracking), to train DST models robustly from noisy labels. ASSIST first generates pseudo labels for each sample in the training set by using an auxiliary model trained on a small clean dataset, then puts the generated pseudo labels and vanilla noisy labels together to train the primary model. We show the validity of ASSIST theoretically. Experimental results also demonstrate that ASSIST improves the joint goal accuracy of DST by up to $28.16\%$ on MultiWOZ 2.0 and $8.41\%$ on MultiWOZ 2.4, compared to using only the vanilla noisy labels.Open AccessASSIST: Towards Label Noise-Robust Dialogue State Tracking
The MultiWOZ 2.0 dataset has greatly boosted the research on dialogue state tracking (DST). However, substantial noise has been discovered in its state annotations. Such noise brings about huge challenges for training DST models robustly. Although several refined versions, including MultiWOZ 2.1-2.4, have been published recently, there are still lots of noisy labels, especially in the training set. Besides, it is costly to rectify all the problematic annotations. In this paper, instead of improving the annotation quality further, we propose a general framework, named ASSIST (lAbel noiSe-robuSt dIalogue State Tracking), to train DST models robustly from noisy labels. ASSIST first generates pseudo labels for each sample in the training set by using an auxiliary model trained on a small clean dataset, then puts the generated pseudo labels and vanilla noisy labels together to train the primary model. We show the validity of ASSIST theoretically. Experimental results also demonstrate that ASSIST improves the joint goal accuracy of DST by up to $28.16\%$ on MultiWOZ 2.0 and $8.41\%$ on MultiWOZ 2.4, compared to using only the vanilla noisy labels. - Open Access
Predicting Engagement in Video Lectures
The explosion of Open Educational Resources (OERs) in the recent years creates the demand for scalable, automatic approaches to process and evaluate OERs, with the end goal of identifying and recommending the most suitable educational materials for learners. We focus on building models to find the characteristics and features involved in context-agnostic engagement (i.e. population-based), a seldom researched topic compared to other contextualised and personalised approaches that focus more on individual learner engagement. Learner engagement, is arguably a more reliable measure than popularity/number of views, is more abundant than user ratings and has also been shown to be a crucial component in achieving learning outcomes. In this work, we explore the idea of building a predictive model for population-based engagement in education. We introduce a novel, large dataset of video lectures for predicting context-agnostic engagement and propose both cross-modal and modality-specific feature sets to achieve this task. We further test different strategies for quantifying learner engagement signals. We demonstrate the use of our approach in the case of data scarcity. Additionally, we perform a sensitivity analysis of the best performing model, which shows promising performance and can be easily integrated into an educational recommender system for OERs.
Open AccessPredicting Engagement in Video Lectures
The explosion of Open Educational Resources (OERs) in the recent years creates the demand for scalable, automatic approaches to process and evaluate OERs, with the end goal of identifying and recommending the most suitable educational materials for learners. We focus on building models to find the characteristics and features involved in context-agnostic engagement (i.e. population-based), a seldom researched topic compared to other contextualised and personalised approaches that focus more on individual learner engagement. Learner engagement, is arguably a more reliable measure than popularity/number of views, is more abundant than user ratings and has also been shown to be a crucial component in achieving learning outcomes. In this work, we explore the idea of building a predictive model for population-based engagement in education. We introduce a novel, large dataset of video lectures for predicting context-agnostic engagement and propose both cross-modal and modality-specific feature sets to achieve this task. We further test different strategies for quantifying learner engagement signals. We demonstrate the use of our approach in the case of data scarcity. Additionally, we perform a sensitivity analysis of the best performing model, which shows promising performance and can be easily integrated into an educational recommender system for OERs.