Mehmet Umut SALUR

Dr. Mehmet Umut SALUR

orcid

Research Focus & Metrics

Academic Career Length

12Years (2015-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
23 March 2020 in IEEE Access
Verified
Journal Article
Open access
MSMehmet Umut SALUR
Mehmet Umut SALUR
ilhan Aydinilhan Aydin
ilhan Aydin
A massive use of social media platforms such as Twitter and Facebook by omnifarious organizations has increased the critical individual feedback on the situation, events, products, and services. However, sentiment classification plays an important role in the user's feedback evaluation. At present, deep learning such as long short-term memory (LSTM), gated recurrent unit (GRU), bidirectionally long short-term memory (BiLSTM) or convolutional neural network (CNN) are prevalently preferred in sentiment classification. Moreover, word embedding such as Word2Vec and FastText is closely examined in text for mapping closely related to the vectors of real numbers. However, both deep learning and word embedding methods have strengths and weaknesses. Combining the strengths of the deep learning models with that of word embedding is the key to high-performance sentiment classification in the field of natural language processing (NLP). In the present study, we propose a novel hybrid deep learning model that strategically combines different word embedding (Word2Vec, FastText, character-level embedding) with different deep learning methods (LSTM, GRU, BiLSTM, CNN). The proposed model extracts features of different deep learning methods of word embedding, combines these features and classifies texts in terms of sentiment. To verify the performance of the proposed model, several deep learning models called basic models were created to perform series of experiments. By comparing, the performance of the proposed model with that of past studies, the proposed model offers better sentiment classification performance.

1 September 2018
Verified
Conference Paper
Closed access
Nashwan Adnan OTHMANNashwan Adnan OTHMAN
Nashwan Adnan OTHMAN
MSMehmet Umut SALUR
Mehmet Umut SALUR
Mehmet KARAKÖSEMehmet KARAKÖSE
Mehmet KARAKÖSE
IAIlhan Aydin
Ilhan Aydin
In these days, real-time object detection and dimensioning of objects is an important issue from many areas of industry. This is a vital topic of computer vision problems. This study presents an enhanced technique for detecting objects and computing their measurements in real time from video streams. We suggested an object measurement technique for real-time video by utilizing OpenCV libraries and includes the canny edge detection, dilation, and erosion algorithms. The suggested technique comprises of four stages: (1) identifying an object to be measured by using canny edge detection algorithm, (2) using morphological operators includes dilation and erosion algorithm to close gaps between edges, (3) find and sort contours, (4) measuring the dimensions of objects. In the implementation of the proposed technique, we designed a system that used OpenCV software library, Raspberry Pi 3 and Raspberry Camera. The proposed technique was nearly achieved 98% success in determines the size of the objects.

15 June 2022 in Neural Computing and Applications
Verified
Journal Article
Closed access
PDF file
MSMehmet Umut SALUR
Mehmet Umut SALUR
ilhan Aydinilhan Aydin
ilhan Aydin

1 September 2018
Verified
Conference Paper
Closed access
IAİlhan Aydin
İlhan Aydin
MSMehmet Sevi̇
Mehmet Sevi̇
MSMehmet Umut SALUR
Mehmet Umut SALUR
Today, social networks have been part of many people's lives. Many activities such as communication, promotion, advertisement, news, agenda creation have started to be done through social networks. Some malicious accounts on Twitter are used for purposes such as misinformation and agenda creation. This is one of the basic problems in social networks. Therefore, detection of malicious account is significant. In this study, machine learning-based methods were used to detect fake accounts that could mislead people. For this purpose, the dataset generated was pre-processed and fake accounts were determined by machine learning algorithms. Decision trees, logistic regression and support vector machines algorithms are used for the detection of fake accounts. Classification performances of these methods are compared and the logistic regression proved to be more successful than the others.

1 September 2019
Verified
Conference Paper
Closed access
MSMehmet Umut SALUR
Mehmet Umut SALUR
IAİlhan Aydin
İlhan Aydin
SAShahed Alali Alghrsi
Shahed Alali Alghrsi
Today, it is possible to see smart solutions based on artificial intelligence in almost every area such as smart homes, smart assistants, smart cities, smart campuses as well as smart tourism centers. Smart tourism centers are mainly tourism solutions that targets more visitors. Visitors, before traveling to any tourist center, conduct research on the social media application, and they are generally evaluating previous visitor comments. Considering that there are many social media platforms, evaluating these comments manually requires great effort and time. In this study, we have developed a SmartSenti application that sentiment analysis which was performed on Twitter posts about tourism centers using machine learning and transfer learning methods. We have collected tweets about five touristic place in Turkey. These tweets are classified as positive or negative after pre-processing and feature extracting phase. The classification results are shared visually with the users through the mobile application that we developed.

1 September 2018
Verified
Conference Paper
Closed access
MSMehmet Umut SALUR
Mehmet Umut SALUR
IAİlhan Aydin
İlhan Aydin
The increase of text based content produced in electronic environment necessitated to classify these contents with advanced algorithms. For text classification, the text contents are reduced to a more structured form by being preprocessed. Many studies have examined how these preprocesses affect the performance of classification algorithms. The main question of this study is how the Convolutional Neural Network (CNN), which is noteworthy with its classification success, is influenced by text preprocessing in the text classification. In this study, the effect of text preprocessing is examined CNN classification performance on two different data sets with Turkish content. As a result of the study, it has been determined that the text preprocessing has a weak positive effect on CNN classification performance.