Yazar "Atagun, Ercan" seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • Küçük Resim Yok
    Öğe
    A Comparison of Data Mining Tools and Classification Algorithms: Content Producers on the Video Sharing Platform
    (Springer International Publishing Ag, 2020) Atagun, Ercan; Argun, Irem Duzdar
    With the development of internet technologies, the use of video sharing sites has increased. Video sharing sites allow users to watch videos of others. In addition, users can create an account to upload content and upload videos. These platforms stand out as the places where individuals are both producers and consumers. In this study, data about YouTube which is a video sharing site was used. The content of the content, which is also called as a channel on YouTube, was made by using a set of producers. The data set with 5000 samples on YouTube channels is taken from Kaggle. The data were classified using 4 different data mining tools such as Weka, RapidMiner, Knime and Orange using Naive Bayes and Random Forest algorithms. The parameters are requested from the user in order to obtain a more efficient result in the application of data mining algorithms and in the data preprocessing steps and in the data mining steps. Although these parameters are common in some data mining software, they are not included in all data mining software. Data mining software provides management of some parameters while other parameters cannot be managed. These changes affect the accuracy value in the study and affect the accuracy value in different ratios. Changing the values of the parameters revealed differences in the accuracy rates obtained. A data mining software model has been proposed by emphasizing to what extent the management of the parameters of the study and the extent of the management of the parameters should be connected to the data mining software developer.
  • Küçük Resim Yok
    Öğe
    Graph-based interpretable dialogue sentiment analysis: A HybridBERT-LSTM framework with semantic interaction explainer
    (Elsevier, 2026) Atagun, Ercan; Temur, Gunay; Birogul, Serdar
    Conversational sentiment analysis in natural language processing faces substantial challenges due to intricate contextual semantics and temporal dependencies within multi-turn dialogues. We present a novel HybridBERT-LSTM architecture that integrates BERT's contextualized embeddings with LSTM's sequential processing capabilities to enhance sentiment classification performance in dialogue scenarios. Our framework employs a dual-pooling mechanism to capture local semantic features and global discourse dependencies, addressing limitations of conventional approaches. Comprehensive evaluation on IMDb benchmark and real-world dialogue datasets demonstrates that HybridBERT-LSTM consistently improves over standalone models (LSTM, BERT, CNN, SVM) across accuracy, precision, recall, and F1-score metrics. The architecture effectively exploits pre-trained contextual representations through bidirectional LSTM layers for temporal discourse modeling. We introduce WordContextGraphExplainer, a graph-theoretic interpretability framework addressing conventional explanation method limitations. Unlike LIME's linear additivity assumptions treating features independently, our approach utilizes perturbation-based analysis to model non-linear semantic interactions. The framework generates semantic interaction graphs with nodes representing word contributions and edges encoding inter-word dependencies, visualizing contextual sentiment propagation patterns. Empirical analysis reveals LIME's inadequacies in capturing temporal discourse dependencies and collaborative semantic interactions crucial for dialogue sentiment understanding. WordContextGraphExplainer explicitly models semantic interdependencies, negation scope, and temporal flow across conversational turns, enabling comprehensive understanding of both word-level contributions and contextual interaction influences on decision-making processes. This integrated framework establishes a new paradigm for interpretable dialogue sentiment analysis, advancing trustworthy AI through high-performance classification coupled with comprehensive explainability.