From Co-occurrence to Lexical Cohesion for Automatic Query Expansion
This paper explores query expansion (QE) in Information Retrieval Systems (IRS), focusing on Pseudo-Relevance Feedback (PRF) to enhance query formulation. While traditional co-occurrence-based QE methods have limitations, this study investigates lexical-based measures for improving query expansion. A Lexical Cohesion-Based Query Expansion (LCBQE) algorithm is proposed and evaluated on the TREC dataset. Experimental results suggest that lexical-based approaches perform as well as or better than co-occurrence methods, demonstrating their potential to enhance retrieval efficiency based on query characteristics.
Change Detection in Remote Sensing Image Data Comparing Algebraic and Machine Learning Methods
This paper explores machine learning techniques for automatic change detection in remote sensing, addressing the need for efficient environmental monitoring. It compares post-classification comparison using a decision tree algorithm with a separability matrix against image differencing based on algebraic techniques. Landsat satellite data is used for pixel-wise computation, and performance is validated through 10-fold cross-validation. Results indicate that the post-classification method achieves higher accuracy, demonstrating its effectiveness in improving change detection and classification performance.
Information Retrieval and Query Expansion for Biomedical Data
This chapter explores the role of information retrieval (IR) in managing the vast and complex data generated in biomedical research, including electronic health records, scientific publications, clinical trials, and experimental studies. It introduces IR principles, emphasizing their significance in extracting relevant biomedical information. Additionally, it examines query expansion techniques to enhance retrieval performance in biomedical information retrieval systems (IRS). By understanding these concepts, researchers can design more effective IR systems, unlocking the full potential of biomedical data and accelerating scientific progress.
Exploring Knowledge Graphs (KG): A Comprehensive Overview
This chapter explores the significance of Knowledge Graphs (KGs) in the biomedical domain, highlighting their role in organizing, integrating, and analyzing complex biomedical data. By structuring knowledge in a graphical format, KGs facilitate drug discovery, clinical decision support, research advancements, drug safety monitoring, and ontology development. The chapter discusses key applications, challenges, and future directions, emphasizing how Knowledge Graphs contribute to breakthrough discoveries and improved patient outcomes.
Ethical Issues in Biomedical Text Mining
This chapter examines the ethical challenges associated with deploying automated and smart healthcare systems, particularly in biomedical text mining. While text-based data is often objective, its sensitivity can be compromised when integrated across medical, technical, and social domains. The automation of natural language processing (NLP) techniques raises ethical concerns in data collection, modeling, and self-learning systems. Ensuring ethical compliance, particularly regarding informed consent and data privacy, is crucial. While anonymization techniques help protect sensitive information, robust ethical frameworks must be implemented to uphold patient rights and data security.
Building Knowledge Graphs in the Biomedical Domain: Methods and Case Studies
This chapter provides a comprehensive guide to building biomedical knowledge graphs (KGs), covering data acquisition, knowledge representation, and ontology development through a practical case study. It explores applications of biomedical KGs, addressing challenges and future directions in the field. By the end, readers will gain insights into how knowledge graphs enhance biomedical research and contribute to improved patient outcomes.
Improving effectiveness of query expansion using information theoretic approach
This paper explores automatic query expansion to improve information retrieval performance using information-theoretic measures. A pseudo-relevance feedback approach is employed, selecting expansion terms from the top N retrieved documents based on co-occurrence analysis.
Machine learning approach for automatic document summarization
This paper investigates a machine learning-based approach for automatic document summarization, focusing on feature selection and learning patterns to determine key information for summaries. Traditional methods rely on various sentence selection features, each with strengths and limitations. Instead of an ad hoc combination, a trainable machine learning model is proposed to adapt across different text genres. The paper presents an architecture for document summarization, using machine learning techniques for efficient and adaptable summary generation.
Genetic algorithm based model for effective document retrieval
This paper proposes a genetic algorithm-based model to optimize similarity measures in information retrieval. By combining cosine, Jaccard, Dice, and Okapi measures with weighted contributions, the approach improves document ranking and retrieval effectiveness. Two fitness functions, non-order-based (recall-precision) and order-based (ranking-aware) are evaluated on the TREC dataset, demonstrating that order-based optimization enhances retrieval performance compared to traditional measures
A Framework for Automatic Query Expansion
This paper presents a framework and computational model for automatic query expansion using pseudo-relevance feedback. The proposed model aims to enhance retrieval efficiency by addressing key aspects of query expansion in a structured manner. Experiments conducted on the TREC dataset demonstrate improved retrieval performance, highlighting the effectiveness of the approach in refining search results.
Toward recommending learning tasks in a learner-centered approach
This paper presents a rule-based recommender system designed to support learner-centered education by assisting students in selecting learning tasks of varying difficulty levels. While offering choice enhances personalization, students may struggle to identify the most suitable tasks. The proposed system provides personalized recommendations, helping learners choose tasks that maximize their learning outcomes and engagement.
Student effort vs. outcome: analysis through Moodle logs
This paper examines whether positive student effort in face-to-face courses supported by Moodle LMS influences course completion rates. By analyzing LMS log data alongside teachers’ diaries, an algorithm was developed to assess student engagement. The results confirm that students demonstrating high effort levels are more likely to successfully complete their courses, highlighting the impact of consistent engagement in LMS-supported learning environments.
A rule-based recommender system to suggest learning tasks
This paper presents a rule-based recommender system designed to support learner-centered learning, where students choose their own topics and tasks. Given the challenge of selecting appropriate learning activities, the system identifies learners with similar characteristics and recommends tasks that have been effective for others. By providing personalized task suggestions, the approach aims to enhance learning outcomes and help learners make informed decisions that maximize their educational progress.
MOOC as supplementary tutoring to public school students learning
This paper explores the integration of Massive Open Online Courses (MOOCs) with a student-centered learning approach using a Learning Management System (LMS). While MOOCs offer free and accessible education, their effectiveness can be improved by allowing students to personalize and take control of their learning. Through the “My Moodle School” project, courses were designed to combine MOOC flexibility with student-driven learning strategies. A preliminary study demonstrated student acceptance and engagement, suggesting that this approach could enhance the learning recovery process and improve overall educational outcomes.
Co-occurrence based predictors for estimating query difficulty
This paper explores query difficulty prediction to assess how reliably an information retrieval (IR) system can handle user queries. Word mismatches between user queries and document vocabulary often hinder retrieval effectiveness, and while query expansion (QE) helps, its impact varies across different query sets. To address this inconsistency, the study investigates query difficulty predictors, introducing two new predictors based on the co-occurrence of query terms. Experiments on TREC collections evaluate their effectiveness, contributing to enhancing query processing reliability and ensuring more robust query expansion strategies.
Selecting Effective Expansion Terms for Better Information Retrieval
This paper explores automatic query expansion to improve information retrieval performance by extracting co-occurring terms from pseudo-relevant documents. The study evaluates different methods for selecting and ranking candidate terms, proposing the use of information-theoretic measures for more effective ranking.
A framework to provide personalization in learning management systems through a recommender system approach
This paper explores personalization in Learning Management Systems (LMS) by integrating recommender systems to tailor learning experiences based on learner profiles. Traditional LMSs present content statically, ignoring factors like learning styles, goals, prior knowledge, abilities, and interests. The proposed framework introduces a flexible integration model that suggests learning objects based on both individual learner data and successful learning patterns from similar profiles. This approach enhances learning efficiency, performance, and satisfaction, reducing learning time while maintaining academic success.
An Empirical Investigation of the Different Levels of Gamification in an Introductory Programming Course
This paper examines how different levels of gamification impact motivation, engagement, and performance in online learning. A study with 450 undergraduates experimentally manipulated gamification intensity based on game elements and content presentation, measuring outcomes at three points. While early-course assessments showed no significant differences, results indicated higher motivation and performance in gamified environments, particularly towards the course’s end. However, engagement was not significantly affected by gamification level. Findings emphasize the importance of implementing the right level of gamification to maximize student motivation and learning outcomes in online education.
Evaluation of awarding badges on students’ engagement in Gamified e-learning systems
This paper examines the role of gamification in enhancing engagement, motivation, and collaboration in educational settings. While game elements are widely used, their structural and contextual implementation remains inconsistent. Through an online gamified study, the research identifies key factors influencing gamification success and failure. Findings indicate that while badges and time spent increase engagement, incorporating feedback, social interaction, and interactive guidance are essential for improved learning outcomes. The study highlights the need for timely feedback, personalized support, and collaboration to maximize gamification’s effectiveness in learner-centered environments.
THESAURUS AND QUERY EXPANSION
This paper explores how query expansion improves information retrieval by addressing challenges like word mismatches and short queries. It examines how thesaurus-based approaches enhance search accuracy by identifying synonyms and semantically related terms, making retrieval more effective in search engines and natural language processing applications.
Multi-Class Pixel Certainty Active Learning Model for Land Cover Classification Using Hyperspectral Imagery
This paper introduces the Pixel Certainty Active Learning (PCAL) model for land cover classification using hyperspectral imagery (HSI). PCAL enhances classification accuracy by leveraging textural pattern information from Extended Differential Patterns (EDP). The approach integrates Distributed Intensity Filtering (DIF) for noise reduction and Histogram Equalization (HE) for image enhancement. Evaluated on the Pavia University and Indian Pines datasets, PCAL outperforms existing methods in classification accuracy and Kappa coefficient, demonstrating its effectiveness in handling pixel variations in remote sensing applications.
Modular Neural Network Learning Using Fuzzy Temporal Database
We proposed a novel mechanism for neural networks’ initial learning and training. Our approach used a fuzzy temporal database as a dynamic repository of information, enabling the network to learn and refine itself more effectively.
A GA Based Approach for Effective Information Retrieval
Our work explored using Genetic Algorithms (GAs) to enhance Information Retrieval (IR) by optimizing document ranking. We applied GAs to determine an optimal set of weights for a combined similarity measure, improving the retrieval of relevant documents while reducing non-relevant ones.
A Framework for Efficient Document Ranking Using Order and Non-Order Based Fitness Function
This paper explores the use of Genetic Algorithms (GAs) to enhance information retrieval by optimizing similarity measures. A combined similarity measure was defined, with GAs learning optimal weights for its components. Both order-based and non-order-based fitness functions were evaluated, showing that incorporating retrieval order improves performance. Experiments on TREC data demonstrated the effectiveness of this approach compared to traditional similarity measures.
WEBLORS–a personalized web-based recommender system
WEBLORS is an adaptive recommender system designed to enhance learning experiences by delivering personalized web-based learning materials. It addresses the challenge of information overload, helping learners access relevant content tailored to their needs, thereby improving learning outcomes and performance.
PLORS: a personalized learning object recommender system
This paper explores how traditional Learning Management Systems (LMS) support course administration but often follow a one-size-fits-all approach, disregarding individual learner profiles such as learning styles, goals, prior knowledge, abilities, and interests. To enhance personalization, recommender systems can suggest relevant learning objects based on a learner’s activity and similar profiles. The proposed personalized learning object recommender system provides tailored content recommendations, helping learners discover valuable resources they may have otherwise overlooked. This personalization enhances engagement, performance, and overall learning experience.
Improving effectiveness of query expansion using information theoretic approach
Automatic query expansion is a widely used method to enhance the performance of information retrieval systems. This paper proposes information-theoretic measures to improve the efficiency of co-occurrence-based automatic query expansion using a pseudo-relevance feedback approach. Expansion terms are selected from the top N retrieved documents and ranked using two information-theoretic methods: Kullback-Leibler divergence (KLD) and a variant of KLD. Experiments conducted on the TREC-1 dataset demonstrate that incorporating these measures can further enhance query expansion effectiveness. Additionally, extensive testing was performed to optimize key parameters, including the number of top N documents used for expansion.
Learning object recommendation system evaluation
This paper evaluates the effectiveness of a learning object recommendation system in enhancing the discovery of educational resources. Learning objects, defined as digital content designed for teaching, are stored in repositories equipped with indexing and retrieval mechanisms. The study applies a recommender algorithm to a repository at the Federal Institute of Rio Grande do Sul and analyzes its impact on user experience. The findings demonstrate how recommender systems can assist users in efficiently locating relevant learning objects, improving accessibility and usability within educational repositories.