In this paper, we introduce an end-to-end Amharic text-line image recognition approach based on recurrent neural networks. Amharic is an indigenous Ethiopic script which follows a unique syllabic writing system adopted from an ancient Geez script. This script uses 34 consonant characters with the seven vowel variants of each (called basic characters) and other labialized characters derived by adding diacritical marks and/or removing parts of the basic characters. These associated diacritics on basic characters are relatively smaller in size, visually similar, and challenging to distinguish from the derived characters. Motivated by the recent success of end-to-end learning in pattern recognition, we propose a model which integrates a feature extractor, sequence learner, and transcriber in a unified module and then trained in an end-to-end fashion. The experimental results, on a printed and synthetic benchmark Amharic Optical Character Recognition (OCR) database called ADOCR, demonstrated that the proposed model outperforms state-of-the-art methods by 6.98% and 1.05%, respectively.

Agriculture is the backbone of the Ethiopian economy. It plays a significant role in the growth of the national economy. Among these, mung bean production takes the highest level of income for most smallholder farmers in all regions of Ethiopia who cultivate mung bean crops. Therefore, this study aims to predict mung bean production using a machine learning algorithm. For this study, the datasets were collected from the Central Statistical Agency of Ethiopia database. A total of 10273 instances were used for the experiment. Python machine learning tool was used to conduct the experiment and build an optimal model. To achieve the objective of this study, different experiments were conducted using Random Forest, Gradient Boosting, and Xgboosting algorithms. In addition, the predictive performances of the classifiers are evaluated and compared using accuracy, precision, recall, F1-score, and confusion matrix. Experimental result shows that the Xgboosting classifiers algorithm achieves the best performance with 98.65 % test accuracy and 99.8 % train accuracy. As a result, the Xgboosting classifier was selected for implementing the model to predict mung bean production. The findings of this study show that the main determinant factors for mung bean production include Meher season, use of extension program, fertilizer used, and fertilizer type. Therefore, the outcome of this research is essential to support the decision-making process of experts and it will enable them take corrective measures on the factors affecting mung bean production.

The main objective of this paper is to identify the factors that influence academic staff's knowledge-sharing intentions and behaviors in Ethiopian higher education. A structural equation model was used to validate the research framework using survey data from 256 respondents. The collected data has been analyzed using Smart PLS software. The result indicates that the factors that affect knowledge-sharing intention include social interaction, reciprocity, management support, reward systems, and reputation. Systems integrated into teaching-learning practice, availability of information technology infrastructure, perceived ease of use, and perceived usefulness are among the technological factors affecting the intention to use web technology for knowledge sharing toward knowledge-sharing behavior. It has also been concluded that management support had the greatest influence, but the reward system did not affect the knowledge-sharing intention. On the other hand, the availability of IT infrastructure has the largest effect on the intention to use web technology, while perceived ease of use has the least. The contributions of this study are twofold. The framework may serve as a roadmap for future researchers and managers considering their strategy to enhance knowledge sharing in HEI. The findings will benefit academic staff, university administrations, and higher education institutions. This study gives managers a solid foundation to promote and support academic staff's usage of web technology for knowledge sharing. The study will also help academic staff enhance their current knowledge-sharing practices.

The Ethiopian Orthodox Tewahedo Church, along with the Eritrean Orthodox Tewahedo Church, uses the ancient language Ge'ez for its liturgy. This preconial language, rich in history, plays a central role in religious texts. With the rise of technological advancements, particularly in optical character recognition (OCR), the need to digitize Ge'ez manuscripts and make them accessible to the public has become increasingly important. In an OCR system, the text is digitized, processed, and segmented, and features are extracted, categorized, and recognized. The first step in character recognition involves digitizing documents via tools such as the Cam Scanner at a resolution of 300 dpi. Once scanned, the image documents are preprocessed to remove noise via a set of low-pass filters. In our study, four low-pass algorithms for noise reduction and four thresholding algorithms for image binarization were tested. On the basis of having the lowest mean square error and highest peak signal-to-noise ratio, we selected bilateral filtering for noise reduction and Otsu’s thresholding for binarization. After preprocessing, the next step is image segmentation, where lines and individual characters are detected via a bounding box or bounding rectangle algorithm. Since the segmented characters vary in shape due to handwriting differences, they are normalized to a uniform size of 32 × 32 pixels. Character features are then extracted via automatic feature extraction methods within a deep learning framework. The recognition model is developed via pretrained deep convolutional neural networks, such as ResNet50 and VGG19, along with CNN-based support vector machines, trained on collected data. The experimental results show that a sequentially developed deep convolutional neural network achieves the highest accuracy of 93.52% in character recognition. However, segmentation errors, often caused by the degradation of document images, negatively impact the performance of the recognition models. Future research should focus on implementing advanced image enhancement and restoration techniques to further improve accuracy.

Tuberculosis remains a persistent public health problem in Ethiopia, which demands innovative solutions for diagnosis and treatment support. Knowledge-based systems, particularly case-based systems, offer valuable decision-making assistance for various diseases. This paper introduces a data mining-driven approach for automatic knowledge extraction to develop a case-based system for tuberculosis diagnosis. Hidden insights are extracted from a tuberculosis dataset through the hybrid knowledge discovery process. The collected data set is pre-processed to fill in missing values, correct outliers, and remove noise. Clustering techniques, such as K-means, Make Density-based clustering, and farthest first, are employed to construct a descriptive model. The K-means algorithm is selected for its superior performance. Integration of the descriptive model into a case-based system is done using a Java-based integrator module. The prototype case-based system demonstrates an accuracy of 90%, which is a promising result. Further work is required to integrate the case-based system with rule-based towards developing a generic knowledge-based system.