Pub Date : 2021-11-22DOI: 10.1109/ict4da53266.2021.9671266
Amare Kassaw, Aysheshim Demilie, Y. Wondie
Nowadays, video streaming has become one of the most dominant services due to increasing interest in watching online television programs and video on demand. Providing this service requires a high speed and high capacity network infrastructure. In this work, we propose quality of experience (QoE) model using fuzzy logic inference system for video streaming services. The proposed model is used to measure the user perception from quality of service (QoS) parameters. The model is essential to replace conventional subjective measurement techniques that are costly and inefficient. In addition, the proposed model is helpful for business decision making, network planning, optimization and operational support activities. The result analysis shows that the stall frequency and the start delay play a major impact on user perception by 33 % and 25 %, respectively. Besides, validation of the results shows that the proposed model is accurate, consistent and linear compared to currently existing models.
{"title":"Fuzzy Logic Inference System for Quality of Experience Modeling for LTE Video Streaming: Case of Addis Ababa LTE Network","authors":"Amare Kassaw, Aysheshim Demilie, Y. Wondie","doi":"10.1109/ict4da53266.2021.9671266","DOIUrl":"https://doi.org/10.1109/ict4da53266.2021.9671266","url":null,"abstract":"Nowadays, video streaming has become one of the most dominant services due to increasing interest in watching online television programs and video on demand. Providing this service requires a high speed and high capacity network infrastructure. In this work, we propose quality of experience (QoE) model using fuzzy logic inference system for video streaming services. The proposed model is used to measure the user perception from quality of service (QoS) parameters. The model is essential to replace conventional subjective measurement techniques that are costly and inefficient. In addition, the proposed model is helpful for business decision making, network planning, optimization and operational support activities. The result analysis shows that the stall frequency and the start delay play a major impact on user perception by 33 % and 25 %, respectively. Besides, validation of the results shows that the proposed model is accurate, consistent and linear compared to currently existing models.","PeriodicalId":371663,"journal":{"name":"2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130888812","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-22DOI: 10.1109/ict4da53266.2021.9672219
Shanko Chura Aredo, Y. Negash, Y. Wondie, Rajaveerappa Devadas, Feyisa Debo, Mahlet Akalu
Massive MIMO (mMIMO) is a variant of MIMO systems in which hundreds and thousands of miniatured antennas can be installed in a single Base Station (BS) serving a number of single antenna user terminals. As the number of transmit antennas (M) equivalently increases with the number of RF chains associated with each antenna elements especially in digital beamforming, the chain exhibits substantial amount of power consumption accordingly. Hence, to alleviate such problems, one of the potential solutions is to reduce the number of RFs or to minimize their power consumption. In this paper, low resolution DAC with hybrid precoder is used for reducing the total power consumption in order to achieve energy efficient downlink mMIMO. The simulation results show that, the power consumption overhead reduces with low DAC resolution and the system achieves more energy and spectral efficiency relative to without DAC resolution. Moreover, considering a given number of users, we compared energy and SE with and without resolution. Finally, the trade-off between EE and SE is analysed for both conditions and has been shown that EE with more resolution surpasses that of with low and without resolution.
{"title":"Adaptive Circuit Power Control for Down-link Massive MIMO Systems","authors":"Shanko Chura Aredo, Y. Negash, Y. Wondie, Rajaveerappa Devadas, Feyisa Debo, Mahlet Akalu","doi":"10.1109/ict4da53266.2021.9672219","DOIUrl":"https://doi.org/10.1109/ict4da53266.2021.9672219","url":null,"abstract":"Massive MIMO (mMIMO) is a variant of MIMO systems in which hundreds and thousands of miniatured antennas can be installed in a single Base Station (BS) serving a number of single antenna user terminals. As the number of transmit antennas (M) equivalently increases with the number of RF chains associated with each antenna elements especially in digital beamforming, the chain exhibits substantial amount of power consumption accordingly. Hence, to alleviate such problems, one of the potential solutions is to reduce the number of RFs or to minimize their power consumption. In this paper, low resolution DAC with hybrid precoder is used for reducing the total power consumption in order to achieve energy efficient downlink mMIMO. The simulation results show that, the power consumption overhead reduces with low DAC resolution and the system achieves more energy and spectral efficiency relative to without DAC resolution. Moreover, considering a given number of users, we compared energy and SE with and without resolution. Finally, the trade-off between EE and SE is analysed for both conditions and has been shown that EE with more resolution surpasses that of with low and without resolution.","PeriodicalId":371663,"journal":{"name":"2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134081174","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-22DOI: 10.1109/ict4da53266.2021.9672247
Amsalu Tessema, E. Alemneh
Identification and generation of Quality Attribute Scenarios (QASs) from Quality Attribute Requirements (QARs) is a critical software engineering technique for defining system specifications and is helpful in facilitating development of Software Architecture (SA) that meets the expected quality. However, identifying QAS types and extracting their components traditionally is a complex task that consumes time and energy. It also requires high budget and is an error-prone task, especially for inexperienced users. This study aims to develop an automatic QASs identification and generation model that extracts QASs from QARs. We used Natural Language Processing (NLP) to preprocess texts and Machine Learning (ML) approaches to identify QAS types, and we built a Custom Named Entity Recognition (CNER) model to generate QAS components. To evaluate the proposed identification model, we used five algorithms. Both SVM and Scholastic Gradient Descent (SGD) classifier algorithms scored 97.7 % accuracy, while LR, KNN, and NB scored 96%, 91.6 %, and 88.8%, respectively. The CNER model achieved 92.3% recall, 93.3% precision, and 92.8% F1-measure score. The results show that automatic identification of QASs from QARs has a potential to replace time taking and error-prone manual work.
{"title":"Automatic Quality Attribute Scenarios Identification and Generation from Quality Attribute Requirements","authors":"Amsalu Tessema, E. Alemneh","doi":"10.1109/ict4da53266.2021.9672247","DOIUrl":"https://doi.org/10.1109/ict4da53266.2021.9672247","url":null,"abstract":"Identification and generation of Quality Attribute Scenarios (QASs) from Quality Attribute Requirements (QARs) is a critical software engineering technique for defining system specifications and is helpful in facilitating development of Software Architecture (SA) that meets the expected quality. However, identifying QAS types and extracting their components traditionally is a complex task that consumes time and energy. It also requires high budget and is an error-prone task, especially for inexperienced users. This study aims to develop an automatic QASs identification and generation model that extracts QASs from QARs. We used Natural Language Processing (NLP) to preprocess texts and Machine Learning (ML) approaches to identify QAS types, and we built a Custom Named Entity Recognition (CNER) model to generate QAS components. To evaluate the proposed identification model, we used five algorithms. Both SVM and Scholastic Gradient Descent (SGD) classifier algorithms scored 97.7 % accuracy, while LR, KNN, and NB scored 96%, 91.6 %, and 88.8%, respectively. The CNER model achieved 92.3% recall, 93.3% precision, and 92.8% F1-measure score. The results show that automatic identification of QASs from QARs has a potential to replace time taking and error-prone manual work.","PeriodicalId":371663,"journal":{"name":"2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132487771","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-22DOI: 10.1109/ict4da53266.2021.9672222
Benson Kituku, Wanjiku Nganga, Lawrence Muchemi
Rule-based grammar development is labor-intensive in terms of time and knowledge requirements, especially for complex morphology and under-resourced languages. Notwithstanding, these grammars are needed for deep natural language processing, generation of well-formed output, or both. To address the challenge, this paper seeks to develop shared multilingual wide-coverage grammar for a subset of Kenyan Bantu languages in Grammatical Framework (GF) by leveraging on cross linguistic similarities using the grammar engineering strategies: grammar porting and grammar sharing. The shared grammar was developed using the morphology-driven approach, where the lexicons are defined first, followed by inflection regular expression and finally the syntax production rules. The resulting congruent Bantu parameterized grammar had shareability for category linearizations, parameters, paradigms, and syntax rules of 100%, 68.75%, 65.3% and 89.57%, respectively, while portability (modification) was exhibited in paradigms, parameter plus syntax rules at 14.29%, 18.75% and 10.43% respectively. The research concludes leveraging on the cross-linguistic similarities of principles and parameters significantly reduces multilingual grammar's development effort and contributes by developing the Bantu parametrized grammar which demonstrates how the effort of developing the rule base has been significantly reduced in languages where data is a scarce commodity.
{"title":"Leveraging on Cross Linguistic Similarities to Reduce Grammar Development Effort for the Under-Resourced Languages: a Case of Kenyan Bantu Languages","authors":"Benson Kituku, Wanjiku Nganga, Lawrence Muchemi","doi":"10.1109/ict4da53266.2021.9672222","DOIUrl":"https://doi.org/10.1109/ict4da53266.2021.9672222","url":null,"abstract":"Rule-based grammar development is labor-intensive in terms of time and knowledge requirements, especially for complex morphology and under-resourced languages. Notwithstanding, these grammars are needed for deep natural language processing, generation of well-formed output, or both. To address the challenge, this paper seeks to develop shared multilingual wide-coverage grammar for a subset of Kenyan Bantu languages in Grammatical Framework (GF) by leveraging on cross linguistic similarities using the grammar engineering strategies: grammar porting and grammar sharing. The shared grammar was developed using the morphology-driven approach, where the lexicons are defined first, followed by inflection regular expression and finally the syntax production rules. The resulting congruent Bantu parameterized grammar had shareability for category linearizations, parameters, paradigms, and syntax rules of 100%, 68.75%, 65.3% and 89.57%, respectively, while portability (modification) was exhibited in paradigms, parameter plus syntax rules at 14.29%, 18.75% and 10.43% respectively. The research concludes leveraging on the cross-linguistic similarities of principles and parameters significantly reduces multilingual grammar's development effort and contributes by developing the Bantu parametrized grammar which demonstrates how the effort of developing the rule base has been significantly reduced in languages where data is a scarce commodity.","PeriodicalId":371663,"journal":{"name":"2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125304127","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-22DOI: 10.1109/ict4da53266.2021.9672240
F. Mekuria, E. Nigussie, E. Schmitt, A. Gonzalez, Tesfa Tegegne, G. Fettweis
In this paper, we present a conceptual system architecture for real-time monitoring, predicting and controlling of invasive water hyacinth in freshwater bodies through the use of emerging technologies. The proposed system is planned to be deployed as one of the rescue efforts to preserve the fresh water lakes of Africa. The case study and the system presented in this paper are based on the Lake Tana, situated near the city of Bahir Dar, in Ethiopia. The rescuing efforts of Lake Tana so far focused on removal of the weed by hand and using harvesting machines. With the weed invasion doubling every two weeks, the current approaches will not be able to control the rapid invasion of the weed, which is causing considerable socioeconomic losses. The proposed system architecture employs networked underwater robots, aerial drones and other environmental sensors for better mapping of the weed coverage in real-time, predicting the floating paths of the weed, and learning the favourable environmental conditions of the lake for eradicating the invasive weed. The advantages of the proposed technical intervention lie not only in accurate monitoring and fast removal of the weed, but also in facilitating data collection for better understanding of the underlying environmental and chemical conditions that facilitate the rapid infestation and growth of the invasive weed.
{"title":"Rescuing the Fresh Water Lakes of Africa through the Use of Drones and Underwater Robots","authors":"F. Mekuria, E. Nigussie, E. Schmitt, A. Gonzalez, Tesfa Tegegne, G. Fettweis","doi":"10.1109/ict4da53266.2021.9672240","DOIUrl":"https://doi.org/10.1109/ict4da53266.2021.9672240","url":null,"abstract":"In this paper, we present a conceptual system architecture for real-time monitoring, predicting and controlling of invasive water hyacinth in freshwater bodies through the use of emerging technologies. The proposed system is planned to be deployed as one of the rescue efforts to preserve the fresh water lakes of Africa. The case study and the system presented in this paper are based on the Lake Tana, situated near the city of Bahir Dar, in Ethiopia. The rescuing efforts of Lake Tana so far focused on removal of the weed by hand and using harvesting machines. With the weed invasion doubling every two weeks, the current approaches will not be able to control the rapid invasion of the weed, which is causing considerable socioeconomic losses. The proposed system architecture employs networked underwater robots, aerial drones and other environmental sensors for better mapping of the weed coverage in real-time, predicting the floating paths of the weed, and learning the favourable environmental conditions of the lake for eradicating the invasive weed. The advantages of the proposed technical intervention lie not only in accurate monitoring and fast removal of the weed, but also in facilitating data collection for better understanding of the underlying environmental and chemical conditions that facilitate the rapid infestation and growth of the invasive weed.","PeriodicalId":371663,"journal":{"name":"2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123876512","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-22DOI: 10.1109/ict4da53266.2021.9671268
Ephrem Tadesse Degu, Rosa Tsegaye Aga
The COVID-19 outbreak is still a challenge in most places because of lack of up-to-date information, primarily, to the people in the world who speak and use underrepresented local languages. Ethiopia is one example of a country where several in-digenous languages are under-represented and under-resourced. Thus, building an interactive interface that responds to users' query using their local language with organized information plays a significant role. In this study, attention-augmented Encoder-Decoder Long Short Term Memory(LSTM) network model has proposed to provide adequate information about the pandemic to the people of Ethiopia by their local language, Amharic. The model converts Amharic COVID-19 related questions into the corresponding structured query language (SQL). The model retrieves information from the Amharic COVID-19 database that has developed for this study. The database contains frequently referenced COVID-19 attributes such as symptoms, prevention, transmission and frequently asked questions. In addition, a parallel Amharic Question-SQL query dataset has been prepared to evaluate the model. The LSTM Network with augmented attention mechanism has shown a clear significant result. In this study, a user interactive interface has also developed. The interface uses the proposed model and provides information about the pandemic to the people with questions in Amharic.
{"title":"Natural Language Interface for Covid-19 Amharic Database Using LSTM Encoder Decoder Architecture with Attention","authors":"Ephrem Tadesse Degu, Rosa Tsegaye Aga","doi":"10.1109/ict4da53266.2021.9671268","DOIUrl":"https://doi.org/10.1109/ict4da53266.2021.9671268","url":null,"abstract":"The COVID-19 outbreak is still a challenge in most places because of lack of up-to-date information, primarily, to the people in the world who speak and use underrepresented local languages. Ethiopia is one example of a country where several in-digenous languages are under-represented and under-resourced. Thus, building an interactive interface that responds to users' query using their local language with organized information plays a significant role. In this study, attention-augmented Encoder-Decoder Long Short Term Memory(LSTM) network model has proposed to provide adequate information about the pandemic to the people of Ethiopia by their local language, Amharic. The model converts Amharic COVID-19 related questions into the corresponding structured query language (SQL). The model retrieves information from the Amharic COVID-19 database that has developed for this study. The database contains frequently referenced COVID-19 attributes such as symptoms, prevention, transmission and frequently asked questions. In addition, a parallel Amharic Question-SQL query dataset has been prepared to evaluate the model. The LSTM Network with augmented attention mechanism has shown a clear significant result. In this study, a user interactive interface has also developed. The interface uses the proposed model and provides information about the pandemic to the people with questions in Amharic.","PeriodicalId":371663,"journal":{"name":"2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122077875","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Plant disease management is an essential step in the process of detecting pathogens in plants. For diseases like Potato's Late Blight, ineffective management could destroy the whole farm within a day. As a result, the total yield per unit of the potato becomes diminished. In this paper, potato's late blight disease detection model was built using the CNN algorithm. The dataset is collected in two ways. The first is preparing a dataset by capturing an image of the leaf from the Holeta potato farm, and the other is using the benchmark dataset. The dataset has two classes: the first class has a healthy class category and the other Late Blight. One of the problems with machine learning is not having enough data. In our case, to train a model publicly available database images of 596 and 430 of our own images were used. To address the problem of a small dataset we have used data augmentation techniques and transfer learning along with 5-fold cross-validation. InceptionV3, VGG16, and VGG19 pretrained models were used for transfer learning techniques. InceptionV3 model achieved 87% score among other pretrained models while testing with unseen data. In the future, the performance of the model could be improved by having a sufficient amount of dataset. Convolutional Neural Network Deep learning Plant disease detection Pretrained model Potato's Late Blight Convolutional Neural Network Deep learning Plant disease detection Pretrained model Potato's Late Blight
{"title":"Transfer Learning and Data Augmentation Based CNN Model for Potato Late Blight Disease Detection","authors":"Natnael Tilahun Sinshaw, Beakal Gizachew Assefa, Sudhir Kumar Mohapatra","doi":"10.1109/ict4da53266.2021.9672243","DOIUrl":"https://doi.org/10.1109/ict4da53266.2021.9672243","url":null,"abstract":"Plant disease management is an essential step in the process of detecting pathogens in plants. For diseases like Potato's Late Blight, ineffective management could destroy the whole farm within a day. As a result, the total yield per unit of the potato becomes diminished. In this paper, potato's late blight disease detection model was built using the CNN algorithm. The dataset is collected in two ways. The first is preparing a dataset by capturing an image of the leaf from the Holeta potato farm, and the other is using the benchmark dataset. The dataset has two classes: the first class has a healthy class category and the other Late Blight. One of the problems with machine learning is not having enough data. In our case, to train a model publicly available database images of 596 and 430 of our own images were used. To address the problem of a small dataset we have used data augmentation techniques and transfer learning along with 5-fold cross-validation. InceptionV3, VGG16, and VGG19 pretrained models were used for transfer learning techniques. InceptionV3 model achieved 87% score among other pretrained models while testing with unseen data. In the future, the performance of the model could be improved by having a sufficient amount of dataset. Convolutional Neural Network Deep learning Plant disease detection Pretrained model Potato's Late Blight Convolutional Neural Network Deep learning Plant disease detection Pretrained model Potato's Late Blight","PeriodicalId":371663,"journal":{"name":"2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127747431","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-22DOI: 10.1109/ict4da53266.2021.9671269
Shanko Chura Aredo, Y. Negash, Y. Wondie, Gelmecha Demissie Jobir, Rajaveerappa Devadas, Bayisa Taye
The primarily goal of future wireless communication systems is to device connections with optimal throughput with minimized latency and enhanced reliability at minimum cost. One of the most important technologies for attaining this goal is massive MIMO (mMIMO). Installing a number of antennas on a single Base Station together with channel coding which is achieved by redundant bits provides reliable communication compared to the classical MIMO in which only limited number of antennas are employed. This paper analyzes the performance of coded and uncoded channel with the effect of M. Furthermore, polar and Low-Density Parity Check Code (LDPC) coding schemes are evaluated and compared with the uncoded channel condition at different number of BS antennas. We have also assessed the performance of massive MIMO at perfect and imperfect channel conditions with different constellation orders. The simulation results show that channel coded data in a perfect channel outperforms the uncoded one and as the number of antennas grows, the bit error rate (BER) is reduced.
{"title":"Probability of Error Minimization Techniques in Downlink Massive MIMO Systems","authors":"Shanko Chura Aredo, Y. Negash, Y. Wondie, Gelmecha Demissie Jobir, Rajaveerappa Devadas, Bayisa Taye","doi":"10.1109/ict4da53266.2021.9671269","DOIUrl":"https://doi.org/10.1109/ict4da53266.2021.9671269","url":null,"abstract":"The primarily goal of future wireless communication systems is to device connections with optimal throughput with minimized latency and enhanced reliability at minimum cost. One of the most important technologies for attaining this goal is massive MIMO (mMIMO). Installing a number of antennas on a single Base Station together with channel coding which is achieved by redundant bits provides reliable communication compared to the classical MIMO in which only limited number of antennas are employed. This paper analyzes the performance of coded and uncoded channel with the effect of M. Furthermore, polar and Low-Density Parity Check Code (LDPC) coding schemes are evaluated and compared with the uncoded channel condition at different number of BS antennas. We have also assessed the performance of massive MIMO at perfect and imperfect channel conditions with different constellation orders. The simulation results show that channel coded data in a perfect channel outperforms the uncoded one and as the number of antennas grows, the bit error rate (BER) is reduced.","PeriodicalId":371663,"journal":{"name":"2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126283266","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Road Traffic Accidents (RTA) are a serious issue of societies resulting in huge losses at the economic and social levels and responsible for millions of deaths and injuries every year in the world. For instance, in Ethiopia, the number of deaths due to traffic accidents is increasing from one year to another. Addis Ababa is one of the popular and known cities that encounter a high number of RTAs due to the increasing number of vehicles and population. The main objective of this paper is to apply machine learning algorithms to predict the accident severity and identify the major causes of accidents in crowded cities (application of Addis Ababa city). The required data are collected from Addis Ababa city police departments and 12316 records of the accident are used for data analysis. We applied seven machine learning classification algorithms (Logistic Regression, Naive Bayes, Decision Tree, Support Vector Machine, K Nearest Neighbor, Random Forest, and AdaBoost) for predicting accident severity and compared the performance to choose the best model. We applied random undersampling and SMOTE oversampling techniques to handle the class imbalance nature of the dependent features and Principal Component Analysis (PCA) for dimension reduction. The experimental result shows that Random Forest achieved a 93.76% F1 score with SMOTE over-sampled data set and about 18% feature size reduction. Moreover, light condition, driving experience, age band of the driver, type of road lane, and types of junctions are identified as major determinant factors of the accident. According to this study, these are major factors to RTA and need to be considered in the design of infrastructure, regulations and policies to reduce accidents.
{"title":"Preventing Traffic Accidents Through Machine Learning Predictive Models","authors":"Tarikwa Tesfa Bedane, Beakal Gizachew Assefa, Sudhir Kumar Mohapatra","doi":"10.1109/ict4da53266.2021.9672249","DOIUrl":"https://doi.org/10.1109/ict4da53266.2021.9672249","url":null,"abstract":"Road Traffic Accidents (RTA) are a serious issue of societies resulting in huge losses at the economic and social levels and responsible for millions of deaths and injuries every year in the world. For instance, in Ethiopia, the number of deaths due to traffic accidents is increasing from one year to another. Addis Ababa is one of the popular and known cities that encounter a high number of RTAs due to the increasing number of vehicles and population. The main objective of this paper is to apply machine learning algorithms to predict the accident severity and identify the major causes of accidents in crowded cities (application of Addis Ababa city). The required data are collected from Addis Ababa city police departments and 12316 records of the accident are used for data analysis. We applied seven machine learning classification algorithms (Logistic Regression, Naive Bayes, Decision Tree, Support Vector Machine, K Nearest Neighbor, Random Forest, and AdaBoost) for predicting accident severity and compared the performance to choose the best model. We applied random undersampling and SMOTE oversampling techniques to handle the class imbalance nature of the dependent features and Principal Component Analysis (PCA) for dimension reduction. The experimental result shows that Random Forest achieved a 93.76% F1 score with SMOTE over-sampled data set and about 18% feature size reduction. Moreover, light condition, driving experience, age band of the driver, type of road lane, and types of junctions are identified as major determinant factors of the accident. According to this study, these are major factors to RTA and need to be considered in the design of infrastructure, regulations and policies to reduce accidents.","PeriodicalId":371663,"journal":{"name":"2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126033857","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-22DOI: 10.1109/ict4da53266.2021.9672229
Tadesse Destaw Belay, A. Ayele, G. Gelaye, Seid Muhie Yimam, Chris Biemann
Amharic is the second-most spoken Semitic language after Arabic and serves as the official working language of the government of Ethiopia. In Amharic writing, there are different characters with the same sound, which are called homophones. The current trend in Amharic NLP research is to normalize homophones into a single representation. This means, instead of character 11We have used the IPA notation for Amharic character transliteration, , and , the character will be used; instead of , and , the character will be replaced; and so on. This was done by the assumption that they are repetitive alphabets as they have the same sound. However, the impact of homophone normalization for Amharic NLP applications is not well studied. When one homophone character is substituted by another, there will be a meaning change and it is against the Amharic writing regulation. For example, the word is “poverty” while means “salvage”. These two words are homophones, but they have different meanings. To study the impacts of homophone normalization, we develop different general-purpose pre-trained embedding models for Amharic using regular and normalized homophone characters. We fine-tune the pre-trained models and build some Amharic NLP applications. For PoS tagging, a model that employs a regular FLAIR embedding model performs better, achieving an F1-score of 77%. For sentiment analysis, the model from regular RoBERTa embedding outperforms the other models with an F1-score of 60%. For IR systems, we achieve an F1-score of 90% using the normalized document. The results show that normalization is highly dependent on the NLP applications. For sentiment analysis and PoS tagging, normalization has negative impacts while it is essential for IR. Our research indicates that normalization should be applied with caution and more effort towards standardization should be given.
{"title":"Impacts of Homophone Normalization on Semantic Models for Amharic","authors":"Tadesse Destaw Belay, A. Ayele, G. Gelaye, Seid Muhie Yimam, Chris Biemann","doi":"10.1109/ict4da53266.2021.9672229","DOIUrl":"https://doi.org/10.1109/ict4da53266.2021.9672229","url":null,"abstract":"Amharic is the second-most spoken Semitic language after Arabic and serves as the official working language of the government of Ethiopia. In Amharic writing, there are different characters with the same sound, which are called homophones. The current trend in Amharic NLP research is to normalize homophones into a single representation. This means, instead of character 11We have used the IPA notation for Amharic character transliteration, , and , the character will be used; instead of , and , the character will be replaced; and so on. This was done by the assumption that they are repetitive alphabets as they have the same sound. However, the impact of homophone normalization for Amharic NLP applications is not well studied. When one homophone character is substituted by another, there will be a meaning change and it is against the Amharic writing regulation. For example, the word is “poverty” while means “salvage”. These two words are homophones, but they have different meanings. To study the impacts of homophone normalization, we develop different general-purpose pre-trained embedding models for Amharic using regular and normalized homophone characters. We fine-tune the pre-trained models and build some Amharic NLP applications. For PoS tagging, a model that employs a regular FLAIR embedding model performs better, achieving an F1-score of 77%. For sentiment analysis, the model from regular RoBERTa embedding outperforms the other models with an F1-score of 60%. For IR systems, we achieve an F1-score of 90% using the normalized document. The results show that normalization is highly dependent on the NLP applications. For sentiment analysis and PoS tagging, normalization has negative impacts while it is essential for IR. Our research indicates that normalization should be applied with caution and more effort towards standardization should be given.","PeriodicalId":371663,"journal":{"name":"2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2021-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134011931","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：