Abstract Transforming several relational tables into a data set to be used as input to a Machine Learning (ML) model is a complex task since the data scientist has to derive many intermediate tables, queries and views in a disorganized manner. This process creates many SQL queries to facilitate the exploration task of the data scientist. Because the provenance of the... Transforming several relational tables into a data set to be used as input to a Machine Learning (ML) model is a complex task since the data scientist has to derive many intermediate tables, queries and views in a disorganized manner. This process creates many SQL queries to facilitate the exploration task of the data scientist. Because the provenance of the intermediate results is not reflected, similar SQL queries tend to be written multiple times causing repeated manual work. In this paper, we propose a tool “ER4ML” assisting data scientists in modeling and visualizing the transformations applied to the relational database before obtaining a dataset feeding the ML models. ER4ML is a diagram flow based on a conceptual view of the database schema and transformations based on powerful extensions of ER diagram in UML notations. In addition, ER4ML tracks data provenance, improving query and data set reuse. Read More Read Less

Abstract We report our models for detecting age, language variety, and gender from social media data in the context of the Arabic author profiling and deception detection shared task (APDA) [32].We build simple models based on pre-trained bidirectional encoders from transformers (BERT). We first fine-tune the pre-trained BERT model on each of the three datasets with shared task released data. Then we augment shared... We report our models for detecting age, language variety, and gender from social media data in the context of the Arabic author profiling and deception detection shared task (APDA) [32].We build simple models based on pre-trained bidirectional encoders from transformers (BERT). We first fine-tune the pre-trained BERT model on each of the three datasets with shared task released data. Then we augment shared task data with in-house data for gender and dialect, showing the utility of augmenting training data. Our best models on the shared task test data are acquired with a majority voting of various BERT models trained under different data conditions. We acquire 54.72% accuracy for age, 93.75% for dialect, 81.67% for gender, and 40.97% joint accuracy across the three tasks.1 Read More Read Less

Abstract This paper focuses on the author profiling task published in the FIRE 2019 (Forum for Information Retrieval Evaluation), which includes automatic identification of the age, gender, and language variety of Arabic tweets. We think the author profiling task as a multi-Classification problem. We have used word and character based on TFIDF features, learned the logistic regression classifier to predict the labels. In the... This paper focuses on the author profiling task published in the FIRE 2019 (Forum for Information Retrieval Evaluation), which includes automatic identification of the age, gender, and language variety of Arabic tweets. We think the author profiling task as a multi-Classification problem. We have used word and character based on TFIDF features, learned the logistic regression classifier to predict the labels. In the final results, our proposed method shows a good performance in terms of age prediction, the accuracy rate is 0.6250. Additionally, we have obtained 0.5111 and 0.9604 accuracy for gender and language variety classifications respectively. In the experiment, We have used the different feature combination and adjusted the feature parameters to test the system. The combination of word and character features can improve the prediction accuracy and enhance the system performance significantly. Read More Read Less

Abstract This paper describes the method for the Classification of Insincere Question(CIQ) in FIRE 2019. In this evaluation, we use an ensemble learning method to unite multiple classification models, including logistic regression model, support vector machine, Naive Bayes, decision tree, K-Nearest Neighbor, Random Forest. The result shows that our classification achieves the 67.32% accuracy rate(rank top 1) on the test dataset.... This paper describes the method for the Classification of Insincere Question(CIQ) in FIRE 2019. In this evaluation, we use an ensemble learning method to unite multiple classification models, including logistic regression model, support vector machine, Naive Bayes, decision tree, K-Nearest Neighbor, Random Forest. The result shows that our classification achieves the 67.32% accuracy rate(rank top 1) on the test dataset. Read More Read Less

Abstract Recent advancements in the Internet technologies have made a tremendous change in the social media. Hate Speech is an attack that is directed towards a group of people based on their religion, gender, colour etc. The offensive content in social media poses a threat to democracy. As these kind of hate speech and offensive content on the web increases day by day, manually... Recent advancements in the Internet technologies have made a tremendous change in the social media. Hate Speech is an attack that is directed towards a group of people based on their religion, gender, colour etc. The offensive content in social media poses a threat to democracy. As these kind of hate speech and offensive content on the web increases day by day, manually monitoring or controlling such hate crimes is a highly challenging task. Most of the existing methodologies focus on English language tweets and only limited work has been reported for Hindi and German language posts. Also, the importance of feature selection methods is not explored much for this problem. In this research work, an enhanced ensemble classifier approach is proposed to identify hate and offensive content posted in Hindi or German languages. In the proposed approach, CHI square based feature selection method is combined with a Random Forest Classifier to classify the tweets. This work was submitted to Hate and Offensive Content Identification (HASOC) task@FIRE2019. From the various experiments conducted on the released HASOC dataset, it is shown that an accuracy of 81% and 64% was achieved on German and Hindi language tweets. Read More Read Less