| An Attention-based Deep Model for Automatic Short Answer Score |
| Pages: 127-132 (6) | [Full Text] PDF (545 KB) |
| T Gong, X Yao |
| School of Computer & Communication Engineering, University of Science and Technology Beijing, Beijing, China |
|
Abstract - In traditional education scenario, scoring assignment is done by human teachers and is time-consuming and laborious. In online education scenario, it is impracticable by human scoring, so automatic scoring is applied broadly. There are two categories of scoring questions: one is short text answer with reference answer, and the other is essay scoring without reference answer. In this work we focus on the former. Existed automatic scoring methods for short text answer used handy craft features which suffered from low accuracy. In order to overcome the problem, we propose a deep learning based method with attention mechanism to automatic scoring for assignment or exam on online education scenario. The method combines pre-trained embedding word vector and RNN model with attention to learn answer vector, and then learned response answer vector and reference answer vector are fed into logistic regression model to predict response answer’s score. Experimental results show that our proposed model achieves a relative 10% increase in performance compared with baseline model. |
| Index Terms - Automatic Scoring, Bidirectional RNN, Attention Mechanism, Word Embedding |
|
C itation - T Gong, X Yao. "An Attention-based Deep Model for Automatic Short Answer Score." International Journal of Computer Science and Software Engineering 8, no. 6 (2019): 127-132. |
| Risk Assessment and Management Method for Distributed Software Development Projects with “Fuzzy Approach” |
| Pages: 133-139 (7) | [Full Text] PDF (457 KB) |
| KU Birant, AH Isik, M Batar, HB Akarsu, AB Tektas |
| Department of Computer Engineering, Dokuz Eylul University, Izmir, TurkeyDepartment of Computer Engineering, Burdur Mehmet Akif Ersoy University, Burdur, TurkeyThe Graduate School of Natural and Applied Sciences, Dokuz Eylül University, Izmir, Turkey |
|
Abstract - Software is a product or a service or a project whose requirements are captured; specification document is prepared by requirements analysis; in/out architecture is designed; related source codes, variables, methods, classes, modules and relationships among these are written; necessary tests and integration operations are done; when needed, maintenance, repair and update operations are executed by software or computer engineers and also that has its own documentation jobs and user manual, and has numeric and textual data, and also multimedia tools in its memory. Software development projects have a large financial burden and need to invest in high volumes. When looked at costs based on the international tangible data on computer software; it was $150 billion in 1985, it was $2 trillion in 2010 and it passed over $5 trillion after 2016. Also, in the year of 2018, just a daily giro of Apple Store was about $250 million. Despite of the costs, expenses and investments that are exponentially increasing every year, the rate of successful development of the software projects is not very high. Based on the “CHAOS” report (international size) prepared in 2016, only 17% of the software development projects were completed in a timely manner, in the allocated budget and in accordance with the requirements. 53% of the projects were completed over time and/or over budget and/or also without fulfilling the requirements exactly. 30% of the software projects cannot have been completed in the development phase and were cancelled. For that software development projects with such high expenses and low success rate can have a better quality structure, a risk assessment and management approach has to be determined for better software risk assessment and management methodology. So, some problems which may form software risks can be recognized and determined on time before causing trouble and endangering for software development projects. In this paper, several software risk assessment approaches underlying software risk management were introduced and explained in detail besides the key – “Fuzzy Approach”. Moreover, that this “Fuzzy Technique” is more useful and more effective for software risk evaluation in comparison with the others was showed and expressed by giving meaningful and general linguistic rules and by applying a demo of software risk assessment under management with “Fuzzy Approach” (designed and developed by 18 original linguistic and logical rules with 15 different risk parameters in Python and in MATLAB) in this article. Also, this paper tried to explain the terms and the statements from the general to the specific as well as its content based on a hierarchical structure. |
| Index Terms - Software Engineering, Software, Software Project Management, Software Risk Management, Software Risk Assessment, Fuzzy Approach |
|
C itation - KU Birant, AH Isik, M Batar, HB Akarsu, AB Tektas. "Risk Assessment and Management Method for Distributed Software Development Projects with “Fuzzy Approach”." International Journal of Computer Science and Software Engineering 8, no. 6 (2019): 133-139. |
| Framework for Handling Data Veracity in Big Data |
| Pages: 140-143 (4) | [Full Text] PDF (296 KB) |
| M Al-Jepoori, ZA Al-Khanjari |
| Computing, Canterbury Christ Church University, Canterbury, UK, Computer Science, Sultan Qaboos University,Muscat, Oman |
|
Abstract - Big Data is the term used for massive amount of data collected by different means and in various formats. Data Veracity refers to the uncertainty of available data; this means that the quality of the collected data cannot be trusted. This paper reports on ongoing research based on using the Semantic Web technology to verify user entered data and increase dependability on Big Data. Validating, cleaning and reducing collected data are the major activities required to enhance the quality of the collected data. |
| Index Terms - Big Data, Veracity, User generated data, automatically collected data, Crowed validation of data, Fake news |
|
C itation - M Al-Jepoori, ZA Al-Khanjari. "Framework for Handling Data Veracity in Big Data." International Journal of Computer Science and Software Engineering 8, no. 6 (2019): 140-143. |