International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 11 Issue: 05 | May 2024
www.irjet.net
p-ISSN: 2395-0072
Automating Assessment with NLP-Powered Answer Checker Akshada Tari1 1Student, Department of Information Technology and Engineering, Goa College of Engineering, Farmagudi, Goa,
India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In the ever-evolving landscape of education, the
how NLP contributes to the efficiency and effectiveness of the grading process.
demand for efficient and accurate assessment tools has grown exponentially. This paper explores the integration of Natural Language Processing (NLP) techniques to develop an Automatic Answer Checker (AAC) with a focus on enhancing the grading process. NLP, a subfield of artificial intelligence, offers a powerful set of tools for understanding and analyzing human language, making it an ideal candidate for automating the assessment of the responses. The proposed AAC system employs advanced NLP ML algorithms to evaluate and grade answers submitted by students. The system aims to streamline the grading process. As education transitions into a more technologically-driven era, the integration of innovative tools like the AAC system holds the potential to enhance the educational experience, ensuring fair and timely evaluation while promoting a deeper understanding of the subject matter.
The rest of the paper is organized as follows: The method proposed which includes sentence transformer, All-mpnetbase-v2, Cosine Similarity is given in Section 2. The Literature survey is covered in Section 3. Conclusions are discussed in Section 4.
2. Related Work [1]. Jagadamba G and Chaya Shree G proposed Artificial Intelligence-based answer verifier to do the job of examiner/evaluator. Artificial Intelligence-based Answer Verifier calculates the score of the student by combining various parameters such as keywords, proper grammar. The value of keywords ranges from 1 to 6 where 1 is for Excellent and 6 is for Very Poor. The values of "grammar" attribute ranges from 0 which is for Improper and 1 which is for Proper. The system is more efficient for answers that are point to point. Provides overall accuracy of 80%. The module is designed and tested for the ‘Cosine Similarity’ algorithm. Cosine Similarity is used to measure the similarity between two non-zero vectors which are the inner product space. The measure is the cosine of the angle between the two vectors i.e., 0° is 1, and less than 1 for any angle in the interval (0, π) radians
Key Words: Natural Language Processing, Cosine Similarity, Sentence Transformer, Automatic Answer Checker
1.INTRODUCTION As the landscape of education evolves, the adoption of online tests and examinations has gained widespread popularity, aiming to alleviate the burdens associated with traditional examination evaluation processes. While online assessments typically focus on objective or multiple-choice questions, the evaluation of subjective-based questions and answers has proven challenging due to the complexities and inefficiencies inherent in the grading process.
[2]. Shreya Singh , Prof. Uday Rote , Omkar Manchekar ,Prof. Sheetal Jagtap ,Ambar Patwardhan, Dr. Hariram Chavan proposed the concept of Artificial Intelligence, OCR, and NLP to solve the problem. The answer sheets of the student is compared to the model answer sheet by the evaluator and will then generate the final score based on multiple parameters(sentence splitting, Jaccard similarity, grammar checking and sentence similarity). For the implementation of the system: cosine similarity and Jaccard similarity was used. The major setback of cosine similarity is it takes into consideration even the repetition of the same words. The measure of cosine similarity is higher primarily due to considering the repetitive similar words multiple times. This can generate a greater similarity level completely based on the number of times the word is repeated. Hence, Jaccard similarity is the better measure of similarity for the system.
Recognizing this limitation, there is a growing demand for innovative solutions that address the assessment of the responses in online exams. In response to this need, our focus turns to the development of an Automatic Answer Checker (AAC) model, designed to evaluate the answers and provide grading. This model incorporates advanced Natural Language Processing (NLP) techniques to comprehend, analyse, and assign weightage to the responses, bridging the gap between traditional human grading and the efficiency of online evaluation. This paper explores the potential of the AAC model to revolutionize the assessment paradigm, offering a solution that ensures the accuracy and reliability of grading for the questions in the modern era of education. The subsequent overview delves deeper into the key components and functionalities of this innovative approach, shedding light on
© 2024, IRJET
|
Impact Factor value: 8.226
[3].Potsangbam Sushila Devi, Sunita Sarkar, Takhellambam Sonamani Singh, Laimayum Dayal Sharma, Chongtham Pankaj and Khoirom Rajib Singh proposed a system designed to evaluate and check identical and semantic related answers
|
ISO 9001:2008 Certified Journal
|
Page 38