Abstract: Recruiting freshers when they are in final semesters through off-campus drive is a time- consuming process as it needs more recruiters to assess them. It requires more recruiters to filter the best candidate. The average time to assess a candidate is 10 to 20 minutes. Then it will take more time to shortlist the best candidates to the company. So with this paper, tried to automate this process of hiring using Natural Language Processing, Bidirectional Encoders, Speech Recognition and Semantic Matching and Supervised Similarity Measuring Techniques. The answers provided by the candidates can be recorded and pre-processed using speech recognition and Bidirectional Encoder Representations from Transformers (BERT). The recorded answers can be checked and matched through semantic sentence matching using the question and answer database designed by the recruiters. This web application would be the fastest and perfect way to hire candidates to the company.
Keywords: semantic similarity, cosine similarity, soft cosine similarity, word embedding, Euclidean distance, Cosine distance, BERT, Soft Cosine Similarity, TF-IDF Vectorizer.
| DOI: 10.17148/IJARCCE.2022.114146